A low cost design of rate controlled JPEG-LS near lossless image compression

摘要

JPEG-LS is the latest JPEG international standard for lossless and near-lossless image compression, featuring in low cost, low complexity with competitive compression performance, which can be expected to have significant impact upon the imaging industry, especially in relation to image compression technology. To this end, we propose a low cost and low complexity rate control algorithm to provide a useful potential for further extended application of this new standard. Specifically, the non-linear prediction adopted by JPEG-LS is further exploited to indicate: (i) the compressibility of each local region inside the input image; (ii) the distortion tolerance according to human visual perception. The objective of our algorithm is to maintain a constant or similar compression ratio (bit-rate) by introducing more information loss for those regions difficult to compress and less information loss for those regions easy to compress. In addition, the information loss is distributed according to the distortion tolerance and thus the highest possible quality can be achieved for the reconstructed images. Extensive experiments reveal that the proposed rate control algorithm is effective, efficient and competitive in terms of quality measurements, when the non-rate controlled JPEG-LS is used as our benchmark.