openCV c++ 中的手动腐蚀图像问题（二进制图像）

Question

I'm writing manual code for erosion or dilatation. And I don't know why the following code does not work.

As you can see, when I'm in the point of (i,j) I go around this point and look for any white pixel. If there is anywehere (in the Little 3x3 matrix (ii,jj) ) white pixel I change my (i, j) point to white.

Does anybody know why it's not working?

I made myself sure that I load the binary image. If I count white and black Pixels summary the gile me the right number of etery Pixels.

 for (int i=1; i<gray.cols-1; i++)
    {
        for(int j=1; j<gray.rows-1; j++)
            {
                ii=i;
                jj=j;

                for (ii-1; ii<i+2; ii++ )
                {
                    for (jj-1; jj<j+2; jj++ )
                    {

                        if (gray.at<uchar>(ii,jj) ==255)
                            gray.at<uchar>(i,j) = 255;
                    };
                };
};

If anybody has an idea how to do it differently?

Answer 1

You should create temporary image to write results of erosion/dilation to. Your problem is that you are writing to original image. Assume that only the first pixel is white. At first iteration you will color second pixel to white. And then in second iteration you will look at newly colored second pixel and color third pixel in white. And so on for all pixels in line.