原文:Win8 Metro(C#)数字图像处理--2.64图像高斯滤波算法



[函数名称]

  高斯平滑滤波器      GaussFilter(WriteableBitmap src,int radius,double sigma)

[算法说明]

  高斯滤波器实质上是一种信号的滤波器,其用途是信号的平滑处理。它是一类根据高斯函数的

形状来选择权重的线性平滑滤波器,该滤波器对于抑制服从正态分布的噪声非常有效。高斯函数

的公式如下所示:

Win8 Metro(C#)数字图像处理--2.64图像高斯滤波算法-LMLPHP

private static double[,] GaussFuc(int r, double sigma)
{
int size = 2 * r + 1;
double[,] gaussResult = new double[size, size];
double k = 0.0;
for (int y = -r, h = 0; y <= r; y++, h++)
{
for (int x = -r, w = 0; x <= r; x++, w++)
{
gaussResult[w, h] = (1.0 / (2.0 * Math.PI * sigma * sigma)) * (Math.Exp(-((double)x * (double)x + (double)y * (double)y) / (2.0 * sigma * sigma)));
k += gaussResult[w, h];
}
}
return gaussResult;
}

我们设置参数r=1,sigma=1.0,则得到一个3*3的高斯模板如下:

Win8 Metro(C#)数字图像处理--2.64图像高斯滤波算法-LMLPHP

这个公式可以理解为先对图像按行进行一次一维高斯滤波,在对结果图像按列进行一次一维高斯滤波,这样速度将大大提高。

一维高斯滤波代码如下(包含归一化):

private static double[] GaussKernel1D(int r, double sigma)
{
double[] filter = new double[2 * r + 1];
double sum = 0.0;
for (int i = 0; i < filter.Length; i++)
{
filter[i] = Math.Exp((double)(-(i - r) * (i - r)) / (2.0 * sigma * sigma));
sum += filter[i];
}
for (int i = 0; i < filter.Length; i++)
{
filter[i] = filter[i] / sum;
}
return filter;
}

[函数代码]

        private static double[] GaussKernel(int radius, double sigma)
{
int length=2*radius+1;
double[] kernel = new double[length];
double sum = 0.0;
for (int i = 0; i < length; i++)
{
kernel[i] = Math.Exp((double)(-(i - radius) * (i - radius)) / (2.0 * sigma * sigma));
sum += kernel[i];
}
for (int i = 0; i < length; i++)
{
kernel[i] = kernel[i] / sum;
}
return kernel;
}
/// <summary>
/// Gauss filter process
/// </summary>
/// <param name="src">The source image.</param>
/// <param name="radius">The radius of gauss kernel,from 0 to 100.</param>
/// <param name="sigma">The convince of gauss kernel, from 0 to 30.</param>
/// <returns></returns>
public static WriteableBitmap GaussFilter(WriteableBitmap src,int radius,double sigma) ////高斯滤波
{
if (src != null)
{
int w = src.PixelWidth;
int h = src.PixelHeight;
WriteableBitmap srcImage = new WriteableBitmap(w, h);
byte[] srcValue = src.PixelBuffer.ToArray();
byte[] tempValue=(byte[])srcValue.Clone();
double[] kernel = GaussKernel(radius, sigma);
double tempB = 0.0, tempG = 0.0, tempR = 0.0;
int rem = 0;
int t = 0;
int v = 0;
double K = 0.0;
for (int y = 0; y < h; y++)
{
for (int x = 0; x < w; x++)
{
tempB = tempG = tempR = 0.0;
for (int k = -radius; k <= radius; k++)
{
rem = (Math.Abs(x + k) % w);
t = rem * 4 + y * w * 4;
K=kernel[k+radius];
tempB += srcValue[t] * K;
tempG += srcValue[t + 1] * K;
tempR += srcValue[t + 2] * K;
}
v = x * 4 + y * w * 4;
tempValue[v] = (byte)tempB;
tempValue[v + 1] = (byte)tempG;
tempValue[v + 2] = (byte)tempR;
}
}
for (int x = 0; x < w; x++)
{
for (int y = 0; y < h; y++)
{
tempB = tempG = tempR = 0.0;
for (int k = -radius; k <= radius; k++)
{
rem = (Math.Abs(y + k) % h);
t = rem * w * 4 + x * 4;
K = kernel[k + radius];
tempB += tempValue[t] * K;
tempG += tempValue[t + 1] * K;
tempR += tempValue[t + 2] * K;
}
v = x * 4 + y * w * 4;
srcValue[v] = (byte)tempB;
srcValue[v + 1] = (byte)tempG;
srcValue[v + 2] = (byte)tempR;
}
}
Stream sTemp = srcImage.PixelBuffer.AsStream();
sTemp.Seek(0, SeekOrigin.Begin);
sTemp.Write(srcValue, 0, w * 4 * h);
return srcImage;
}
else
{
return null;
}
}

Win8 Metro(C#)数字图像处理--2.64图像高斯滤波算法-LMLPHP

最后,分享一个专业的图像处理网站(微像素),里面有很多源代码下载:

05-12 03:44