Low rank approximation of grayscale image based on SVD 2

1. a matrix A: Original Grayscale Data
A is a 448 by 794 regular matrix with rank(A)=448, i.e., A has 448 singular values.
2. A20: the best approximation by matirices of rank 20
10 singular values from the largest
3. A50: the best approximation by matirices of rank 50
20 singular values from the largest
4. A100: the best approximation by matirices of rank 100
50 singular values from the largest