The study of Color Quantization

• Main Authors: Ming-Gong Lee, 李銘恭
• Other Authors: Yu-Chin Hu
• Format: Others
• Language: zh-TW
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/40658810522182428767

碩士 === 靜宜大學 === 資訊管理學系研究所 === 93 === Thanks to the rapid progress of WWW, and the technology of computer software and hardware, multimedia information has involved in our daily life. Considering the limitations of the bandwidth, monitor display or storage requirement, it is essential for multimedia information to be compressed. Since the color quantization for compression is simple and fast, it is applied extensively. In the encoding process of color quantization for compression, it consists of two steps: palette design and pixel mapping, the color on the palette can be concerned to the reconstructed image quality. Besides the pixel mapping in color quantization will affect performance of the image compression. In this paper, we proposed two palette design methods, dynamic difference of cell splitting and stable color palette, the first, it combine the splitting and clustering technique, so the executing time is quite fast and the quantized error is small. In the later, it is an improving method for LBG algorithm, from the experimental results, it show that the executing time is only 1.28 % of LBG algorithm when the palette size is 256, and the image quality is still very well. In pixel mapping procedure, we proposed an algorithm that is equivalent to the full search algorithm. In our method, we using the sum of R, G and B spectrum for filleting other colors, from the experimental results, it show the execution time is only 6.19% when palette size is 256.