图像处理之：最大类间方差法法进行图像二值化处理

 方法简介：

g=ω0ω1(μ0-μ1)^2
对于图像I(x,y),前景(即目标)和背景的分割阈值记作T,属于前景的像素点数占整幅图像的比例记为ω0,其平均灰度μ0;
背景像素点数占整幅图像的比例为ω1,其平均灰度为μ1,类间方差记为g。采用遍历的方法得到使类间方差最大的阈值T

int Process::image_binary(unsigned char* buffer, int w, int h)
{
	int height = h;
	int width = w;
	int size = h*w;
	int histogram[256] = { 0 };
	int tmp_val = 0;
	for (int i = 0; i<height; i++)  //计算直方图
	{
		for (int j = 0; j<width; j++)
		{
			histogram[(int)buffer[i*width + j]]++;
		}
	}
	//计算阈值
	int gSum0 = 0, gSum1 = 0, N0 = 0, N1 = 0;
	float w0 = 0.0, w1 = 0.0, u0 = 0.0, u1 = 0.0, g = 0.0, tempg = 0.0;
	int thr = 0;
	for (int i = 0; i<256; i++)
	{
		gSum0 = 0;
		gSum1 = 0;
		N0 += histogram[i];
		w0 = (float)N0 / size;
		N1 = size - N0;
		w1 = 1 - w0;
		for (int j = 0; j <= i; j++)
		{
			gSum0 += j*histogram[j];
		}
		u0 = (float)gSum0 / N0;
		for (int k = i + 1; k<256; k++)
		{
			gSum1 += k*histogram[k];
		}
		u1 = (float)gSum1 / N1;
		g = w0*w1*(u0 - u1)*(u0 - u1);
		if (tempg<g)
		{
			tempg = g;
			thr = i;
		}
	}
	for (int i = 0; i<height; i++)  //二值化处理
	{
		for (int j = 0; j<width; j++)
		{
			tmp_val = buffer[i*width + j];
			if (tmp_val <= thr)
				buffer[i*width + j] = 0;
			else
				buffer[i*width + j] = 255;
		}
	}
	return 0;
}

处理效果对比：