Implementation of VQ Algorithms on a Reconfigurable Array Processor

摘要

Vector quantization is being widely used in image data compression applicationsdue to the facts that it is capable of achieving fractional bit rates with reasonable complexity and that the decoding is a very simple table look-up scheme. In image encoding, a vector quantizer accepts a block of pixels and outputs an address of the best matching tile stored in a codebook. The matching algorithm requires a large number of basic arithmetic operations in typical applications. Since real time coding is required in many video applications, the need for dedicated processing architectures arises naturally. This paper investigates the mapping of VQ algorithms onto an array processor to achieve near real-time compression of video images.