Hardware-Constrained Hybrid Coding of Video Imagery

Abstract

The design, implementation, and performance of a video bandwidth compression system is described. In this system, compression is obtained by several methods including the use of DCT/DCPM hybrid coding, frame rate reduction, and resolution reduction. The overall compression ratio is up to 1000:1. The hardware-constrained design of the DCT and the DPCM is described and a new method is derived to solve the optimum integer bit-assignment problem associated with the block quantization process in the DPCM. Computer simulation results are presented which predict that the performance of the system using the derived optimal bit assignment method is superior to those obtained by other bit assignment methods. The real-time hybrid coding system design is optimized for a set of ?modified? average statistics to compress a wide variety of input video images. This approach eliminates the problem of nonzero dc mean value which could otherwise cause serious degradations in the system performance. The compression system is fully implemented and the quality of the reconstructed video as predicted by computer simulation has been demonstrated by the actual hardware performance. The PSNR of the reconstructed imagery is in excess of 36 dB at 2 bits per pixel.