Image Denoising And Ireconstruction Based On Compressed Sensing

With the rapid development of information technology,the traditional signal processing mode is unable to meet the amounts of processing data's quickly increasing.It's a critical time to develop an efficient sampling theory,and compressed sensing(CS)theory is proposed.The people's requirements of image products are increasing with the development of living standards.However,it's inevitable that the image will be contaminated by the image noise,and the image reconstruction is unavoidable during the image transmission.Image denoising and image reconstruction are very important steps in improving the image quality.A research on compressed sensing theory in image denoising and image reconstruction was accomplished in this paper.In the first place,some basic concepts and the framework of CS theory were introduced.The principle and method of the signal sparse representation and the Fourier transform were presented especially.The process of measurement matrix designigng was proposed.Then three kinds of minimum lp norm signal reconstruction algorithms were analysed,the residual of the l1 norm reconstructon algorithm was calculated especially.In the second place,an exhaustive study was made on the greedy algorithms and convex relaxation methods,such as Matching Pursuit,Orthogonal Matching Pursuit,Regularized Orthogonal Matching Pursuit and Basic Pursuit.The flow chats of them were researched and some experiments of the reconstruction were carried out.Besides,some other popular algorithms were introduced briefly.In the third place,six kinds of image noises with different Probability Density Function were discussed.Then the subjective evaluations and objective evaluations of image quality were elaborated.The effects of CS image denoising were compared with the traditional Wavelet image denoising.The experimental results revealed that the CS method displays more flexibility and robustness applied in different image noises.The last but not the least,three kinds of image sparse representation methods were expounded and the reconstruction results of them were compared by computing time and visual quality.Diverse area images were used to study the computing time and visual quality with various sampling rates(SR=0.2,0.4,0.8).The CS theory was introduced into many useful areas and it displayed good consequences.The CS image reconstruction can recover the image in highly visual quality with fewer data and the CS theory indicated excellent ability of image reconstruction.