EmguCV：如何从图像创建协方差矩阵？

Question

I have an Image<Gray, Byte> and i want to calculate the covariance matrix for the image.

Therefore i use the function CvInvoke.cvCalcCovarMatrix(imageptr, 2, cov, avg, COVAR_METHOD.CV_COVAR_NORMAL);

Here my code:

       Image<Gray, Byte> image_gray = _image.Convert<Gray, Byte>();
       Matrix<float> cov = new Matrix<float>(image_gray.Rows, image_gray.Cols);
       Matrix<float> avg = new Matrix<float>(image_gray.Cols, 1);
       Matrix<float>[] input = new Matrix<float>[image_gray.Rows];
       float[] temp = new float[image_gray.Rows];


       for (int j = 0; j <image_gray.Rows; j++)
       {
           for (int i = 0; i < image_gray.Cols; i++)
           {
                temp[i] = (float)image_gray[j, i].Intensity;

            }

            input[j] =new Matrix<float>(temp);
       }

        IntPtr[] imageptr = Array.ConvertAll<Matrix<Single>, IntPtr>(input, delegate(Matrix<Single> mat) { return mat.Ptr; });
        CvInvoke.cvCalcCovarMatrix(imageptr, 2, cov, avg, COVAR_METHOD.CV_COVAR_NORMAL);

The problem is, that the result of the covariance matrix elements are always null.

Answer 1

The mistake was the 2nd parameter of cvCalcCovarMatrix(imageptr,count, cov, avg, COVAR_METHOD.CV_COVAR_NORMAL);

Correct is:

cvCalcCovarMatrix(imageptr, image_gray.Cols, cov, avg, COVAR_METHOD.CV_COVAR_NORMAL);