The Bayer Raw Data Compression Algorithm Based on Vector Quantization

摘要：

-This paper proposed a Bayer CFA raw data direct compression algorithm based on vector quantization. This algorithm firstly divided image data as vector using division block method for CFA raw data according to the character of Bayer pattern, then encoded each vector block using the trained codebooks and got compressed data. In the process of training codebooks, this paper adopted a large number of typical Bayer CFA images forming training space, and ensured the best proper distortion measure function for Bayer raw data through experiments. It also improved the traditional LBG algorithm, optimized the obtained codebooks by the incorporate codeword method, and finally got good performance codebooks. Through many experiments, this direct compression method acquired the lower distortion and bit rate, and interpolated color images could produce excellent results in terms of both subjective and objective image quality measures.

关键词： color filter array Bayer pattern raw data vector quantization LBG algorithm

作者: ZHOU Yingfeng LI Junfeng

作者单位: Engineering College, Jinggangshan University, Jian, Jiangxi, China, 343009

会议类型: 国际会议

会议名称: 第七届国际测试技术研讨会

会议地点: 北京

会议语种:英文

在线出版日期: 2007-08-05（万方平台首次上网日期，不代表论文的发表时间）