Super Resolution Based on Inter-frame Translation Model

Inter-frame motion estimation is a crucial step in the super resolution reconstruction. However, the requirement of accuracy and complex computation for motion estimation is an important problem. In order to achieve accurate estimation as well as low complexity, we present a weighted inter-frame translation model. This paper mainly discusses a video sequences super resolution algorithm based on inter-frame translation model without motion-estimation. At first, the theory of Kalman filtering is introduced. Secondly, a weighted translational motion model instead of the explicit motion estimation. In order to reduce the complexity of computation, the weighted translational observation model is used to simplify the state transition equations. Thirdly, the quality of recovery image is improved due to the employment of Kalman filter framework for the super resolution algorithm. The advantages of the proposed algorithm are demonstrated by the simulation result, more specifically, the PSNR of the proposed algorithm is improved about 1 dB.