Dynamic AIFV Coding
Hiraoka Tomotaka, H. Yamamoto · 2018
In this paper, we propose two types of dynamic AIFV (almost instantaneous fixed-to-variable length) coding schemes for stationary memoryless sources with unknown probability distribution such that the dynamic AIFV code trees are constructed from the dynamic Huffman code tree. The one is based on the AIFV code with two code trees and the other is based on a simplified AIFV-m code with m code trees. The proposed dynamic AIFV coding can be implemented with almost the same complexity as the dynamic Huffman coding, and it can attain better compression rate than the dynamic Huffman code when the probability of the most likely source symbol is larger than about 0.62.