Color Image Coding Using Differential Trellis-Coded Vector Quantization

• Main Authors: Shao-Wei Hsu, 許紹偉
• Other Authors: Kuen-Tsair Lay
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/43328277022883732301

碩士 === 國立臺灣科技大學 === 電子工程系 === 92 === ABSTRACT Image compression is the process of reducing the number of bits required to represent an image. Vector Quantization (VQ) is one method to perform this operation. With this method a set of data points is encoded by a reduced set of reference vectors (the codebook). Vector quantization is useful in compressing data that arises in a wide range of applications and it can achieve better compression performance than conventional coding techniques, which are based on the encoding of scalar quantization. Trellis coded quantization (TCQ) is a relatively new scheme which was proposed to cope with memoryless and Gaussian-Markov source in the early 1990’s. It is derived by modification to trellis coded modulation (TCM), which is a joint coding and modulation scheme proposed for reducing bit error rate (BER) with no need of bandwidth expansion in digital communication. Basically, the TCQ encoder is the TCM demodulator, and the TCQ decoder id the TCM modulator. In this thesis, we focus on trellis coded vector quantization (TCVQ). The novel feature of TCVQ is the partition of an expanded set of vector quantization symbols into subsets and the labeling of the trellis branches with elements in these subsets. This is done in such a way that the minimum Euclidean distance among the output levels assigned to the same branch is maximized. However, set partition is not a simple task in the vector case and multidimensional spaces. So we need to find an efficient algorithm to solve this problem and improve the performance of TCVQ. Then we propose to add new features to conventional TCVQ：namely, differential trellis coded vector quantization(DTCVQ). For comparison purpose, we consider, TCQ, VQ, DVQ, TCVQ, DTCVQ, respectively. Performance comparisons among the aforementioned schemes are made by simulation. In experiments, the following phenomena is observed: first, TCVQ scheme is slightly better than VQ scheme. Second, DTCVQ scheme is much better than TCVQ scheme, at the cost of more computation in the coding process. Keyword：vector quantization, trellis coded quantization, Trellis coded vector quantization, differential trellis coded vector quantization