Region-activity-based pyramidal image coder using generalizedrank-ordered prediction filter

摘要

This paper presents a lossy image compression scheme which employsa pyramidal coding technique with segmentation based on the localactivity levels. A novel prediction method is introduced, renderingsignificantly reduced variance of the quantizer input. Consequently, thecoding efficiency is much enhanced due to the greatly reducedquantization distortion. Both theoretical analysis and simulationresults show that the proposed scheme yields high quality performance.In particular, a reconstruction PSNR of 30.5 dB is achieved at a bitrate of 0.181 bpp for a 512×512 image “Lenna” beforeentropy coding