Ricean code based compression method for Bayer CFA images

Abstract

Generally on CCD Bayer CFA images, compression is performed after demosaicing. Nowadays, for better image quality compression-first schemes are preferred over the conventional demosaicing-first schemes. In some high-end photography applications, original CFA images are required; in such cases lossless compression of CFA images is necessary. A fair performance is obtained for CFA images by lossless image compression methods like JPEG-LS, JPEG-2000, etc. The proposed method mainly aims at exploiting a context matching technique to rank the neighboring pixels when predicting a pixel in a CFA image. It reorders the neighboring samples such that closest neighboring samples of the same color are predicted on higher context similarity. Adaptive color difference estimation follows the adaptive codeword generation technique to adjust the divisor of rice code for encoding the prediction residues. From Simulation results, the proposed algorithm achieved a better compression performance as compared with conventional lossless CFA image coding methods. The experimental results are obtained to prove the proposed method is having best average compression ratio as compared with the latest lossless Bayer image compression algorithms using MATLAB, a technical computing language.