A video super-resolution reconstruction method based on sub-pixel registration
Yang Xiao-min · Journal of Optoelectronics·laser · 2009
To improve the spatial resolution of a video,a super-resolution reconstruction method is proposed utilizing the movement information between frames.For the reconstruction of the whole video,a segmentation reconstruction model based on sliding window is proposed.In each sliding window,firstly,a movement registration with sub-pixel precision is performed on the adjacent frames;Secondly,the super-resolution reconstruction is implemented using the iterative back projection algorithm.In the registration algorithm,a method based on four parameter rigid transformation model is proposed.The iterative solving method and the Gauss pyramid image model are used to estimate the movement parameters from coarseness to fine.The reconstruction is performed on simulated image and real-taken color video respectively.The experimental results show that this registration algoritm gets better precision,and the reconstruction algorithm achieves higher PSNR,and the reconstructed video has better visual effect and higher resolution ability.It can be extensively applied to the super-resolution reconstruction of video sequences in which the frames are different from each other mainly by translation and rotation.