摘要
嵌入式零树小波编码算法(EZW)在扫描过程中为了判断小波系数是零树根还是孤零,需要对系数进行重复扫描,这样不仅浪费时间,而且会产生大量冗余的符号流。针对EZW算法存在的不足,提出了一种改进的EZW算法,通过增加2个标识符,来避免连续出现零树根,对扫描产生的符号流采用哈夫曼编码。实验表明,改进型的算法,减少了扫描时间,提高了压缩比和编码效率。
In the analyzing process of EZW algorithm, it is found that the current probability of zerotree root is gradually increased with the accumulation of continuous ordered scanning for the wavelet coefficients of image. In this paper, an improved EZW algorithm for the phenomenon of successive zerotree root is proposed. The improved algorithm uses six symbols instead of four symbols in EZW, and the newly-added symbols are used to avoid the successive appearance of zerotree roots. The symbol stream, produced by scanning, is coded with Huffman code and then Run-length code is used to recode the binary symbol stream and implement the amplitude quantization. Experimental results show that the improved algorithm could greatly reduce the number of zerotree roots through re-classifying the significant symbols, save the scanning time, and promote the compression ratio and coding efficiency while having no influence on other symbols.
出处
《通信技术》
2013年第2期29-31,34,共4页
Communications Technology
关键词
小波变换
零树结构
图像压缩
wavelet transform
zerotree structure
image compression
