A fast algorithm for Huffman decoding based on a recursion Huffman tree

This paper focuses on the time efficiency of Huffman decoding. In this paper, we utilize numerical interpretation to speed up the decoding process. The proposed algorithm firstly transforms the given Huffman tree into a recursion Huffman tree. Then, with the help of the recursion Huffman tree, the algorithm has the possibility to decode more than one symbol at a time if the minimum code length is less than or equal to half of the width of the processing unit. When the minimum code length is larger than the half of the width of the processing unit, the proposed method can still increase the average symbols decoded in one table access (thus speeding up the decoding time). In fact, the experimental results of the test files show that the average number of decoded symbols at one time for the proposed method ranges from 1.91 to 2.13 when the processing unit is 10. The experimental comparisons show that, compared to the conventional binary tree search method and the level-compressed Huffman decoding method, the decoding time of the proposed method is a great improvement.

► The algorithm has the possibility to decode more than one symbol at a time.
► The proposed method increase the average symbols decoded in one table access.
► A heuristic strategy is proposed to deal with the problem of rereading Huffman stream.