Utilização de algorítmos genéticos na geração de tabelas de quantização Y Cb Cr para o padrão JPEG e avalização de artefatos de blocos

Abstract

The JPEG standard has been used for a long time as the algorithm of reference for the compression of monochrome images, and that compression would be applied separately to components of color images. In the process of image compression JPEG utilizes standard quantization table in an independent way to the application. Using JPEG, the quality loss of the reconstructed image is a direct function of the quantization process and this can be seen in the peak signal to noise ratio (PSNR) as in terms of subjective quality. A method for the generation of quantization tables for the JPEG algorithm applicable to monochrome and color images is showed and it is addressed to specific applications, that is, compression of natural images, etc. Another innovation is the use of Genetic Algorithms in the search for a quantization table linked to a class of images and this provides a superior PSNR, if compared with that obtained by the JPEG algorithm with its standard quantization table. The comparisson between the algorithm proposed in this Project and the JPEG standart is also done through the application of a blockiness metric for compressed images, highlighting which algorithm will provide the lowest blocking artifact.