A Systolic Architecture for Real-Time Composite Video Image Coding
- 1 October 1986
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
Abstract
Secure and reliable image transmission is achieved using a representation of video images in digital form and digital communication techniques. However, the digital representation results in a large increase in the transmission bandwidth required. Hence the need for coding or compression techniques that will reduce the bandwidth required or equivalently the bit-rate involved as much as possible for a given level of reconstructed image quality. From Information Theory, we know that it is possible to obtain better performance by coding vectors of picture elements instead of coding the individual pixels. Nevertheless, application of Vector Quantizers is limited due to the high computational complexity. With curent VLSI technology this can be changed. The repetitive nature of VQ error computation and codebook vectors comparison points to us that using special structure such as systolic type architecture, VQ can be VLSI implemented to make real-time applications possible. Along with full search VQ, we emphasize on the implementation of multistage VQ which can significantly reduce the computation load due to much smaller codebooks required. Examples of processed images are presented to show the validity of such an approach.Keywords
This publication has 6 references indexed in Scilit:
- Image coding using vector quantizationPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2005
- Multiple stage vector quantization for speech codingPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2005
- Vector quantizationIEEE ASSP Magazine, 1984
- Vector quantization: A pattern-matching technique for speech codingIEEE Communications Magazine, 1983
- Why systolic architectures?Computer, 1982
- An Algorithm for Vector Quantizer DesignIEEE Transactions on Communications, 1980