In this project I studied the blind deconvolution problem and presented two algorithms proposed as a solution. Iterative blind deconvolution converges to an image by alternating between the image and Fourier domains, enforcing known constraints in each. Simulated annealing approaches blind deconvolution as an optimization problem, tries to minimize an objective function which is the difference between blurred image and convolution of estimates.We have seen that using both of the algorithms, the originals can be recovered with minimal partial information.
However both of the algorithms have some drawbacks. Although Iterative blind deconvolution is fast, it has uncertain convergence and uniqueness properties. On the other hand simulated annealing method has known convergence properties but it suffers from computational complexity. For realistic size images it will take excessive amount of time to converge to a solution.