LZCE算法:竞争进化的Lempel-Ziv算法

LZCE:Lempel Ziv Algorithm with Competitive Evolution

给出一组在ＬＺ算法基础上改进而来的具有竞争进化功能的字典算法：ＬＺＣＥ（Ｌｅｍ ｐｅｌ－ＺｉｖＡｌｇｏｒｉｔｈｍ ｗ ｉｔｈ Ｃｏｍ ｐｅｔｉｔｉｖｅ Ｅｖｏｌｕｔｉｏｎ）．在算法构成上，根据图像压缩应用的特点，将ＬＺ算法（Ｖ－Ｆ编码）和Ｈｕｆｆｍ ａｎ 编码（Ｆ－Ｖ 编码）相结合，实现了基于先验知识的统计编码与基于学习的ＬＺ算法的相互竞争，构造出可综合利用局域相关性、广域相关性和先验统计知识的竞争进化的字典算法；在二维相关性利用上，经ＤＣＴ域引入ＬＺＣＥ算法，较好地解决了一维ＬＺ算法难以利用二维相关性这一问题．实验结果表明，对于相同的ＰＳＮＲ，该方法可将ＪＰＥＧ 码率降低２０％

This paper proposed a group of algorithms:LZCE(Lempel Ziv algorithms with competitive evolution).By combining the fixed to variable length code based on a priori with the variable to fixed length code based on learning,a competitive evolution was introduced to the LZ algorithm ,which makes the LZCE efficient for compressing the images within a wide range of distributions.By utilizing the LZCE algorithm in DCT domain,the following defects of the existing LZ algorithms can be remedied:1 D LZ algorithms have a poor compression efficiency;2 D LZ algorithms have a poor processing speed.The preliminary experiment shows that by adopting LZCE algorithm into JPEG entropy coding unit and keeping the same PSNR,about 20% of the bit rate can be reduced.