The Performance Analysis of Universal Inspection Registration for Video MSRR Scheme Established on Stochastic Maximum A Posteriori Framework

Traditionally, the general MSRR (Multi-frame Super Resolution Reconstruction) schemes can be effectively implemented on the video with a simple shifting motion pattern because the conventional inspected model in video MSRR scheme is established on an ordinary shifting pattern. For handling with any real and complex inter-frame motion patterns, the universal inspection registration, established on a fast affine block-based transform, has been proposed. Cooperated with the universal inspection registration, this paper proposes the video MSRR scheme established on stochastic maximum a posteriori (MAP) framework. By mathematically cooperating with the ordinary Tikhonov prior function, the refined SR image can be enumerated by optimizing the MAP error function with an optimized nonlinear programing technique. Using two standard video sequences such as Susie and Foreman with four noise models at several noise energy, the refined SR image products from computer testing expose that the MSRR schemes with the proposed universal inspection registration outperforms than the antecedent MSRR schemes with the an ordinary shifting inspection registration.