多阶上下文自适应二进制算术编码实现

提出了一种基于多阶上下文自适应的二进制算术编码算法.考虑了编码符号序列之间的相关性,用符号序列的大概率值代替单个符号的较小概率值,由此建立了多阶上下文概率模型,降低符号间的冗余度. 将此模型应用到二进制算术编码中,形成了一种比传统二进制算术编码更有效的方法,能让平均码长变短并使得码字的信息量逼近符号的熵率,从而显著提高二进制算术编码的数据压缩比(可高达90%).对不同类型的数据文件进行实验测试,结果显示其压缩编码效果良好.

Realization of multi-order context-based adaptive binary arithmetic coding

An adaptive binary arithmetic coding(ABAC) algorithm was introduced which was based on multi-order context.The correlation between the coded symbol sequences was considered and a multi-order context probability model(MOCPM) was built to reduce redundancy between symbols by using bigger probability value of symbol sequence instead of smaller probability value of single symbol.A novel method is produced by utilizing MOCPM in ABAC,which is more effective than traditional methods.It enables to shorten average code length and makes the coded symbol almost its entropy rate,so a notable improvement is achieved in the data compression of binary arithmetic coding.The data compression ratio can be as high as 90 %.Experimental results on different data types demonstrate our algorithm is highly effective in coding.