Noise Removal Algorithm Based on Code Shrinkage Techniques

摘要

The research reported in the paper aims the development of a methodology for noise removal in image restoration. The noise removal technique is based on a transform that decorrelates the components of the noise followed by the use of a shrinkage process in order to remove the noise. The second part of the paper presents a LMS optimal PCA compression/decompression scheme, where the noise is annihilated in the feature space. The experimental results are performed on data represented by monochrome images decomposed in blocks of 8 × 8 size and liniarization.