Rank Minimization

Summary

The key to construct effective denoising method is on how to make full use of the prior information in the image. This chapter discusses the development of the matrix rank minimization‐based image denoising methods. Singular value decomposition (SVD) has long been a basic tool for signal processing and analysis. The chapter shows how to infer the noise level according to image singular values out of SVD. It discusses how to denoise image by hard thresholding the singular values of the noisy image and presents a signal localization method to improve the SVD hard threshold denoising performance. The chapter explores the application of random SVD low‐rank image matrix approximation to achieve SAP corrupted image recovery. It also discusses the application of iterative adjustment for performance improvement, which can apply to any of the SVD hard thresholding methods.