A Hybrid Coding Scheme for Color Image Compression

• Main Authors: Ta-Jung Kuo, 郭大榮
• Other Authors: Ching-Huang Wei
• Format: Others
• Language: en_US
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/18416180146372995425

碩士 === 國立高雄第一科技大學 === 電腦與通訊工程所 === 93 === Two hybrid image-coding schemes are proposed by the combination of the advantages of four coding schemes: BTC, VQ, DCT coding, and predictive coding. The experiments show that the hybrid coding schemes can obtain high compression ratio, saved VQ codebook searching time, and the competitive image quality. BTC has the property of low computational complexity and the edge preservation. DCT has the property of high image quality and high compression ratio. VQ has the property of high compression ratio and the moderate fidelity. The input image is first coded through BTC to generate a bit-map and both high-mean and low-mean sub-images. The high-mean and low-mean sub-images are encoded through the DCT coding scheme. The predictive scheme here is used to reduce the required bit rate because the property of similarity exists between two neighboring blocks in a high-correlated image. The bit map generated by BTC is encoded by VQ and block predictive coding scheme. Especially the usage of block predictive coding scheme here in bit map is not only to reduce the bit rate but also to save the codebook searching time about 25% blocks than using VQ alone in bit map. Because it just need to simply assign two-bit indicator to stand for the block. For coding the color image, a common codebook is used to encoding the three color-component planes in order to reduce the storage space for codebook without any explicit degradation in image quality.