A still image encoder based on adaptive resolution vector quantization employing needless calculation elimination architecture

Abstract

-We have developed an advanced vector quantization (VQ) encoding hardware for still image encoding systems. By utilizing needless calculation elimination method, computational cost of VQ encoding is reduced to 40% or less, while maintaining the accuracy of full-search VQ. We have successfully implemented the advanced encoding method and Adaptive resolution VQ (AR-VQ), which realizes compression ratio over 1/200 while maintaining image quality, into a still image encoding processor. The processor can compress still image of 1600 x 2400 pixels within one second, which is 60 times faster than software implementation on current PCs.