Application of Genetic Algorithm to Vector Quantization

• Main Authors: Ying-Tzu Hung, 洪櫻慈
• Other Authors: Shing-Tai Pan
• Format: Others
• Language: zh-TW
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/63708062339502629860

碩士 === 樹德科技大學 === 資訊工程學系 === 93 === Vector quantization (VQ) is an efficient image-coding technique because of its high compression ratio for distorted compression technique. However, the method of VQ always costs much encoding time and the large amount of MSE computations. Consequently, some optimization methods are used to speedup the training procedure. It is well known that Genetic Algorithm (GA) is an efficient global search method. Hence, this thesis will improve GVQ algorithm and propose a new gene classification vector quantification (GCVQ) algorithm. In the past the codebook in VQ is mostly produced from training set. Consequently, the codevectors of codebook posess only the characteristic of the image which the training set is generated. The codebook could not be suitable for all images. So in this thesis, CVQ and MRVQ are used to implement the GCVQ algorithm, and it will produces the global codebook.