Research On Image Super-resolution Algorithm Based On Sparse Representation

The purpose of image super-resolution reconstruction is to transform low-resolution images into clear,high-resolution image.In recent years,with the development of sparse representation theory,the image reconstruction techniques based on sparse become the main methods.However,this method depends on the selection of image blocks,and can not make good use of the prior information of the image.In response to this phenomenon,This article based on the original image reconstruction,in-depth study of the following two methods:1.Explored and studied Image Super-resolution Reconstruction Based on Wavelet Transform and Sparse Representation;Using the VO model,decomposes the image into structural and texture parts,Image reconstruction using Haar wavelet transform for the structural part,The texture part adopts sparse representation of image super-resolution reconstruction,and gray feature matrix is used to extract rich texture features,To better recovery of image edge details.Linearly weighting the two partial reconstruction results and performing non-local optimization to obtain the reconstructed high resolution image.Simulation results show that the algorithm can effectively recover the texture and edge details of the image while satisfying the reconstruction efficiency.2.Explored and studied Remote Sensing Image Reconstruction Based on Optimal Direction Multi-Dictionary Learning;In order to improve the speed of dictionary training,dictionary training using the MOD method with faster convergence speed.Due to the complexity of remote sensing images,using the VO model decomposition are reconstructed using sparse representations,to make the structural texture features of remote sensing images more obvious.Simulation results show that the algorithm can effectively improve the reconstruction efficiency of remote sensing images,The texture information of the image structure is clearer and more in line with human visual observation.