An improvement in DCT coding based on SVD
Zhu Ren-yi · Tv Engineering · 2004
This article utilizes mathematical principles and physical characteristics possessed in SVD(singular value decomposition) image to improve the DCT coding. Its fundamental process is that the source image is divided into n×n blocks at the beginning, the coder then computes the differences of the blocks in the way of raster scanning sequence. When the standard deviation of the block is larger than the standard one, the SVD coding is adopted, and only larger eigenvalues and its eigenvectors are coded and transported at the same time,in order to improve the compression ratio and reduce the computational complexity. Alternatively, the DCT coding is employed when the standard deviation is smaller. Experimental results obtained through computer simulation show that the proposed method has a better performance over the conventional DCT scheme in terms of the compression ratio and the PSNR and so on.