An efficient parallel algorithm for high resolution color image reconstruction

This paper studies the application of preconditioned conjugate gradient methods in high resolution color image reconstruction problems. The high resolution color images are reconstructed from multiple undersampled, shifted, degraded color frames with subpixel displacements. The resulting degradation matrices are spatially variant. The preconditioners are derived by taking the cosine transform approximation of the degradation matrices. The resulting preconditioning matrices allow the use of fast transform methods. We show how the methods can be implemented on parallel computers, and we demonstrate their parallel efficiency using experiments on a sixteen processor IBM SP-2.