Image processing: Interative optimization problem by a gradient descent approach

I have to find the image X that minimizes the following cost function:

f=||A-(abs(X).^2-conj(X).*B)||^2

where:

- A, B and X are MxN matrices

- A and B are known images

- || is the usual norm operator

The problem has to be solved iteratively using a gradient (respect to conj(X)) descent approach. The gradient respect to conj(X) is:

g=-2*(A-(abs(X).^2-conj(X).*B))(X+B)

I'm absolutely blocked, I will be very grateful for any help provided.

Answer 1

[asim asrar]

Dear sir,

Did you get any clue to solve this problem, as i am also struck with similar type of problem.