Gaussian Blur

Posts Tagged 'Gaussian Blur'

C# How to: Weighted Difference of Gaussians

Article Purpose

It is the purpose of this article to illustrate the concept of Difference of Gaussians Edge Detection. This article extends the conventional implementation of Difference of Gaussian algorithms through the application of equally sized matrix kernels only differing by a weight factor.

Frog: Kernel 5×5, Weight1 0.1, Weight2 2.1

Sample Source Code

This article is accompanied by a sample source code Visual Studio project which is available for download here.

Using the Sample Application

This article relies on a sample application included as part of the accompanying sample source code. The sample application serves as a practical implementation of the concepts explored throughout this article.

The sample application user interface enables the user to configure and control the implementation of a Difference of Gaussians Edge Detection Image filter. The configuration options exposed through the sample application’s user interface can be detailed as follows:

Load/Save Images – When executing the sample application users are able to load source/input images from the local file system through clicking the Load Image button. If desired, the sample application enables users to save resulting images to the local file system through clicking the Save Image button.
Kernel Size – This option relates to the size of the matrix kernels that is to be implemented when performing Gaussian Blurring through image convolution. Smaller matrix kernels are faster to compute and generally result in image edges detected in the source/input image to be expressed through thinner gradient edges. Larger matrix kernels can be computationally expensive to compute as kernel sizes increase. In addition, the edges detected in source/input images will generally be expressed as thicker gradient edges in resulting images.
Weight Values – The sample application calculates Gaussian matrix kernels and in doing so implements a weight factor. A Weight Factor determines the blur intensity observed in result images after having applied image convolution. Higher weight factors result in a more intense level of Gaussian Blurring being applied. As expected, lower weight factors values result in a less intense level of Gaussian Blurring being applied. If the value of the first weight factor exceeds the value of the second weight factor resulting images will be generated with a Black background and edges being indicated in White. In a similar fashion, when the second weight factor value exceeds that of the first weight factor resulting images will be generated with a White background and edges being indicated in Black. The greater the difference between the first and second weight factor values result in a greater degree of image noise removal. When weight factor values only differ slightly, resulting images may be prone to image noise.

The following image is screenshot of the Weighted Difference of Gaussians sample application in action:

Frog: Kernel 5×5, Weight1 1.8, Weight2 0.1

The Gaussian Blur algorithm can be described as one of the most popular and widely implemented methods of image blurring. From Wikipedia we gain the following excerpt:

A Gaussian blur (also known as Gaussian smoothing) is the result of blurring an image by a Gaussian function. It is a widely used effect in graphics software, typically to reduce image noise and reduce detail. The visual effect of this blurring technique is a smooth blur resembling that of viewing the image through a translucent screen, distinctly different from the bokeh effect produced by an out-of-focus lens or the shadow of an object under usual illumination. Gaussian smoothing is also used as a pre-processing stage in computer vision algorithms in order to enhance image structures at different scales.

Mathematically, applying a Gaussian blur to an image is the same as convolving the image with a Gaussian function. This is also known as a two-dimensional Weierstrass transform.

Take Note: The Gaussian Blur algorithm has the attribute of smoothing image detail/definition whilst also having an edge preservation attribute. When applying a Gaussian Blur to an image a level of image detail/definition will be blurred/smoothed away, done in a fashion that would exclude/preserve image gradient edges.

Frog: Kernel 5×5, Weight1 2.7, Weight2 0.1

Difference of Gaussians Edge Detection

Difference of Gaussian refers to a specific method of image edge detection. Difference of Gaussians, common abbreviated as DoG, functions through the implementation of Gaussian Image Blurring.

A clear and concise description can be found on the Difference of Gaussians Wikipedia Article Page:

In imaging science, difference of Gaussians is a feature enhancement algorithm that involves the subtraction of one blurred version of an original image from another, less blurred version of the original. In the simple case of grayscale images, the blurred images are obtained by convolving the original grayscale images with Gaussian kernels having differing standard deviations. Blurring an image using a Gaussian kernel suppresses only high-frequency spatial information. Subtracting one image from the other preserves spatial information that lies between the range of frequencies that are preserved in the two blurred images. Thus, the difference of Gaussians is a band-pass filter that discards all but a handful of spatial frequencies that are present in the original grayscale image.

In a conventional sense Difference of Gaussians involves applying Gaussian Blurring to images created as copies of the original source/input image. There must be a difference in the size of the kernels implemented when applying image convolution. A typical example would be applying a 3×3 Gaussian blur on one image copy whilst applying a 5×5 Gaussian blur on another image copy. The final step requires creating a result image populated by subtracting the two blurred image copies. The results obtained from subtraction represents the edges forming part of the source/input image.

This article extends beyond the conventional method of implementing Difference of Gaussians Edge Detection. The implementation illustrated in this article retains the core concept of subtracting values which have been blurred to different intensities. The implementation method explored here differs from the conventional method in the sense that the matrix kernels implemented do not differ in size. Both matrix kernels are in fact required to have the same size dimensions.

The matrix kernels implemented are equal in terms of their size dimensions, although kernel values are different. Expressed from another angle: equally sized matrix kernels of which one represents a more intense level of blurring than the other. A matrix kernels’ resulting intensity can be determined by the weight factor implemented when calculating the kernel values.

Frog: Kernel 5×5, Weight1 3.7, Weight2 0.2

The advantages of implementing equally sized kernels can be described as follows:

Single Convolution implementation: Image Convolution involves executing several nested code loops. Application performance can be severely negatively impacted when executing large nested loops. The conventional method of implementing Difference of Gaussians generally involves having to implement two instances of image convolution, once per image copy. The Difference of Gaussians method implemented in this article executes the code loops related to image convolution only once. Considering the kernels are equal in size, both can be iterated within the same set of loops.

Eliminating Image subtraction: In conventional Difference of Gaussians implementations images expressing differing intensity levels of Gaussian Blurring have to be subtracted. The Difference of Gaussian implementation method described in this article eliminates the need to perform image subtraction. When applying image convolution using both kernels simultaneously the two results obtained, one from each kernel, can be subtracted and assigned to the result image. In addition, through calculating both kernel results at the same time further reduces the need to create two temporary source image copies.

Frog: Kernel 5×5, Weight1 2.4, Weight2 0.3

Difference of Gaussians Edge Detection Required Steps

When implementing a Difference of Gaussians Edge Detection several steps are required, those steps are detailed as follows:

Calculate Kernels – Before implementing image convolution two matrix kernels have to be calculated. The calculated kernels are required to be of equal size and differ in Gaussian Blur intensity. The sample application allows the user to configure kernel intensity through updating the weight values, expressed as Weight 1 and Weight 2.
Convert Source Image to Grayscale – Applying image convolution on grayscale images outperforms convolution on RGB images. When converting an RGB pixel to a grayscale pixel, colour components are combined to form a single gray level intensity. In other words a grayscale image consists of a third of the number of pixels when compared to the RGB image from which the grayscale image had been rendered. In the case of an ARGB image the derived grayscale image will be expressed in 25% of the number of pixels forming part of the source ARGB image. When applying image convolution the number of processor cycles increases when the image pixel count increases.
Perform Convolution Implementing Thresholds – Using the newly created grayscale image perform image convolution for both kernels calculated in the first step. The result value equates to subtracting the two results obtained from convolution. If the result value exceeds the difference between the first and second kernel weight value, the resulting pixel should be set to White, if not, set the result pixel to Black.

Frog: Kernel 5×5, Weight1 2.1, Weight2 0.5

Calculating Gaussian Convolution Kernels

The sample application implements Gaussian Blur Kernel calculations. The matrix kernels implemented in image convolution are calculated at runtime, as opposed to being hard coded. Being able to dynamically construct convolution kernels has the advantage of providing a greater degree of control in runtime regarding image convolution application.

Several steps are involved in calculating Gaussian Blur Matrix Kernels. The first required step being to determine the Matrix Kernel Size and Weight. The size and weight factor of a matrix kernel comprises the two configurable values implemented when calculating Gaussian Blur Kernels. In the case of this article and the sample application those values will be configured by the user through the sample application’s user interface.

The formula implemented in calculating Gaussian Blur Kernels can be expressed as follows:

The formula contains a number of symbols, which define how the filter will be implemented. The symbols forming part of the Gaussian Kernel formula are described in the following list:

G(x y) – A value calculated using the Gaussian Kernel formula. This value forms part of a Kernel, representing a single element.
π – Pi, one of the better known members of the Greek alphabet. The mathematical constant defined as 22 / 7.
σ – The lower case version of the Greek alphabet letter Sigma. This symbol simply represents a threshold or factor value, as specified by the user.
e – The formula references a lower case e symbol. The symbol represents Euler’s number. The value of Euler’s number has been defined as a mathematical constant equating to 2.71828182846.
x, y – The variables referenced as x and y relate to pixel coordinates within an image. y Representing the vertical offset or row and x represents the horizontal offset or column.

Note: The formula’s implementation expects x and y to equal zero values when representing the coordinates of the pixel located in the middle of the kernel.

Frog: Kernel 7×7, Weight1 0.1, Weight2 2.0

Implementing Gaussian Kernel Calculations

The sample application defines the GaussianCalculator.Calculate method. This method accepts two parameters, kernel size and kernel weight. The following code snippet details the implementation:

public static double[,] Calculate(int lenght, double weight) 
{
    double[,] Kernel = new double [lenght, lenght]; 
    double sumTotal = 0; 

  
    int kernelRadius = lenght / 2; 
    double distance = 0; 

  
    double calculatedEuler = 1.0 /  
    (2.0 * Math.PI * Math.Pow(weight, 2)); 

  
    for (int filterY = -kernelRadius; 
         filterY <= kernelRadius; filterY++) 
    {
        for (int filterX = -kernelRadius; 
            filterX <= kernelRadius; filterX++) 
        {
            distance = ((filterX * filterX) +  
                       (filterY * filterY)) /  
                       (2 * (weight * weight)); 

  
            Kernel[filterY + kernelRadius,  
                   filterX + kernelRadius] =  
                   calculatedEuler * Math.Exp(-distance); 

  
            sumTotal += Kernel[filterY + kernelRadius,  
                               filterX + kernelRadius]; 
        } 
    } 

  
    for (int y = 0; y < lenght; y++) 
    { 
        for (int x = 0; x < lenght; x++) 
        { 
            Kernel[y, x] = Kernel[y, x] *  
                           (1.0 / sumTotal); 
        } 
    } 

  
    return Kernel; 
}

Frog: Kernel 3×3, Weight1 0.1, Weight2 1.8

Implementing Difference of Gaussians Edge Detection

The sample source code defines the DifferenceOfGaussianFilter method. This method has been defined as an extension method targeting the Bitmap class. The following code snippet provides the implementation:

public static Bitmap DifferenceOfGaussianFilter(this Bitmap sourceBitmap,  
                                                int matrixSize, double weight1, 
                                                double weight2) 
{
    double[,] kernel1 =  
    GaussianCalculator.Calculate(matrixSize,  
    (weight1 > weight2 ? weight1 : weight2)); 

 
    double[,] kernel2 =  
    GaussianCalculator.Calculate(matrixSize,  
    (weight1 > weight2 ? weight2 : weight1)); 

 
    BitmapData sourceData = sourceBitmap.LockBits(new Rectangle (0, 0, 
                             sourceBitmap.Width, sourceBitmap.Height), 
                                               ImageLockMode.ReadOnly, 
                                         PixelFormat.Format32bppArgb); 

 
    byte[] pixelBuffer = new byte [sourceData.Stride * sourceData.Height]; 
    byte[] resultBuffer = new byte [sourceData.Stride * sourceData.Height]; 
    byte[] grayscaleBuffer = new byte [sourceData.Width * sourceData.Height]; 

 
    Marshal.Copy(sourceData.Scan0, pixelBuffer, 0, pixelBuffer.Length); 
    sourceBitmap.UnlockBits(sourceData); 

 
    double rgb = 0; 

 
    for (int source = 0, dst = 0;  
         source < pixelBuffer.Length && dst < grayscaleBuffer.Length;  
         source += 4, dst++) 
    { 
        rgb = pixelBuffer * 0.11f; 
        rgb += pixelBuffer * 0.59f; 
        rgb += pixelBuffer * 0.3f; 

 
        grayscaleBuffer[dst] = (byte)rgb; 
    }

 
    double color1 = 0.0; 
    double color2 = 0.0; 

 
    int filterOffset = (matrixSize - 1) / 2; 
    int calcOffset = 0; 

 
    for (int source = 0, dst = 0;  
         source < grayscaleBuffer.Length && dst + 4 < resultBuffer.Length;  
         source++, dst += 4) 
    { 
        color1 = 0; 
        color2 = 0; 

         
        for (int filterY = -filterOffset; 
                filterY <= filterOffset; filterY++) 
        {
            for (int filterX = -filterOffset; 
                filterX <= filterOffset; filterX++) 
            {
                calcOffset = source + (filterX) + 
                             (filterY * sourceBitmap.Width); 

 
                calcOffset = (calcOffset < 0 ? 0 :  
                             (calcOffset >= grayscaleBuffer.Length ?  
                             grayscaleBuffer.Length - 1 : calcOffset)); 

 
                color1 += (grayscaleBuffer[calcOffset]) * 
                           kernel1[filterY + filterOffset, 
                           filterX + filterOffset]; 

 
                color2 += (grayscaleBuffer[calcOffset]) * 
                           kernel2[filterY + filterOffset, 
                           filterX + filterOffset]; 
            } 
        }

 
        color1 = color1 - color2; 
        color1 = (color1 >= weight1 - weight2 ? 255 : 0); 

 
        resultBuffer[dst] = (byte)color1; 
        resultBuffer[dst + 1] = (byte)color1; 
        resultBuffer[dst + 2] = (byte)color1; 
        resultBuffer[dst + 3] = 255; 
    }

 
    Bitmap resultBitmap = new Bitmap(sourceBitmap.Width, sourceBitmap.Height); 

 
    BitmapData resultData = resultBitmap.LockBits(new Rectangle (0, 0, 
                             resultBitmap.Width, resultBitmap.Height), 
                                              ImageLockMode.WriteOnly, 
                                         PixelFormat.Format32bppArgb); 

 
    Marshal.Copy(resultBuffer, 0, resultData.Scan0, resultBuffer.Length); 
    resultBitmap.UnlockBits(resultData); 

 
    return resultBitmap; 
}

Frog: Kernel 3×3, Weight1 2.1, Weight2 0.7

Sample Images

This article features a number of sample images. All featured images have been licensed allowing for reproduction. The following image files feature as sample images:

Panamanian Golden Frog – Licensed under the Creative Commons Attribution 2.0 Generic license. Attribution: Brian Gratwicke. Download from Wikipedia.
Dendropsophus Microcephalus – Licensed under the Creative Commons Attribution 2.0 Generic license. Attribution: Brian Gratwicke. Download from Wikipedia.
Tyler’s Tree Frog – Has been released into the public domain by its author, LiquidGhoul. This applies worldwide. Download from Wikipedia.
Mimic Poison Frog – Licensed under the Creative Commons Attribution-Share Alike 3.0 Unported license. Download from Wikipedia.
Phyllobates Terribilis – This file is licensed under the Creative Commons Attribution-Share Alike 2.0 Germany license. Attribution: Wilfried Berns. Download from Wikipedia.

Panamanian Golden Frog

Dendropsophus Microcephalus

Tyler’s Tree Frog

Mimic Poison Frog

Phyllobates Terribilis

C# How to: Image Blur

Published June 9, 2013 Augmented Reality , C# , Code Samples , Extension Methods , Graphic Filters , Graphics , How to , Image Convolution , Image Filters , Image Processing , Learn Everyday , Microsoft , New Version , Opensource , Update 29 Comments
Tags: Augmented Reality, Bitmap ARGB, Bitmap Filters, Bitmap.LockBits, BitmapData, Box Blur, C#, Calculating Kernels, Code Sample, Converting Images, Convolution matrix, Feature extraction, Gaussian, Gaussian Blur, Gaussian Kernels, Graphic Filters, Image, Image Blur, Image processing, Image Smoothing, Image Transform, Machine vision, Matrix, Mean Filter, Median Filter, Motion Blur, Pixel Filters

Article Purpose

This article serves to provides an introduction and discussion relating to Image Blurring methods and techniques. The Image Blur methods covered in this article include: Box Blur, Gaussian Blur, Mean Filter, Median Filter and Motion Blur.

Daisy: Mean 9×9

Sample Source Code

This article is accompanied by a sample source code Visual Studio project which is available for download here.

Using the Sample Application

This article is accompanied by a sample application, intended to provide a means of testing and replicating topics discussed in this article. The sample application is a Windows Forms based application of which the user interface enables the user to select an Image Blur type to implement.

When clicking the Load Image button users are able to browse the local file system in order to select source/input images. In addition users are also able to save blurred result images when clicking the Save Image button and browsing the local file system.

Daisy: Mean 7×7

The sample application provides the user with the ability to select the method of image blurring to implement. The combobox dropdown located on the right-hand side of the user interface lists all of the supported methods of image blurring. When a user selects an item from the combobox, the associated blur method will be implemented on the preview image.

The image below is a screenshot of the Image Blur Filter sample application in action:

Image Blur Overview

The process of image blurring can be regarded as reducing the sharpness or crispness defined by an image. Image blurring results in image detail/definition being perceived as less distinct. Images are often blurred as a method of smoothing an image.

Images perceived as too crisp/sharp can be softened by applying a variety of image blurring techniques and intensity levels. Often images are smoothed/blurred in order to remove/reduce image noise. In image edge detection implementations better results are often achieved when first implementing noise reduction through smoothing/blurring. Image blurring can even be implemented in a fashion where results reflect image edge detection, a method known as Difference of Gaussians.

In this article and the accompanying sample source code all methods of image blurring supported have been implemented through image convolution, with the exception of the Median filter. Each of the supported methods in essence only represent a different convolution matrix kernel. The image blurring technique capable of achieving optimal results will to varying degrees be dependent on the features present in the specified source/input image. Each method provides a different set of desired properties and compromises. In the following sections an overview of each method will be discussed.

Daisy: Mean 9×9

Mean Filter/Box Blur

The Mean Filter also sometimes referred to as a Box Blur represents a fairly simplistic implementation and definition. A Mean Filter definition can be found on Wikipedia as follows:

A box blur is an image filter in which each pixel in the resulting image has a value equal to the average value of its neighbouring pixels in the input image. It is a form of low-pass ("blurring") filter and is a convolution.

Due to its property of using equal weights it can be implemented using a much simpler accumulation algorithm which is significantly faster than using a sliding window algorithm.

Mean Filter as a title relates to all weight values in a convolution kernel being equal, therefore the alternate title of Box Blur. In most cases a Mean Filter matrix kernel will only contain the value one. When performing image convolution implementing a Mean Filter kernel, the factor value equates to the 1 being divided by the sum of all kernel values.

The following is an example of a 5×5 Mean Filter convolution kernel:

The kernel consist of 25 elements, therefore the factor value equates to one divided by twenty five.

The Mean Filter Blur does not result in the same level of smoothing achieved by other image blur methods. The Mean Filter method can also be susceptible to directional artefacts.

Daisy Mean 5×5

Gaussian Blur

The Gaussian method of image blurring is a popular and often implemented filter. In contrast to the Box Blur method Gaussian Blurring produce resulting images appearing to contain a more uniform level of smoothing. When implementing image edge detection a Gaussian Blur is often applied to source/input images resulting in noise reduction. The Gaussian Blur has a good level of image edge preservation, hence being used in edge detection operations.

From Wikipedia we gain the following description:

A Gaussian blur (also known as Gaussian smoothing) is the result of blurring an image by a Gaussian function. It is a widely used effect in graphics software, typically to reduce image noise and reduce detail. The visual effect of this blurring technique is a smooth blur resembling that of viewing the image through a translucent screen, distinctly different from the bokeh effect produced by an out-of-focus lens or the shadow of an object under usual illumination. Gaussian smoothing is also used as a pre-processing stage in computer vision algorithms in order to enhance image structures at different scales

A potential drawback to implementing a Gaussian blur results from the filter being computationally intensive. The following matrix kernel represents a 5×5 Gaussian Blur. The sum total of all elements in the kernel equate to 159, therefore a factor value of 1.0 / 159.0 will be implemented.

Daisy: Gaussian 5×5

Median Filter Blur

The Median Filter is classified as a non-linear filter. In contrast to the other methods of image blurring discussed in this article the Median Filter implementation does not involve convolution or a predefined matrix kernel. The following description can be found on Wikipedia:

In signal processing, it is often desirable to be able to perform some kind of noise reduction on an image or signal. The median filter is a nonlinear digital filtering technique, often used to remove noise. Such noise reduction is a typical pre-processing step to improve the results of later processing (for example, edge detection on an image). Median filtering is very widely used in digital image processing because, under certain conditions, it preserves edges while removing noise.

Daisy: Median 7×7

As the name implies, the Median Filter operates by calculating the median value of a pixel group also referred to as a window. Calculating a Median value involves a number of steps. The required steps are listed as follows:

Iterate each pixel that forms part of the source/input image.
In relation to the pixel currently being iterated determine neighbouring pixels located within the bounds defined by the window size. The window location should be offset in order to align the window’s middle pixel and the pixel currently being iterated.
Neighbouring pixels located within the bounds defined by the window should be added to a one dimensional neighbourhood array. Once all value have been added, the array should be sorted by value.
The pixel value located at the middle of the sorted neighbourhood array qualifies as the Median value. The newly determined Median value should be assigned to the pixel currently being iterated.
Repeat the steps listed above until all pixels within the source/input image have been iterated.

Similar to the Gaussian Blur filter the Median Filter has the ability to smooth image noise whilst providing edge preservation. Depending on the window size implemented and the physical dimensions of input/source images the Median Filter can be computationally expensive.

Daisy: Median 9×9

Motion Blur

The sample source implements Motion Blur filters. Motion blurring in the traditional sense has been association with photography and video capturing. Motion Blurring can often be observed in scenarios where rapid movements are being captured to photographs or video recording. When recording a single frame, rapid movements could result in the image changing before the frame being captured has completed.

Motion Blurring can be synthetically imitated through the implementation of Digital Motion Blur filters. The size of the matrix kernel provided when implementing image convolution affects the filter intensity perceived in result images. Relating to Motion Blur filters the size of the kernel specified in convolution influences the perception and appearance of how rapidly movement had occurred to have blurred the resulting image. Larger kernels produce the appearance of more rapid motion, whereas smaller kernels result in less rapid motion being perceived.

Daisy: Motion Blur 7×7 135 Degrees

Depending on the kernel specified the ability exists to create the appearance of movement having occurred in a certain direction. The sample source code implements Motion Blur filters at 45 degrees, 135 degrees and in both directions simultaneously.

The kernel listed below represents a 5×5 Motion Blur filter occurring at 45 degrees and 135 degrees:

Image Blur Implementation

The sample source code implements all of the concepts explored throughout this article. The source code definition can be grouped into 4 sections: ImageBlurFilter method, ConvolutionFilter method, MedianFilter method and the Matrix class. The following article sections relate to the 4 main source code sections.

The ImageBlurFilter extension method has the purpose of invoking the correct blur filter method and relevant method parameters. This method acts as a method wrapper providing the technical implementation details required when performing a specified blur filter.

The definition of the ImageBlurFilter extension method as follows:

 public static Bitmap ImageBlurFilter(this Bitmap sourceBitmap,  
                                             BlurType blurType) 
{  
     Bitmap resultBitmap = null; 

       
     switch (blurType) 
     {  
         case BlurType.Mean3x3: 
              {  
                 resultBitmap = sourceBitmap.ConvolutionFilter( 
                                Matrix.Mean3x3, 1.0 / 9.0, 0); 
              } break; 
         case BlurType.Mean5x5: 
              {  
                 resultBitmap = sourceBitmap.ConvolutionFilter( 
                                Matrix.Mean5x5, 1.0 / 25.0, 0); 
              } break; 
         case BlurType.Mean7x7: 
              {  
                 resultBitmap = sourceBitmap.ConvolutionFilter( 
                                Matrix.Mean7x7, 1.0 / 49.0, 0); 
              } break; 
         case BlurType.Mean9x9: 
              {  
                 resultBitmap = sourceBitmap.ConvolutionFilter( 
                                Matrix.Mean9x9, 1.0 / 81.0, 0); 
              } break; 
         case BlurType.GaussianBlur3x3: 
              {  
                 resultBitmap = sourceBitmap.ConvolutionFilter( 
                         Matrix.GaussianBlur3x3, 1.0 / 16.0, 0); 
              } break; 
         case BlurType.GaussianBlur5x5: 
              {  
                 resultBitmap = sourceBitmap.ConvolutionFilter( 
                        Matrix.GaussianBlur5x5, 1.0 / 159.0, 0); 
              } break; 
         case BlurType.MotionBlur5x5: 
              {  
                 resultBitmap = sourceBitmap.ConvolutionFilter( 
                           Matrix.MotionBlur5x5, 1.0 / 10.0, 0); 
              } break; 
         case BlurType.MotionBlur5x5At45Degrees: 
              {  
                 resultBitmap = sourceBitmap.ConvolutionFilter( 
                 Matrix.MotionBlur5x5At45Degrees, 1.0 / 5.0, 0); 
              } break; 
         case BlurType.MotionBlur5x5At135Degrees: 
              {  
                 resultBitmap = sourceBitmap.ConvolutionFilter( 
                 Matrix.MotionBlur5x5At135Degrees, 1.0 / 5.0, 0); 
              } break; 
         case BlurType.MotionBlur7x7: 
              {  
                 resultBitmap = sourceBitmap.ConvolutionFilter( 
                 Matrix.MotionBlur7x7, 1.0 / 14.0, 0); 
              } break; 
         case BlurType.MotionBlur7x7At45Degrees: 
              {  
                 resultBitmap = sourceBitmap.ConvolutionFilter( 
                 Matrix.MotionBlur7x7At45Degrees, 1.0 / 7.0, 0); 
              } break; 
         case BlurType.MotionBlur7x7At135Degrees: 
              {  
                 resultBitmap = sourceBitmap.ConvolutionFilter( 
                 Matrix.MotionBlur7x7At135Degrees, 1.0 / 7.0, 0); 
              } break; 
         case BlurType.MotionBlur9x9: 
              {  
                 resultBitmap = sourceBitmap.ConvolutionFilter( 
                 Matrix.MotionBlur9x9, 1.0 / 18.0, 0); 
              } break; 
         case BlurType.MotionBlur9x9At45Degrees: 
              {  
                 resultBitmap = sourceBitmap.ConvolutionFilter( 
                 Matrix.MotionBlur9x9At45Degrees, 1.0 / 9.0, 0); 
              } break; 
         case BlurType.MotionBlur9x9At135Degrees: 
              {  
                 resultBitmap = sourceBitmap.ConvolutionFilter( 
                 Matrix.MotionBlur9x9At135Degrees, 1.0 / 9.0, 0); 
              } break; 
         case BlurType.Median3x3: 
              {  
                 resultBitmap = sourceBitmap.MedianFilter(3); 
              } break; 
         case BlurType.Median5x5: 
              {  
                 resultBitmap = sourceBitmap.MedianFilter(5); 
              } break; 
         case BlurType.Median7x7: 
              {  
                 resultBitmap = sourceBitmap.MedianFilter(7); 
              } break; 
         case BlurType.Median9x9: 
              {  
                 resultBitmap = sourceBitmap.MedianFilter(9); 
              } break; 
         case BlurType.Median11x11: 
              {  
                 resultBitmap = sourceBitmap.MedianFilter(11); 
              } break; 
      }  

   
     return resultBitmap; 
}

Daisy: Motion Blur 9×9

The Matrix class serves as a collection of various kernel definitions. The Matrix class and all public properties are defined as static. The definition of the Matrix class as follows:

     public static class Matrix 
    {  
         public static double[,] Mean3x3 
         {  
             get 
             {  
                 return new double[,]   
                { {  1, 1, 1, },  
                  {  1, 1, 1, },  
                  {  1, 1, 1, }, }; 
             }  
         }  

  
         public static double[,] Mean5x5 
         {  
             get 
             {  
                 return new double[,]   
                { {  1, 1, 1, 1, 1 },  
                  {  1, 1, 1, 1, 1 }, 
                  {  1, 1, 1, 1, 1 }, 
                  {  1, 1, 1, 1, 1 }, 
                  {  1, 1, 1, 1, 1 }, }; 
            }  
        }  

    
        public static double[,] Mean7x7 
        {  
             get 
             {  
                return new double[,]   
                { {  1, 1, 1, 1, 1, 1, 1 },  
                  {  1, 1, 1, 1, 1, 1, 1 }, 
                  {  1, 1, 1, 1, 1, 1, 1 }, 
                  {  1, 1, 1, 1, 1, 1, 1 }, 
                  {  1, 1, 1, 1, 1, 1, 1 }, 
                  {  1, 1, 1, 1, 1, 1, 1 }, 
                  {  1, 1, 1, 1, 1, 1, 1 }, }; 
             }  
        }  

    
         public static double[,] Mean9x9 
         {  
             get 
             {  
                return new double[,]   
                { {  1, 1, 1, 1, 1, 1, 1, 1, 1 },  
                  {  1, 1, 1, 1, 1, 1, 1, 1, 1 }, 
                  {  1, 1, 1, 1, 1, 1, 1, 1, 1 }, 
                  {  1, 1, 1, 1, 1, 1, 1, 1, 1 }, 
                  {  1, 1, 1, 1, 1, 1, 1, 1, 1 }, 
                  {  1, 1, 1, 1, 1, 1, 1, 1, 1 }, 
                  {  1, 1, 1, 1, 1, 1, 1, 1, 1 }, 
                  {  1, 1, 1, 1, 1, 1, 1, 1, 1 }, 
                  {  1, 1, 1, 1, 1, 1, 1, 1, 1 }, }; 
            }  
        }  

    
        public static double[,] GaussianBlur3x3 
        {  
            get 
            {  
                return new double[,]   
                { {  1, 2, 1, },  
                  {  2, 4, 2, },  
                  {  1, 2, 1, }, }; 
            }  
        }  

    
        public static double[,] GaussianBlur5x5 
        {  
            get 
            {  
                return new double[,]   
                { {  2, 04, 05, 04, 2 },  
                  {  4, 09, 12, 09, 4 },  
                  {  5, 12, 15, 12, 5 }, 
                  {  4, 09, 12, 09, 4 }, 
                  {  2, 04, 05, 04, 2 }, }; 
            }  
        }  

    
        public static double[,] MotionBlur5x5 
        {  
             get 
            {  
                return new double[,]   
                { {  1, 0, 0, 0, 1 },  
                  {  0, 1, 0, 1, 0 },  
                  {  0, 0, 1, 0, 0 }, 
                  {  0, 1, 0, 1, 0 }, 
                  {  1, 0, 0, 0, 1 }, }; 
            }  
        }  

    
        public static double[,] MotionBlur5x5At45Degrees 
        {  
             get 
            {  
                return new double[,]   
                { {  0, 0, 0, 0, 1 },  
                  {  0, 0, 0, 1, 0 },  
                  {  0, 0, 1, 0, 0 }, 
                  {  0, 1, 0, 0, 0 }, 
                  {  1, 0, 0, 0, 0 }, }; 
            }  
        }  

    
        public static double[,] MotionBlur5x5At135Degrees 
        {  
             get 
            {  
                return new double[,]   
                { {  1, 0, 0, 0, 0 },  
                  {  0, 1, 0, 0, 0 },  
                  {  0, 0, 1, 0, 0 }, 
                  {  0, 0, 0, 1, 0 }, 
                  {  0, 0, 0, 0, 1 }, }; 
            }  
        }  

    
        public static double[,] MotionBlur7x7 
        {  
            get 
            {  
                return new double[,]   
                { {  1, 0, 0, 0, 0, 0, 1 },  
                  {  0, 1, 0, 0, 0, 1, 0 },  
                  {  0, 0, 1, 0, 1, 0, 0 }, 
                  {  0, 0, 0, 1, 0, 0, 0 }, 
                  {  0, 0, 1, 0, 1, 0, 0 }, 
                  {  0, 1, 0, 0, 0, 1, 0 }, 
                  {  1, 0, 0, 0, 0, 0, 1 }, }; 
            }  
        }  

    
        public static double[,] MotionBlur7x7At45Degrees 
        {  
            get 
            {  
                return new double[,]   
                { {  0, 0, 0, 0, 0, 0, 1 },  
                  {  0, 0, 0, 0, 0, 1, 0 },  
                  {  0, 0, 0, 0, 1, 0, 0 }, 
                  {  0, 0, 0, 1, 0, 0, 0 }, 
                  {  0, 0, 1, 0, 0, 0, 0 }, 
                  {  0, 1, 0, 0, 0, 0, 0 }, 
                  {  1, 0, 0, 0, 0, 0, 0 }, }; 
            }  
        }  

    
        public static double[,] MotionBlur7x7At135Degrees 
        {  
            get 
            {  
                return new double[,]   
                { {  1, 0, 0, 0, 0, 0, 0 },  
                  {  0, 1, 0, 0, 0, 0, 0 },  
                  {  0, 0, 1, 0, 0, 0, 0 }, 
                  {  0, 0, 0, 1, 0, 0, 0 }, 
                  {  0, 0, 0, 0, 1, 0, 0 }, 
                  {  0, 0, 0, 0, 0, 1, 0 }, 
                  {  0, 0, 0, 0, 0, 0, 1 }, }; 
            }  
        }  

    
        public static double[,] MotionBlur9x9 
        {  
            get 
            {  
                return new double[,]   
                { { 1, 0, 0, 0, 0, 0, 0, 0, 1, }, 
                  { 0, 1, 0, 0, 0, 0, 0, 1, 0, }, 
                  { 0, 0, 1, 0, 0, 0, 1, 0, 0, }, 
                  { 0, 0, 0, 1, 0, 1, 0, 0, 0, }, 
                  { 0, 0, 0, 0, 1, 0, 0, 0, 0, }, 
                  { 0, 0, 0, 1, 0, 1, 0, 0, 0, }, 
                  { 0, 0, 1, 0, 0, 0, 1, 0, 0, }, 
                  { 0, 1, 0, 0, 0, 0, 0, 1, 0, }, 
                  { 1, 0, 0, 0, 0, 0, 0, 0, 1, }, }; 
            }  
        }  

    
        public static double[,] MotionBlur9x9At45Degrees 
        {  
             get 
            {  
                return new double[,]   
                { { 0, 0, 0, 0, 0, 0, 0, 0, 1, }, 
                  { 0, 0, 0, 0, 0, 0, 0, 1, 0, }, 
                  { 0, 0, 0, 0, 0, 0, 1, 0, 0, }, 
                  { 0, 0, 0, 0, 0, 1, 0, 0, 0, }, 
                  { 0, 0, 0, 0, 1, 0, 0, 0, 0, }, 
                  { 0, 0, 0, 1, 0, 0, 0, 0, 0, }, 
                  { 0, 0, 1, 0, 0, 0, 0, 0, 0, }, 
                  { 0, 1, 0, 0, 0, 0, 0, 0, 0, }, 
                  { 1, 0, 0, 0, 0, 0, 0, 0, 0, }, }; 
            }  
        }  

    
        public static double[,] MotionBlur9x9At135Degrees 
        {  
             get 
            {  
                return new double[,]   
                { { 1, 0, 0, 0, 0, 0, 0, 0, 0, }, 
                  { 0, 1, 0, 0, 0, 0, 0, 0, 0, }, 
                  { 0, 0, 1, 0, 0, 0, 0, 0, 0, }, 
                  { 0, 0, 0, 1, 0, 0, 0, 0, 0, }, 
                  { 0, 0, 0, 0, 1, 0, 0, 0, 0, }, 
                  { 0, 0, 0, 0, 0, 1, 0, 0, 0, }, 
                  { 0, 0, 0, 0, 0, 0, 1, 0, 0, }, 
                  { 0, 0, 0, 0, 0, 0, 0, 1, 0, }, 
                  { 0, 0, 0, 0, 0, 0, 0, 0, 1, }, }; 
            }  
        }  
    }

Daisy: Median 7×7

The MedianFilter extension method targets the Bitmap class. The MedianFilter method applies a Median Filter using the specified Bitmap and matrix size (window size), returning a new Bitmap representing the filtered image.

The definition of the MedianFilter extension method as follows:

 public static Bitmap MedianFilter(this Bitmap sourceBitmap, 
                                   int matrixSize) 
{ 
     BitmapData sourceData = 
                sourceBitmap.LockBits(new Rectangle(0, 0, 
                sourceBitmap.Width, sourceBitmap.Height), 
                ImageLockMode.ReadOnly, 
                PixelFormat.Format32bppArgb); 

   
     byte[] pixelBuffer = new byte[sourceData.Stride * 
                                   sourceData.Height]; 

   
     byte[] resultBuffer = new byte[sourceData.Stride * 
                                    sourceData.Height]; 

   
     Marshal.Copy(sourceData.Scan0, pixelBuffer, 0, 
                                pixelBuffer.Length); 

   
     sourceBitmap.UnlockBits(sourceData); 

   
     int filterOffset = (matrixSize - 1) / 2; 
     int calcOffset = 0; 

   
     int byteOffset = 0; 

   
     List<int> neighbourPixels = new List<int>(); 
     byte[] middlePixel; 

   
     for (int offsetY = filterOffset; offsetY < 
         sourceBitmap.Height - filterOffset; offsetY++) 
     { 
         for (int offsetX = filterOffset; offsetX < 
             sourceBitmap.Width - filterOffset; offsetX++) 
         { 
             byteOffset = offsetY * 
                          sourceData.Stride + 
                          offsetX * 4; 

   
             neighbourPixels.Clear(); 

   
             for (int filterY = -filterOffset; 
                 filterY <= filterOffset; filterY++) 
             {
                 for (int filterX = -filterOffset; 
                     filterX <= filterOffset; filterX++) 
                 {

   
                     calcOffset = byteOffset + 
                                  (filterX * 4) + 
                                  (filterY * sourceData.Stride); 

   
                     neighbourPixels.Add(BitConverter.ToInt32( 
                                      pixelBuffer, calcOffset)); 
                 } 
             } 

   
             neighbourPixels.Sort(); 
  
             middlePixel = BitConverter.GetBytes( 
                                neighbourPixels[filterOffset]); 

   
             resultBuffer[byteOffset] = middlePixel[0]; 
             resultBuffer[byteOffset + 1] = middlePixel[1]; 
             resultBuffer[byteOffset + 2] = middlePixel[2]; 
             resultBuffer[byteOffset + 3] = middlePixel[3]; 
         } 
     }

   
     Bitmap resultBitmap = new Bitmap (sourceBitmap.Width, 
                                      sourceBitmap.Height); 

   
     BitmapData resultData = 
                resultBitmap.LockBits(new Rectangle  (0, 0, 
                resultBitmap.Width, resultBitmap.Height), 
                ImageLockMode.WriteOnly, 
                PixelFormat.Format32bppArgb); 

   
     Marshal.Copy(resultBuffer, 0, resultData.Scan0, 
                                resultBuffer.Length); 

   
     resultBitmap.UnlockBits(resultData); 

   
     return resultBitmap; 
}

Daisy: Motion Blur 9×9

The sample source code performs image convolution by invoking the ConvolutionFilter extension method.

The definition of the ConvolutionFilter extension method as follows:

private static Bitmap ConvolutionFilter(this Bitmap sourceBitmap, 
                                          double[,] filterMatrix, 
                                               double factor = 1, 
                                                    int bias = 0) 
{ 
    BitmapData sourceData = sourceBitmap.LockBits(new Rectangle(0, 0, 
                             sourceBitmap.Width, sourceBitmap.Height), 
                                               ImageLockMode.ReadOnly, 
                                         PixelFormat.Format32bppArgb); 

   
    byte[] pixelBuffer = new byte[sourceData.Stride * sourceData.Height]; 
    byte[] resultBuffer = new byte[sourceData.Stride * sourceData.Height]; 

   
    Marshal.Copy(sourceData.Scan0, pixelBuffer, 0, pixelBuffer.Length); 
    sourceBitmap.UnlockBits(sourceData); 

   
    double blue = 0.0; 
    double green = 0.0; 
    double red = 0.0; 

   
    int filterWidth = filterMatrix.GetLength(1); 
    int filterHeight = filterMatrix.GetLength(0); 

   
    int filterOffset = (filterWidth - 1) / 2; 
    int calcOffset = 0; 

   
    int byteOffset = 0; 

   
    for (int offsetY = filterOffset; offsetY < 
        sourceBitmap.Height - filterOffset; offsetY++) 
     { 
        for (int offsetX = filterOffset; offsetX < 
            sourceBitmap.Width - filterOffset; offsetX++) 
         { 
            blue = 0; 
            green = 0; 
            red = 0; 

   
            byteOffset = offsetY * 
                         sourceData.Stride + 
                         offsetX * 4; 

   
            for (int filterY = -filterOffset; 
                filterY <= filterOffset; filterY++) 
             { 
                for (int filterX = -filterOffset; 
                    filterX <= filterOffset; filterX++) 
                 { 

   
                    calcOffset = byteOffset + 
                                 (filterX * 4) + 
                                 (filterY * sourceData.Stride); 

   
                    blue += (double)(pixelBuffer[calcOffset]) * 
                            filterMatrix[filterY + filterOffset, 
                                                filterX + filterOffset]; 

   
                    green += (double)(pixelBuffer[calcOffset + 1]) * 
                             filterMatrix[filterY + filterOffset, 
                                                filterX + filterOffset]; 

   
                    red += (double)(pixelBuffer[calcOffset + 2]) * 
                           filterMatrix[filterY + filterOffset, 
                                              filterX + filterOffset]; 
                 } 
             }

   
            blue = factor * blue + bias; 
            green = factor * green + bias; 
            red = factor * red + bias; 

   
            blue = (blue > 255 ? 255 : 
                   (blue < 0 ? 0 : 
                    blue)); 

   
            green = (green > 255 ? 255 : 
                    (green < 0 ? 0 : 
                     green)); 

   
            red = (red > 255 ? 255 : 
                  (red < 0 ? 0 : 
                   red));  

   
            resultBuffer[byteOffset] = (byte)(blue); 
            resultBuffer[byteOffset + 1] = (byte)(green); 
            resultBuffer[byteOffset + 2] = (byte)(red); 
            resultBuffer[byteOffset + 3] = 255; 
         } 
     }

   
    Bitmap resultBitmap = new Bitmap(sourceBitmap.Width, sourceBitmap.Height); 

   
    BitmapData resultData = resultBitmap.LockBits(new Rectangle (0, 0, 
                             resultBitmap.Width, resultBitmap.Height), 
                                              ImageLockMode.WriteOnly, 
                                         PixelFormat.Format32bppArgb); 

   
    Marshal.Copy(resultBuffer, 0, resultData.Scan0, resultBuffer.Length); 
    resultBitmap.UnlockBits(resultData); 

   
    return resultBitmap; 
}

Sample Images

This article features a number of sample images. All featured images have been licensed allowing for reproduction.

The sample images featuring an image of a yellow daisy is licensed under the Creative Commons Attribution-Share Alike 2.5 Generic license and can be downloaded from Wikimedia.org.

The sample images featuring an image of a white daisy is licensed under the Creative Commons Attribution-Share Alike 3.0 Unported license and can be downloaded from Wikipedia.

The sample images featuring an image of a pink daisy is licensed under the Creative Commons Attribution-Share Alike 2.5 Generic license and can be downloaded from Wikipedia.

The sample images featuring an image of a purple daisy is licensed under the Creative Commons Attribution-ShareAlike 3.0 License and can be downloaded from Wikipedia.

The Original Image

Daisy: Gaussian 3×3

Daisy: Gaussian 5×5

Daisy: Mean 3×3

Daisy: Mean 5×5

Daisy: Mean 7×7

Daisy: Mean 9×9

Daisy: Median 3×3

Daisy: Median 5×5

Daisy: Median 7×7

Daisy: Median 9×9

Daisy: Median 11×11

Daisy: Motion Blur 5×5

Daisy: Motion Blur 5×5 45 Degrees

Daisy: Motion Blur 5×5 135 Degrees

Daisy: Motion Blur 7×7

Daisy: Motion Blur 7×7 45 Degrees

Daisy: Motion Blur 7×7 135 Degrees

Daisy: Motion Blur 9×9

Daisy: Motion Blur 9×9 45 Degrees

Daisy: Motion Blur 9×9 135 Degrees

C# How to: Calculating Gaussian Kernels

Published June 8, 2013 Augmented Reality , C# , Code Samples , Extension Methods , Graphic Filters , Graphics , How to , Image Arithmetic , Image Filters , Image Processing , Microsoft , New Version , Opensource , Tip , Update 31 Comments
Tags: Augmented Reality, Bitmap ARGB, Bitmap Filters, Bitmap.LockBits, BitmapData, C#, Calculate Matrix, Calculating Kernels, Code Sample, Converting Images, Convolution matrix, Feature extraction, Gaussian, Gaussian Blur, Gaussian Formula, Gaussian Kernels, Graphic Filters, Image, Image Blur, Image processing, Image Smoothing, Image Transform, Machine vision, Matrix, Pixel Filters

Article Purpose

This purpose of this article is to explain and illustrate in detail the requirements involved in calculating Gaussian Kernels intended for use in image convolution when implementing Gaussian Blur filters. This article’s discussion spans from exploring concepts in theory and continues on to implement concepts through C# sample source code.

Ant: Gaussian Kernel 5×5 Weight 19

Sample Source Code

This article is accompanied by a sample source code Visual Studio project which is available for download here

Using the Sample Application

A Sample Application forms part of the accompanying sample source code, intended to implement the topics discussed and also provides the means to replicate and test the concepts being illustrated.

The sample application is a Windows Forms based application which provides functionality enabling users to generate/calculate Gaussian Kernels. Calculation results are influenced through user specified options in the form of: Kernel Size and Weight.

Ladybird: Gaussian Kernel 5×5 Weight 5.5

In the sample application and related sample source code when referring to Kernel Size, a reference is being made relating to the physical size dimensions of the kernel/matrix used in convolution. When higher values are specified in setting the Kernel Size, the resulting output image will reflect a greater degree of blurring. Kernel Sizes being specified as lower values result in the output image reflecting a lesser degree of blurring.

In a similar fashion to the Kernel size value, the Weight value provided when generating a Kernel results in smoother/more blurred images when specified as higher values. Lower values assigned to the Weight value has the expected result of less blurring being evident in output images.

Prey Mantis: Gaussian Kernel 13×13 Weight 13

The sample application has the ability to provide the user with a visual representation implementing the calculated kernel value blurring. Users are able to select source/input image from the local file system by clicking the Load Image button. When desired, users are able to save blurred/filtered images to the local file system by clicking the Save Image button.

The image below is screenshot of the Gaussian Kernel Calculator sample application in action:

Calculating Gaussian Convolution Kernels

The formula implemented in calculating Gaussian Kernels can be implemented in C# source code fairly easily. Once the method in which the formula operates has been grasped the actual code implementation becomes straight forward.

The Gaussian Kernel formula can be expressed as follows:

The formula contains a number of symbols, which define how the filter will be implemented. The symbols forming part of the Gaussian Kernel formula are described in the following list:

G(x y) – A value calculated using the Gaussian Kernel formula. This value forms part of a Kernel, representing a single element.
π – Pi, one of the better known members of the Greek alphabet. The mathematical constant defined as 22 / 7.
σ – The lower case version of the Greek alphabet letter Sigma. This symbol simply represents a threshold or factor value, as specified by the user.
e – The formula references a lower case e symbol. The symbol represents Euler’s number. The value of Euler’s number has been defined as a mathematical constant equating to 2.71828182846.
x, y – The variables referenced as x and y relate to pixel coordinates within an image. y Representing the vertical offset or row and x represents the horizontal offset or column.

Note: The formula’s implementation expects x and y to equal zero values when representing the coordinates of the pixel located in the middle of the kernel.

Ladybird: Gaussian Kernel 13×13 Weight 9.5

When calculating the kernel elements, the coordinate values expressed by x and y should reflect the distance in pixels from the middle pixel. All coordinate values must be greater than zero.

In order to gain a better grasp on the Gaussian kernel formula we can implement the formula in steps. If we were to create a 3×3 kernel and specified a weighting value of 5.5 our calculations can start off as indicated by the following illustration:

The formula has been implement on each element forming part of the kernel, 9 values in total. Coordinate values have now been replaced with actual values, differing for each position/element. Calculating zero to the power of two equates to zero. In the scenario above indicating zeros which express exponential values might help to ease initial understanding, as opposed to providing simplified values and potentially causing confusing scenarios. The following image illustrates the calculated values of each kernel element:

Ant: Gaussian Kernel 9×9 Weight 19

An important requirement to take note of at this point being that the sum total of all the elements contained as part of a kernel/matrix must equate to one. Looking at our calculated results that is not the case. The kernel needs to be modified in order to satisfy the requirement of having a sum total value of 1 when adding together all the elements of the kernel.

At this point the sum total of the kernel equates to 0.046322548968. We can correct the kernel values, ensuring the sum total of all kernel elements equate to 1. The kernel values should be updated by multiplying each element by one divided by the current kernel sum. In other words each item should be multiplied by:

1.0 / 0.046322548968

After updating the kernel by multiplying each element with the values mentioned above, the result as follows:

We have now successfully calculated a 3×3 Gaussian Blur kernel matrix which implements a weight value of 5.5. Implementing the Gaussian blur has the following effect:

Rose: Gaussian Kernel 3×3 Weight 5.5

The Original Image

The calculated Gaussian Kernel can now be implemented when performing image convolution.

Implementing Gaussian Kernel Calculations

In this section of the article we will be exploring how to implement Gaussian Blur kernel calculations in terms of C# code. Defined as part of the sample source code the definition of the static MatrixCalculator class, exposing the static Calculate method. All of the formula calculation tasks discussed in the previous section have been implemented within this method.

As parameter values the method expects a value indicating the kernel size and a value representing the Weight value. The Calculate method returns a two dimensional array of type double. The return value array represents the calculated kernel.

The definition of the MatrixCalculator.Calculate method as follows:

public static double[,] Calculate(int length, double weight) 
{
    double[,] Kernel = new double [length, lenght]; 
    double sumTotal = 0; 

  
    int kernelRadius = lenght / 2; 
    double distance = 0; 

  
    double calculatedEuler = 1.0 /  
    (2.0 * Math.PI * Math.Pow(weight, 2)); 

  
    for (int filterY = -kernelRadius; 
         filterY <= kernelRadius; filterY++) 
    {
        for (int filterX = -kernelRadius; 
            filterX <= kernelRadius; filterX++) 
        {
            distance = ((filterX * filterX) +  
                       (filterY * filterY)) /  
                       (2 * (weight * weight)); 

  
            Kernel[filterY + kernelRadius,  
                   filterX + kernelRadius] =  
                   calculatedEuler * Math.Exp(-distance); 

  
            sumTotal += Kernel[filterY + kernelRadius,  
                               filterX + kernelRadius]; 
        } 
    } 

  
    for (int y = 0; y < lenght; y++) 
    { 
        for (int x = 0; x < lenght; x++) 
        { 
            Kernel[y, x] = Kernel[y, x] *  
                           (1.0 / sumTotal); 
        } 
    } 

  
    return Kernel; 
}

Ladybird: Gaussian Kernel 19×19 Weight 9.5

The sample source code provides the definition of the ConvolutionFilter extension method, targeting the Bitmap class. This method accepts as a parameter a two dimensional array representing the matrix kernel to implement when performing image convolution. The matrix kernel value passed to this function originates from the calculated Gaussian kernel.

Detailed below is the definition of the ConvolutionFilter extension method:

public static Bitmap ConvolutionFilter(this Bitmap sourceBitmap,  
                                         double[,] filterMatrix,  
                                              double factor = 1,  
                                                   int bias = 0)  
{ 
    BitmapData sourceData = sourceBitmap.LockBits(new Rectangle(0, 0, 
                            sourceBitmap.Width, sourceBitmap.Height), 
                                              ImageLockMode.ReadOnly,  
                                        PixelFormat.Format32bppArgb); 

   
    byte[] pixelBuffer = new byte[sourceData.Stride * sourceData.Height]; 
    byte[] resultBuffer = new byte[sourceData.Stride * sourceData.Height]; 

   
    Marshal.Copy(sourceData.Scan0, pixelBuffer, 0, pixelBuffer.Length); 
    sourceBitmap.UnlockBits(sourceData); 

   
    double blue = 0.0; 
    double green = 0.0; 
    double red = 0.0; 

   
    int filterWidth = filterMatrix.GetLength(1); 
    int filterHeight = filterMatrix.GetLength(0); 

   
    int filterOffset = (filterWidth-1) / 2; 
    int calcOffset = 0; 

   
    int byteOffset = 0; 

   
    for (int offsetY = filterOffset; offsetY <  
        sourceBitmap.Height - filterOffset; offsetY++) 
    {
        for (int offsetX = filterOffset; offsetX <  
            sourceBitmap.Width - filterOffset; offsetX++) 
        {
            blue = 0; 
            green = 0; 
            red = 0; 

   
            byteOffset = offsetY *  
                         sourceData.Stride +  
                         offsetX * 4; 

   
            for (int filterY = -filterOffset;  
                filterY <= filterOffset; filterY++) 
            { 
                for (int filterX = -filterOffset; 
                    filterX <= filterOffset; filterX++) 
                { 

   
                    calcOffset = byteOffset +  
                                 (filterX * 4) +  
                                 (filterY * sourceData.Stride); 

   
                    blue += (double  )(pixelBuffer[calcOffset]) * 
                            filterMatrix[filterY + filterOffset,  
                                                filterX + filterOffset]; 

   
                    green += (double  )(pixelBuffer[calcOffset + 1]) * 
                             filterMatrix[filterY + filterOffset,  
                                                filterX + filterOffset]; 

   
                    red += (double  )(pixelBuffer[calcOffset + 2]) * 
                           filterMatrix[filterY + filterOffset,  
                                              filterX + filterOffset]; 
                } 
            } 

   
            blue = factor * blue + bias; 
            green = factor * green + bias; 
            red = factor * red + bias; 

   
            blue = (blue > 255 ? 255 : (blue < 0 ? 0 : blue)); 
            green = (green > 255 ? 255 : (green < 0 ? 0 : green)); 
            red = (red > 255 ? 255 : (red < 0 ? 0 : blue)); 

   
            resultBuffer[byteOffset] = (byte)(blue); 
            resultBuffer[byteOffset + 1] = (byte)(green); 
            resultBuffer[byteOffset + 2] = (byte)(red); 
            resultBuffer[byteOffset + 3] = 255; 
        }
    }

   
    Bitmap resultBitmap = new Bitmap(sourceBitmap.Width, sourceBitmap.Height); 

   
    BitmapData resultData = resultBitmap.LockBits(new Rectangle (0, 0, 
                             resultBitmap.Width, resultBitmap.Height), 
                                              ImageLockMode.WriteOnly, 
                                         PixelFormat.Format32bppArgb); 

   
    Marshal.Copy(resultBuffer, 0, resultData.Scan0, resultBuffer.Length); 
    resultBitmap.UnlockBits(resultData); 

   
    return resultBitmap; 
}

Ant: Gaussian Kernel 7×7 Weight 19

Sample Images

This article features a number of sample images. All featured images have been licensed allowing for reproduction.

The sample images featuring an image of a prey mantis is licensed under the Creative Commons Attribution-Share Alike 3.0 Unported license and can be downloaded from Wikipedia.

The sample images featuring an image of an ant has been released into the public domain by its author, Sean.hoyland. This applies worldwide. In some countries this may not be legally possible; if so: Sean.hoyland grants anyone the right to use this work for any purpose, without any conditions, unless such conditions are required by law. The original image can be downloaded from Wikipedia.

The sample images featuring an image of a ladybird (ladybug or lady beetle) is licensed under the Creative Commons Attribution-Share Alike 2.0 Generic license and can be downloaded from Wikipedia.

The sample images featuring an image of a wasp is licensed under the Creative Commons Attribution-Share Alike 3.0 Unported license and can be downloaded from Wikimedia.org.

The Original Image

Wasp Gaussian Kernel 3×3 Weight 9.25

Wasp Gaussian Kernel 5×5 Weight 9.25

Wasp Gaussian Kernel 7×7 Weight 9.25

Wasp Gaussian Kernel 9×9 Weight 9.25

Wasp Gaussian Kernel 11×11 Weight 9.25

Wasp Gaussian Kernel 13×13 Weight 9.25

Wasp Gaussian Kernel 15×15 Weight 9.25

Wasp Gaussian Kernel 17×17 Weight 9.25

Wasp Gaussian Kernel 19×19 Weight 9.25

C# How to: Image Unsharp Mask

Published May 18, 2013 Augmented Reality , C# , Code Samples , Extension Methods , Graphic Filters , Graphics , How to , Image Convolution , Image Filters , Microsoft , Opensource , Tip 32 Comments
Tags: Augmented Reality, Bitmap ARGB, Bitmap Filters, Bitmap.LockBits, BitmapData, BitmapData.Stride, C#, C# Convolution, Code Sample, Computer vision, Convolution filters, Convolution Kernel, Convolution matrix, Edge enhance, Extension Methods, Feature extraction, Gaussian, Gaussian Blur, Graphic Filters, Image Blur, Image processing, Image Sharpness, Image Transform, Machine vision, Matrix, Mean Filter, Pixel Filters, Pixel Manipulation, Unsharp, Unsharp Masking

Article purpose

The purpose of this article is to explore and illustrate the concept of Image Unsharp Masking. This article implements image convolution in the form of a 3×3 Gaussian Blur, 5×5 Gaussian Blur, 3×3 Mean filter and a 5×5 Mean filter.

Sample Source code

This article is accompanied by a sample source code Visual Studio project which is available for download here.

Using the Sample Application

The sample source code associated with this article includes a Windows Forms based sample application implementing the concepts explored throughout this article.

When using the Image Unsharp Mask sample application users can select a source/input image from the local file system by clicking the Load Image button. The dropdown combobox at the bottom of the screen allows the user to select an unsharp masking variation. On the right hand side of the screen users can specify the level/intensity of resulting image sharpness.

Clicking the Save Image button allows a user to save resulting images to the local file system. The image below is a screenshot of the Image Unsharp Mask sample application in action:

What is Image Unsharp Masking?

A good definition of Image Unsharp Masking can be found on Wikipedia:

Unsharp masking (USM) is an image manipulation technique, often available in digital image processing software.

The "unsharp" of the name derives from the fact that the technique uses a blurred, or "unsharp", positive image to create a "mask" of the original image. The unsharped mask is then combined with the negative image, creating an image that is less blurry than the original. The resulting image, although clearer, probably loses accuracy with respect to the image’s subject. In the context of signal-processing, an unsharp mask is generally a linear or nonlinear filter that amplifies high-frequency components.

In this article we implement Image Unsharp Masking by first creating a blurred copy of a source/input image then subtracting the blurred image from the original image, which is known as the mask. Increased image sharpness is achieved by adding a factor of the mask to the original image.

Applying a Convolution Matrix filter

The sample source code provides the definition for the ConvolutionFilter extension method targeting the Bitmap class. This method is invoked when implementing image blurring. The definition of the ConvolutionFilter extension method as follows:

 private static Bitmap ConvolutionFilter(Bitmap sourceBitmap,  
                                     double[,] filterMatrix,  
                                          double factor = 1,  
                                               int bias = 0,  
                                     bool grayscale = false )  
{ 
     BitmapData sourceData = sourceBitmap.LockBits(new Rectangle (0, 0, 
                              sourceBitmap.Width, sourceBitmap.Height), 
                                                ImageLockMode.ReadOnly,  
                                          PixelFormat.Format32bppArgb); 

  
     byte[] pixelBuffer = new byte[sourceData.Stride * sourceData.Height]; 
     byte[] resultBuffer = new byte[sourceData.Stride * sourceData.Height]; 

  
     Marshal.Copy(sourceData.Scan0, pixelBuffer, 0, pixelBuffer.Length); 
     sourceBitmap.UnlockBits(sourceData); 

  
     if (grayscale == true) 
     {
         float rgb = 0; 

  
         for (int k = 0; k < pixelBuffer.Length; k += 4) 
         { 
             rgb = pixelBuffer[k] * 0.11f; 
             rgb += pixelBuffer[k + 1] * 0.59f; 
             rgb += pixelBuffer[k + 2] * 0.3f; 

  
             pixelBuffer[k] = (byte )rgb; 
             pixelBuffer[k + 1] = pixelBuffer[k]; 
             pixelBuffer[k + 2] = pixelBuffer[k]; 
             pixelBuffer[k + 3] = 255; 
         } 
     } 

  
     double blue = 0.0; 
     double green = 0.0; 
     double red = 0.0; 

  
     int filterWidth = filterMatrix.GetLength(1); 
     int filterHeight = filterMatrix.GetLength(0); 

  
     int filterOffset = (filterWidth-1) / 2; 
     int calcOffset = 0; 

  
     int byteOffset = 0; 

  
     for (int offsetY = filterOffset; offsetY <  
         sourceBitmap.Height - filterOffset; offsetY++) 
     { 
         for (int  offsetX = filterOffset; offsetX <  
             sourceBitmap.Width - filterOffset; offsetX++) 
         { 
             blue = 0; 
             green = 0; 
             red = 0; 

  
             byteOffset = offsetY *  
                          sourceData.Stride +  
                          offsetX * 4; 

   
             for (int  filterY = -filterOffset;  
                 filterY <= filterOffset; filterY++) 
             {
                 for (int  filterX = -filterOffset; 
                     filterX <= filterOffset; filterX++) 
                 { 

   
                     calcOffset = byteOffset +  
                                  (filterX * 4) +  
                                  (filterY * sourceData.Stride); 

  
                     blue += (double)(pixelBuffer[calcOffset]) * 
                             filterMatrix[filterY + filterOffset,  
                                          filterX + filterOffset]; 

  
                     green += (double)(pixelBuffer[calcOffset + 1]) * 
                              filterMatrix[filterY + filterOffset,  
                                           filterX + filterOffset]; 

   
                     red += (double)(pixelBuffer[calcOffset + 2]) * 
                            filterMatrix[filterY + filterOffset,  
                                         filterX + filterOffset]; 
                 }
             }

   
             blue = factor * blue + bias; 
             green = factor * green + bias; 
             red = factor * red + bias; 

  
             if (blue > 255) 
             { blue = 255; } 
             else if (blue < 0) 
             { blue = 0; } 

  
             if (green > 255) 
             { green = 255; } 
             else if (green < 0) 
             { green = 0; } 

  
             if (red > 255) 
             { red = 255; } 
             else if (red < 0) 
             { red = 0; } 

  
             resultBuffer[byteOffset] = (byte )(blue); 
             resultBuffer[byteOffset + 1] = (byte )(green); 
             resultBuffer[byteOffset + 2] = (byte )(red); 
             resultBuffer[byteOffset + 3] = 255; 
        }
    } 

  
     Bitmap resultBitmap = new Bitmap(sourceBitmap.Width,
                                      sourceBitmap.Height); 
  
     BitmapData resultData = resultBitmap.LockBits(new Rectangle (0, 0, 
                              resultBitmap.Width, resultBitmap.Height), 
                                               ImageLockMode.WriteOnly, 
                                          PixelFormat.Format32bppArgb); 

  
     Marshal.Copy(resultBuffer, 0, resultData.Scan0, resultBuffer.Length); 
     resultBitmap.UnlockBits(resultData); 

   
     return resultBitmap; 
}

Subtracting and Adding Images

An important step required when implementing Image Unsharp Masking comes in the form of creating a mask by subtracting a blurred copy from the original image and then adding a factor of the mask to the original image. In order to achieve increased performance the sample source code combines the process of creating the mask and adding the mask to the original image.

The SubtractAddFactorImage extension method iterates every pixel that forms part of an image. In a single step the blurred pixel is subtracted from the original pixel, multiplied by a user specified factor and then added to the original pixel. The definition of the SubtractAddFactorImage extension method as follows:

private static Bitmap SubtractAddFactorImage( 
                              this Bitmap subtractFrom, 
                                  Bitmap subtractValue, 
                                   float factor = 1.0f) 
{ 
    BitmapData sourceData =  
               subtractFrom.LockBits(new Rectangle (0, 0, 
               subtractFrom.Width, subtractFrom.Height), 
               ImageLockMode.ReadOnly, 
               PixelFormat.Format32bppArgb); 

  
    byte[] sourceBuffer = new byte[sourceData.Stride *  
                                   sourceData.Height]; 

  
    Marshal.Copy(sourceData.Scan0, sourceBuffer, 0,  
                                sourceBuffer.Length); 

  
    byte[] resultBuffer = new byte[sourceData.Stride *  
                                   sourceData.Height]; 

  
    BitmapData subtractData =  
               subtractValue.LockBits(new Rectangle (0, 0, 
               subtractValue.Width, subtractValue.Height), 
               ImageLockMode.ReadOnly, 
               PixelFormat.Format32bppArgb); 

  
    byte[] subtractBuffer = new byte[subtractData.Stride * 
                                     subtractData.Height]; 

  
    Marshal.Copy(subtractData.Scan0, subtractBuffer, 0, 
                                 subtractBuffer.Length); 

  
    subtractFrom.UnlockBits(sourceData); 
    subtractValue.UnlockBits(subtractData); 

  
    double blue = 0; 
    double green = 0; 
    double red = 0; 

  
    for (int k = 0; k < resultBuffer.Length && 
                   k < subtractBuffer.Length; k += 4) 
    { 
        blue = sourceBuffer[k] +  
              (sourceBuffer[k] - 
               subtractBuffer[k]) * factor; 

  
        green = sourceBuffer[k + 1] +  
               (sourceBuffer[k + 1] - 
                subtractBuffer[k + 1]) * factor; 

  
        red = sourceBuffer[k + 2] +  
             (sourceBuffer[k + 2] - 
              subtractBuffer[k + 2]) * factor; 

  
        blue = (blue < 0 ? 0 : (blue > 255 ? 255 : blue)); 
        green = (green < 0 ? 0 : (green > 255 ? 255 : green)); 
        red = (red < 0 ? 0 : (red > 255 ? 255 : red)); 

  
        resultBuffer[k] = (byte )blue; 
        resultBuffer[k + 1] = (byte )green; 
        resultBuffer[k + 2] = (byte )red; 
        resultBuffer[k + 3] = 255; 
    }

  
    Bitmap resultBitmap = new Bitmap (subtractFrom.Width,  
                                     subtractFrom.Height); 

  
    BitmapData resultData =  
               resultBitmap.LockBits(new Rectangle (0, 0, 
               resultBitmap.Width, resultBitmap.Height), 
               ImageLockMode.WriteOnly, 
               PixelFormat.Format32bppArgb); 

  
    Marshal.Copy(resultBuffer, 0, resultData.Scan0,  
                               resultBuffer.Length); 

  
    resultBitmap.UnlockBits(resultData); 

  
    return resultBitmap; 
}

Matrix Definition

The image blurring convolution filters implemented by the sample source code relies on static matrix/kernel values defined in the Matrix class. The variants of blurring implemented are: 3×3 Gaussian, 5×5 Gaussian, 3×3 Mean and 5×5 Mean. The definition of the Matrix class is detailed by the following code snippet:

public static class Matrix
{
    public static double[,] Gaussian3x3
    {
        get
        {
            return new double[,]
            { { 1, 2, 1, }, 
              { 2, 4, 2, }, 
              { 1, 2, 1, }, };
        }
    }

  
    public static double[,] Gaussian5x5Type1
    {
        get
        {
            return new double[,]
            { { 2, 04, 05, 04, 2 }, 
              { 4, 09, 12, 09, 4 },
              { 5, 12, 15, 12, 5 },
              { 4, 09, 12, 09, 4 },
              { 2, 04, 05, 04, 2 }, };
        }
    }

  
    public static double[,] Mean3x3
    {
        get
        {
            return new double[,]
            { { 1, 1, 1, }, 
              { 1, 1, 1, }, 
              { 1, 1, 1, }, };
        }
    }

  
    public static double[,] Mean5x5
    {
        get
        {
            return new double[,]
            { { 1, 1, 1, 1, 1 }, 
              { 1, 1, 1, 1, 1 },
              { 1, 1, 1, 1, 1 },
              { 1, 1, 1, 1, 1 },
              { 1, 1, 1, 1, 1 }, };
        }
    }
}

Implementing Image Unsharpening

This article explores four variants of Image Unsharp Masking, relating to the four types of image blurring discussed in the previous section. The sample source code defines the following Image Unsharp Masking extension methods: UnsharpGaussian3x3, UnsharpGaussian5x5, UnsharpMean3x3 and UnsharpMean5x5. All four methods are defined as extension methods targeting the Bitmap class. When looking at the sample images in the following section you will notice the correlation between increased image blurring and enhanced image sharpness. The definition as follows:

public static Bitmap UnsharpGaussian3x3( 
                                 this Bitmap sourceBitmap,  
                                 float factor = 1.0f) 
{
    Bitmap blurBitmap = ExtBitmap.ConvolutionFilter( 
                                  sourceBitmap,  
                                  Matrix.Gaussian3x3,  
                                  1.0 / 16.0); 

   
    Bitmap resultBitmap =  
           sourceBitmap.SubtractAddFactorImage( 
                        blurBitmap, factor); 

   
    return resultBitmap; 
 } 

   
public static Bitmap UnsharpGaussian5x5( 
                                 this Bitmap sourceBitmap, 
                                 float factor = 1.0f) 
{
    Bitmap blurBitmap = ExtBitmap.ConvolutionFilter( 
                                  sourceBitmap,  
                                  Matrix.Gaussian5x5Type1,  
                                  1.0 / 159.0); 

   
    Bitmap resultBitmap = 
           sourceBitmap.SubtractAddFactorImage( 
                        blurBitmap, factor); 

   
    return resultBitmap; 
} 
 
public static Bitmap UnsharpMean3x3( 
                                 this Bitmap sourceBitmap, 
                                 float factor = 1.0f) 
{ 
    Bitmap blurBitmap = ExtBitmap.ConvolutionFilter( 
                                  sourceBitmap,  
                                  Matrix.Mean3x3,  
                                  1.0 / 9.0); 

   
    Bitmap resultBitmap = 
           sourceBitmap.SubtractAddFactorImage( 
                        blurBitmap, factor); 

   
    return resultBitmap; 
}

   
public static Bitmap UnsharpMean5x5( 
                                 this Bitmap sourceBitmap, 
                                 float factor = 1.0f) 
{ 
    Bitmap blurBitmap = ExtBitmap.ConvolutionFilter( 
                                  sourceBitmap,  
                                  Matrix.Mean5x5,  
                                  1.0 / 25.0); 

   
    Bitmap resultBitmap = 
           sourceBitmap.SubtractAddFactorImage( 
                        blurBitmap, factor); 

   
    return resultBitmap; 
}

Sample Images

The input/source image used in rendering the sample images shown in this article is licensed under the Creative Commons Attribution-Share Alike 3.0 Unported license and can be downloaded from Wikipedia:

The Original Image

Unsharp Gaussian 3×3

Unsharp Gaussian 5×5

Unsharp Mean 3×3

Unsharp Gaussian 5×5

C# How to: Difference Of Gaussians

Published May 18, 2013 Augmented Reality , C# , Code Samples , Edge Detection , Extension Methods , Graphic Filters , Graphics , How to , Image Convolution , Image Filters , Microsoft , New Version , Opensource 31 Comments
Tags: Augmented Reality, Bitmap ARGB, Bitmap Filters, Bitmap.LockBits, BitmapData, Blob detection, C#, Change detection, Code Sample, Computer vision, Convolution filters, Convolution Kernel, Convolution matrix, Difference of Gaussians, DoG, DoG Edge Detection, Edge Detect, Edge enhance, Feature extraction, Gaussian, Gaussian Blur, Graphic Filters, Image Noise, Image processing, Image Transform, Machine vision, Matrix, Pixel Filters, Pixel Manipulation, Step detection

Article purpose

In this article we explore the concept of Difference of Gaussians edge detection. This article implements image convolution as a means of achieving Gaussian blurring. All of the concepts explored are implemented by accessing and manipulating the raw pixel data exposed by an image, no GDI+ or conventional drawing code is required.

Sample source code

This article is accompanied by a sample source code Visual Studio project which is available for download here.

Using the Sample Application

The concepts explored in this article can be easily replicated by making use of the Sample Application, which forms part of the associated sample source code accompanying this article.

When using the Difference Of Gaussians sample application you can specify a input/source image by clicking the Load Image button. The dropdown combobox towards the bottom middle part of the screen relates the various edge detection methods discussed.

If desired a user can save the resulting edge detection image to the local file system by clicking the Save Image button.

The following image is screenshot of the Difference Of Gaussians sample application in action:

What is Difference of Gaussians?

Difference of Gaussians, commonly abbreviated as DoG, is a method of implementing image edge detection. Central to the Difference of Gaussians method of edge detection is the application of Gaussian image blur.

From Wikipedia we gain the following explanation:

In imaging science, Difference of Gaussians is a feature enhancement algorithm that involves the subtraction of one blurred version of an original image from another, less blurred version of the original. In the simple case of grayscale images, the blurred images are obtained by convolving the original grayscale images with Gaussian kernels having differing standard deviations. Blurring an image using a Gaussian kernel suppresses only high-frequency spatial information. Subtracting one image from the other preserves spatial information that lies between the range of frequencies that are preserved in the two blurred images. Thus, the difference of Gaussians is a band-pass filter that discards all but a handful of spatial frequencies that are present in the original grayscale image.

In simple terms Difference of Gaussians can be implemented by applying two Gaussian blurs of different intensity levels to the same source image. The resulting image is then created by subtracting the two images of different Gaussian blurring.

Applying a Convolution Matrix filter

In the sample source code accompanying this article Gaussian image blurring is applied by invoking the ConvolutionFilter method. This method accepts a two dimensional array of type double representing the convolution matrix/kernel. This method is also capable of first converting source images to grayscale, which can be specified as a method parameter. Resulting images sometimes tend to be very dark, which can be corrected by specifying a suitable bias value.

The following code snippet provides the implementation of the ConvolutionFilter method:

 private static Bitmap ConvolutionFilter(Bitmap sourceBitmap,  
                                     double[,] filterMatrix,  
                                          double factor = 1,  
                                               int bias = 0,  
                                     bool grayscale = false )  
{ 
     BitmapData sourceData = sourceBitmap.LockBits(new Rectangle (0, 0, 
                              sourceBitmap.Width, sourceBitmap.Height), 
                                                ImageLockMode.ReadOnly,  
                                          PixelFormat.Format32bppArgb); 

  
     byte[] pixelBuffer = new byte[sourceData.Stride * sourceData.Height]; 
     byte[] resultBuffer = new byte[sourceData.Stride * sourceData.Height]; 

  
     Marshal.Copy(sourceData.Scan0, pixelBuffer, 0, pixelBuffer.Length); 
     sourceBitmap.UnlockBits(sourceData); 

  
     if (grayscale == true) 
     {
         float rgb = 0; 

  
         for (int k = 0; k < pixelBuffer.Length; k += 4) 
         { 
             rgb = pixelBuffer[k] * 0.11f; 
             rgb += pixelBuffer[k + 1] * 0.59f; 
             rgb += pixelBuffer[k + 2] * 0.3f; 

  
             pixelBuffer[k] = (byte )rgb; 
             pixelBuffer[k + 1] = pixelBuffer[k]; 
             pixelBuffer[k + 2] = pixelBuffer[k]; 
             pixelBuffer[k + 3] = 255; 
         } 
     } 

  
     double blue = 0.0; 
     double green = 0.0; 
     double red = 0.0; 

  
     int filterWidth = filterMatrix.GetLength(1); 
     int filterHeight = filterMatrix.GetLength(0); 

  
     int filterOffset = (filterWidth-1) / 2; 
     int calcOffset = 0; 

  
     int byteOffset = 0; 

  
     for (int offsetY = filterOffset; offsetY <  
         sourceBitmap.Height - filterOffset; offsetY++) 
     { 
         for (int  offsetX = filterOffset; offsetX <  
             sourceBitmap.Width - filterOffset; offsetX++) 
         { 
             blue = 0; 
             green = 0; 
             red = 0; 

  
             byteOffset = offsetY *  
                          sourceData.Stride +  
                          offsetX * 4; 

   
             for (int  filterY = -filterOffset;  
                 filterY <= filterOffset; filterY++) 
             {
                 for (int  filterX = -filterOffset; 
                     filterX <= filterOffset; filterX++) 
                 { 

   
                     calcOffset = byteOffset +  
                                  (filterX * 4) +  
                                  (filterY * sourceData.Stride); 

  
                     blue += (double)(pixelBuffer[calcOffset]) * 
                             filterMatrix[filterY + filterOffset,  
                                          filterX + filterOffset]; 

  
                     green += (double)(pixelBuffer[calcOffset + 1]) * 
                              filterMatrix[filterY + filterOffset,  
                                           filterX + filterOffset]; 

   
                     red += (double)(pixelBuffer[calcOffset + 2]) * 
                            filterMatrix[filterY + filterOffset,  
                                         filterX + filterOffset]; 
                 }
             }

   
             blue = factor * blue + bias; 
             green = factor * green + bias; 
             red = factor * red + bias; 

  
             if (blue > 255) 
             { blue = 255; } 
             else if (blue < 0) 
             { blue = 0; } 

  
             if (green > 255) 
             { green = 255; } 
             else if (green < 0) 
             { green = 0; } 

  
             if (red > 255) 
             { red = 255; } 
             else if (red < 0) 
             { red = 0; } 

  
             resultBuffer[byteOffset] = (byte )(blue); 
             resultBuffer[byteOffset + 1] = (byte )(green); 
             resultBuffer[byteOffset + 2] = (byte )(red); 
             resultBuffer[byteOffset + 3] = 255; 
        }
    } 

  
     Bitmap resultBitmap = new Bitmap(sourceBitmap.Width,
                                      sourceBitmap.Height); 
  
     BitmapData resultData = resultBitmap.LockBits(new Rectangle (0, 0, 
                              resultBitmap.Width, resultBitmap.Height), 
                                               ImageLockMode .WriteOnly, 
                                          PixelFormat.Format32bppArgb); 

  
     Marshal.Copy(resultBuffer, 0, resultData.Scan0, resultBuffer.Length); 
     resultBitmap.UnlockBits(resultData); 

   
     return resultBitmap; 
}

The Gaussian Matrix

The sample source code defines three Gaussian matrix/kernel values, a 3×3 matrix and two slightly different 5×5 matrices. The Gaussian3x3 matrix requires a factor of 1 / 16, the Gaussian5x5Type1 matrix a factor of 1 / 159 and the factor required by the Gaussian5x5Type2 equates to 1 / 256.

public static class Matrix
{
    public static double[,] Gaussian3x3
    {
        get
        {
            return new double[,]  
            { { 1, 2, 1, }, 
              { 2, 4, 2, },
              { 1, 2, 1, }, };
        }
    }

  
    public static double[,] Gaussian5x5Type1
    {
        get  
        {
            return new double[,]  
             { { 2, 04, 05, 04, 2  }, 
               { 4, 09, 12, 09, 4  }, 
               { 5, 12, 15, 12, 5  },
               { 4, 09, 12, 09, 4  },
               { 2, 04, 05, 04, 2  }, };
        }
    }

  
    public static double[,] Gaussian5x5Type2
    {
        get 
        {
            return new   double[,] 
             { {  1,   4,  6,  4,  1  }, 
               {  4,  16, 24, 16,  4  }, 
               {  6,  24, 36, 24,  6  },
               {  4,  16, 24, 16,  4  },
               {  1,   4,  6,  4,  1  }, };
        }
    }
}

Subtracting Images

When implementing the Difference of Gaussians method of edge detection after having applied two varying levels of Gaussian blurring the resulting images need to be subtracted. The sample source code associated with this article implements the SubtractImage extension method when subtracting images.

The following code snippet details the implementation of the SubtractImage extension method:

private static void SubtractImage(this Bitmap subtractFrom,  
                                  Bitmap subtractValue,  
                                  bool invert = false,
                                  int bias = 0) 
{ 
    BitmapData sourceData = 
               subtractFrom.LockBits(new Rectangle(0, 0, 
               subtractFrom.Width, subtractFrom.Height), 
               ImageLockMode.ReadWrite, 
               PixelFormat.Format32bppArgb); 

   
    byte[] resultBuffer = new byte[sourceData.Stride *
                                  sourceData.Height]; 

  
    Marshal.Copy(sourceData.Scan0, resultBuffer, 0,
                               resultBuffer.Length); 

   
    BitmapData subtractData =
               subtractValue.LockBits(new Rectangle(0, 0, 
               subtractValue.Width, subtractValue.Height), 
               ImageLockMode.ReadOnly, 
               PixelFormat.Format32bppArgb); 

   
    byte[] subtractBuffer = new byte[subtractData.Stride *  
                                     subtractData.Height]; 

   
    Marshal.Copy(subtractData.Scan0, subtractBuffer, 0,  
                                 subtractBuffer.Length); 

   
    subtractValue.UnlockBits(subtractData); 

   
    int blue = 0; 
    int green = 0; 
    int red = 0; 

   
    for (int k = 0; k < resultBuffer.Length &&  
                    k < subtractBuffer.Length; k += 4) 
    { 
        if (invert == true ) 
        { 
            blue = 255 - resultBuffer[k] -  
                   subtractBuffer[k] + bias; 

   
            green = 255 - resultBuffer[k + 1] -  
                    subtractBuffer[k + 1] + bias; 

   
            red = 255 - resultBuffer[k + 2] -  
                  subtractBuffer[k + 2] + bias; 
        } 
        else  
        { 
            blue = resultBuffer[k] -  
                   subtractBuffer[k] + bias; 

   
            green = resultBuffer[k + 1] -  
                    subtractBuffer[k + 1] + bias; 

   
            red = resultBuffer[k + 2] -  
                  subtractBuffer[k + 2] + bias; 
        } 

   
        blue = (blue < 0 ? 0 : (blue > 255 ? 255 : blue)); 
        green = (green < 0 ? 0 : (green > 255 ? 255 : green)); 
        red = (red < 0 ? 0 : (red > 255 ? 255 : red)); 

   
        resultBuffer[k] = (byte )blue; 
        resultBuffer[k + 1] = (byte )green; 
        resultBuffer[k + 2] = (byte )red; 
        resultBuffer[k + 3] = 255; 
    } 

   
    Marshal.Copy(resultBuffer, 0, sourceData.Scan0,  
                               resultBuffer.Length); 

   
    subtractFrom.UnlockBits(sourceData); 
}

Difference of Gaussians Extension methods

The sample source code implements Difference of Gaussians edge detection by means of two extension methods: DifferenceOfGaussians3x5Type1 and DifferenceOfGaussians3x5Type2. Both methods are virtually identical, the only difference being the 5×5 Gaussian matrix being implemented.

Both methods create two new images, each having a Gaussian blur of different levels of intensity applied. The two new images are subtracted in order to create a single resulting image.

The following source code snippet provides the implementation of the DifferenceOfGaussians3x5Type1 and DifferenceOfGaussians3x5Type2 extension methods:

public static Bitmap DifferenceOfGaussians3x5Type1(
                                  this Bitmap sourceBitmap,  
                                  bool grayscale = false, 
                                  bool invert = false, 
                                  int bias = 0) 
{
    Bitmap bitmap3x3 = ExtBitmap.ConvolutionFilter(sourceBitmap, 
                       Matrix.Gaussian3x3, 1.0 / 16.0, 
                       0, grayscale); 

  
    Bitmap bitmap5x5 = ExtBitmap.ConvolutionFilter(sourceBitmap, 
                       Matrix.Gaussian5x5Type1, 1.0 / 159.0,
                       0, grayscale); 

  
    bitmap3x3.SubtractImage(bitmap5x5, invert, bias); 

  
    return bitmap3x3; 
} 

  
public static Bitmap DifferenceOfGaussians3x5Type2(
                                  this Bitmap sourceBitmap,  
                                  bool grayscale = false, 
                                  bool invert = false, 
                                  int bias = 0) 
{ 
    Bitmap bitmap3x3 = ExtBitmap.ConvolutionFilter(sourceBitmap, 
                       Matrix.Gaussian3x3, 1.0 / 16.0, 
                       0, true ); 

  
    Bitmap bitmap5x5 = ExtBitmap.ConvolutionFilter(sourceBitmap, 
                       Matrix.Gaussian5x5Type2, 1.0 / 256.0,
                       0, true ); 

 
    bitmap3x3.SubtractImage(bitmap5x5, invert, bias); 

  
    return bitmap3x3; 
}

Sample Images

The original sample image used in this article is licensed under the Creative Commons Attribution-Share Alike 2.0 Generic license. The original author is attributed as Andrew Dunn – http://www.andrewdunnphoto.com/

The Original Image

Difference Of Gaussians 3×5 Type1

Difference Of Gaussians 3×5 Type2

Difference Of Gaussians 3×5 Type1 Bias 128

Difference Of Gaussians 3×5 Type 2 Bias 96

Posts Tagged 'Gaussian Blur'

Article Purpose

Sample Source Code

Using the Sample Application

Gaussian Blur

Difference of Gaussians Edge Detection

Difference of Gaussians Edge Detection Required Steps

Calculating Gaussian Convolution Kernels

Implementing Gaussian Kernel Calculations

Implementing Difference of Gaussians Edge Detection

Sample Images

Related Articles and Feedback

Rate this:

Share this:

Article Purpose

Sample Source Code

Using the Sample Application

Image Blur Overview

Mean Filter/Box Blur

Gaussian Blur

Median Filter Blur

Motion Blur

Image Blur Implementation

Sample Images

Related Articles and Feedback

Rate this:

Share this:

Article Purpose

Sample Source Code

Using the Sample Application

Calculating Gaussian Convolution Kernels

Implementing Gaussian Kernel Calculations

Sample Images

Related Articles and Feedback

Rate this:

Share this:

Article purpose

Sample Source code

Using the Sample Application

What is Image Unsharp Masking?

Applying a Convolution Matrix filter

Subtracting and Adding Images

Matrix Definition

Implementing Image Unsharpening

Sample Images

Related Articles and Feedback

Rate this:

Share this:

Article purpose

Sample source code

Using the Sample Application

What is Difference of Gaussians?

Applying a Convolution Matrix filter

The Gaussian Matrix

Subtracting Images

Difference of Gaussians Extension methods

Sample Images

Related Articles and Feedback

Rate this:

Share this:

Dewald Esterhuizen

RSS Feed

Recent Posts

Blog Stats

Follow SoftwareByDefault via Email

Categories

Top Posts & Pages

Archives

C# How to Articles

SoftwareByDefault Online References

MSDN Articles

Categories

Tags