In this paper, a novel state-based dynamic multi-alphabet arithmetic coding algorithm which adapts efficiently to the locally occurring symbol statistics is presented. The proposed coding algorithm is applicable to sources such as raw images or transformed images that locally produce a smaller set of symbols from a large alphabet, or any other source characterized by very large alphabet size and highly skewed distributions. The performance of the proposed algorithm is compared with two standard entropy coding schemes, CA-2D-VLC and CABAC, that have recently appeared in the literature, in terms of compression ratio, peak signal-to-noise ratio and subjective quality. The simulation results obtained are encouraging, paving the way for further research and hardware implementation of this algorithm. It is found that the proposed algorithm achieves an increase in the compression ratio of about 45% over the compared standards for the similar peak signal-to-noise ratio and subjective quality.
Block truncation coding (BTC) technique is a simple and fast image compression algorithm since complicated transforms are not used. The principle used in BTC algorithm is to use two-level quantiser that adapts to local properties of the image while preserving the first-or first-and second-order statistical moments. The parameters transmitter or stored in the BTC algorithm are statistical moments and bitplane yielding good quality images at a bitrate of 2 bits per pixel (bpp). In this paper, two algorithms for modified BTC (MBTC) are proposed for reducing the bitrate below 2 bpp. The principal used in the proposed algorithms is to use the ratio of moments which is a smaller value when compared to absolute moments. The ratio values are then entropy coded. The bitplane is also coded to remove the correlation among the bits. The proposed algorithms are compared with MBTC and the algorithms obtained by combining JPEG standard with MBTC in terms of bitrate, peak signal-to-noise ratio (PSNR) and subjective quality. It is found that the reconstructed images obtained using the proposed algorithms yield better results.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.