Image Super-resolution Reconstruction Algorithm Based On Adaptive Regularization

As a means to store and exchange visual information,images are the most important information carrier for human beings to know the world.With the rapid development of the Internet industry and the upgrading of emerging electronic products,people have higher and higher definition requirements for images.However,in the daily life,low-resolution images are often produced due to the influence of hardware equipment,shooting environment and unpredictable factors.In order to solve this problem,Super Resolution reconstruction technology(SR)was born.This technique utilizes the redundant complementary information between sequential Low Resolution(LR)images in the same motion scene and reconstructs a High Resolution(HR)image based on the software method to improve low resolution image quality.The main research contents of this paper are as follows:(1)An adaptive bilateral total variation regularization,namely ABTV regularization algorithm,is proposed.Based on bilateral total variation(BTV)regularization,the algorithm introduces an adaptive weight matrix of the same size as the high resolution image,which can assign an adaptive weight for each pixel.It adjusts the weight according to the position of the pixel in the images to achieve image edge information while suppressing image noise.(2)This paper is to verify the validity of the L1-L2 mixed norm model for filtering out Gaussian and Laplace mixed noise in real low-resolution sequence images.Then,ABTV regularization algorithm was combined with L1-L2 mixed norm model to conduct experiments on simulated low-resolution images sequences and actual image sequences.(3)An adaptive regularization parameter selection method is proposed.The method makes full use of the image information solved by the previous iteration in the super-resolution reconstruction process to adaptive select the regularization parameters in the image reconstruction process.The method is combined with the ABTV regularization algorithm to prove the effective of the method.