Image Compression Using Two Stage Block Coding

摘要

Two Stage Block Truncation Coding(TSBTC) exploiting two-component source model and visual masking effect of human observer is proposed in this paper. In TSBTC, an input image is first coded by Block Truncation Coding (BTC) over 8~*8 pixel blocks with a 4:1 subsampling / interpolation and then the coarsely coded image passes through a 3~*3 FIR low pass filter to have the out-of-band noise removed, by which the discontinuous component is established. The remainder component is coded by BTC over 4~*4 pixel blocks in which the two representive gray levels are nonuniform quantized. Simulation on tested images shows that image quality is improved remarkably by TSBTC compared with that by BTC at the same bit rate or less, while TSBTC still keeps all the advantages of BTC.