A high throughput context-based adaptive arithmetic codec for JPEG2000

Abstract

In JPEG2000 image compression standard, context-based arithmetic coding is adopted for improving compression rate. Efficient and high throughput codec: design is strongly required for handling high-resolution images. We propose an efficient architecture for context-based adaptive arithmetic coding, which can achieve high throughput performance. Based on parallel leading zeros detection and bit-stuffing handling, symbol can be encoded and decoded within one cycle. Therefore, throughput rate can be increased as high as codec operating frequency. The modified encoding and decoding flows are compatible with JPEG2000 standard. For 0.35u 1P4M CMOS technology, encoding and decoding rate can run up to 200Msymbol/sec and 175Msymbol/sec respectively. They are only cost 6.8K and 9.8K gate count. These high throughput performances can meet high-resolution real time application requirement.