摘要
算术编码是基于统计的、无损数据压缩效率最高的方法．对于算术编码的进位问题，目前广泛使用的是Ｒｉｓｓａｎｅｎ和Ｌａｎｇｄｏｎ提出的比特填充技术．本文提出进位陷阶技术，不必人为插入填充比特就可以解决进位问题，因而能够得到一个确切的数，并使解码端得到很好的简化．以进位陷阱的思想为基础，本文提出算术编码的一种简捷的终止技术，称为中值终止技术，并重新构造了算术编码和解码算法．本文讨论了算术编码的分析性质，得到一系列有趣的结果，包括算术编码的区间套性质、算术编码的收敛性以及串的算术编码数和算术编码映射的概念，这些分析性质在本文一些重要结论的证明中得到了应用．
Arithmetic coding is the most powerful technique for lossless data compression. The carry-over problem is inherent in arithmetic coding. Since Rissanen and Langdon proposed their classical bit-stuffing technique, no further improvement was made on this problem. This paper presents a novel solution, named the carry-trap technique, which works without a deliberately inserted stuffed-bit.Based on the concept of the trap-bit, a concise termination method, named medium termination technique, is proposed, and arithmetic encoding and decoding algorithms are reconstructed. For the first time, the paper discusses the analytic property of arithmetic coding, and gets a series of interesting results, which play important roles in the proofs of some of the theorems.
出处
《计算机学报》
EI
CSCD
北大核心
1997年第11期966-973,共8页
Chinese Journal of Computers
