Study On Restoration Method Of Compressed Image Based On Sparse Regularization

Image compression technology can effectively reduce the amount of data,thus,it can save the communication channel and improve the transfer rate of information.It has been widely used in signal analysis and processing field.Image compression and decompression process,however,will cause the image noise,fuzzy image quality degradation,which will directly affect the application performance of the image.Therefore,it is necessary to carry out study of compressed image restoration processing.This thesis is based on the principle of image compression.It makes the thorough analysis caused by compression of image quality degradation characteristics,then,it builds compressed image restoration processing model and method.Concrete study contents are as follows:Firstly,starting from the image degradation model,it combines with the principle of image compression coding.Then it analyzes the effect of image compression algorithm based on discrete cosine transform on image quality.Secondly,the compressed image is sparse representation.For the sparse representation,a key part is the choice of a dictionary.For image and dictionary adaptive problem,it presents an adaptive dictionary learning method,and then it combines with the adaptive dictionary to construct the sparse adaptive presentation.This method has solved the problem in compressed sensing system how to use the observed value image block in constructing adaptive dictionary to construct adaptive dictionary.Thirdly,considering the image quality degradation characteristics caused by compression,this paper builds sparse regularization prior based on sparse representation theory and compressed sensing,and brings it into the regularized image sparse restoration model,at the same time,it researches the solving the model of regularization method.So it gives the compressed image restoration method based on sparse regularization.Finally,analyze the effect of regular parameter on the compressed image restoration and choose the appropriate regular parameter.Then,we make comparison with several classical restoration algorithms to our new method in this paper.And based on image quality objective evaluation and subjective evaluation as well as the time performance,we illustrate the processing performance of the new compressed image restoration algorithm.