Lossy image compression using SVD coding, compressive autoencoders, and prediction error-vector quantization

Abstract

Now-a-days due to the huge increase in the size of image data Lossy Image Compression is highly used to reduce the image size but without having huge data loss. Image compression using SVD coding algorithm, Compressive Encoders and using prediction Error and Vectorization ratio are proved to have numerous application in image compression. Image compression using SVD coding algorithm involves refactoring of a digital image into three matrixes. Refactoring is achieved by using singular values, and the image is represented with a smaller set of values. Though, encoders cannot directly optimize due to the inherentNon-differentiability of the compression loss but it is out performing recently proposed approaches based on RNNs. The PE-VQ method is based on Prediction Error and Vector Quantisation techniques where image performance is determined using compression ratio and PSNR values using databases namely CLEF med 2009, Corel 1k and standard images like Lena, Barbara etc. Thus, in this research article a comparative study of these three techniquesis carried out where their image quality and compression ratio is examined by using the PSNR values and compression ratios.