Singular value decomposition based image compression technique

Abstract

The use of matrix in image processing is progressively increasing in many years. Singular Value Decomposition (SVD) is one such image compression technique that performs on matrices. In this thesis, we investigated an algorithm which is used to compress the colorful images for estimating the effect of information content of the compressed images. In order to evaluate the effect, we calculated the CR, PSNR, and MSE values of the compressed images with respect to the reference images using MATLAB. We calculated MSE values for the compressed images using JPEG2000 and JPEG and compared them with SVD. Finally, MOS are calculated for different compressed images using SVD and compared them with the existing MOS in the database of JPEG2000 and JPEG.