a) Let h, and h be to orthogonal convolution operators: -2 -1 0] 6:1 -1 0 1 0 12 h 0-1 10 -1 21 0 m What is the difference between the estimates of gradient magnitude and gradient direction that we find based on the results from the operators h, and h, as compared to the ordinary Sobel operators h, and hy: [10-17 h, 20-2 = [10 -1 -2 -17 0 0 0 2 1 b) Are the filters h, and h separable into ID-filters? What about h, and hy? Give reasons for your answers.