Chun-Yang Ho1, Chaur-Heh Hsieh2, Chung-Woei Chao3 1Kuang-Wu Institute of Technology and Commerce Peito, 112, Taipei, Taiwan, R. O. C.
2Department of Information Engineering, I-Shou University Kaohsiung, 840, Taiwan, R.O.C.
3Ministry of National Defense Taipei, Taiwan, R. O. C.
Received:
March 1, 1999
Accepted:
June 17, 1999
Publication Date:
September 1, 1999
Download Citation:
||https://doi.org/10.6180/jase.1999.2.3.04
Recently, vector quantization (VQ) has been received great attention for image compression. In VQ, each image block is encoded into a quantization index. Search-order coding (SOC) is an efficient compression technique for the VQ indexes. This paper presents a modification of SOC that exploits the correlation of vector-quantized indexes efficiently. For an input index, a novel searching scheme is employed to find out the matched index from the previous indexes and then the corresponding search order of the matched index is sent to the decoder. The search order is encoded with a simple and efficient variable-length code. Experimental results show that the proposed scheme significantly reduces the bit rate without extra coding error. Moreover, the extra computation introduced is negligible.ABSTRACT
Keywords:
Image compression, Vector quantization, Codevector
REFERENCES