Research On The Application Of Compressed Sensing In Super-srsolution Image Reconstruction

With the continuous development of multimedia technology,the requirement of image quality is getting higher and higher.Super resolution image reconstruction(SR)produces high quality and high resolution image without improving the imaging precision of data acquire equipment,has become a focus recently.In recent years,compressed sensing(CS)theory,which is based on sparse signal representation,has made noticeable achevements in image de-noising,radar imaging and so on.It can break through the inherent limitations of traditional method to apply compressed sensing to super-resolution image reconstruction which has great research value.Compressed sensing and its application in super-resolution image reconstruction are studied in this paper,key issues in single-frame image super-resolution reconstruction and multi-frame image super-resolution reconstruction are also been discussed.This paper proposed a single-frame image super-resolution reconstruction framework based on compressed sensing theory which solves the problem that edge fuzzy phenomenon exists in the compressed image.In this framework,over-complete dictionary is deployed to replace wavelet basis which results in an improvement of image sparse representation.What's more,the joint training method between observation matrix and sparse matrix is introduced to reduce the correlation between them which achieves great reconstruction results.When it comes to multi-frame image super-resolution reconstruction,image degradation model is briefly reviewed.In order to reduce data volume while ensuring the quality of reconstruction,the joint sparse model in distributed compressed sensing is applied to multi-frame image super resolution reconstruction based on the analysis that information redundancy exists among low-resolution image sequences.