High Efficient and Real Time Huffman Codec Used In Handwriting Short Message Service

The author presents in this paper an enhanced Huffman codec with high speed, high efficiency, and low memory usage for use with a camera-equipped mobile telephone in a "handwriting short message service" (HSMS) system. The proposed codec relies on the fact that images transferred using HSMS usually consist of long runs of white pixels and short runs of back pixels. In the encoding process, a single-side growing 12-bit Huffman code-table is defined to efficiently reduce the amount of redundant data, and the number of necessary flags within the bit stream minimized. Real-time Huffman decoding is achieved through a group-based fast matching algorithm to accelerate searching for a symbol in the continuous coding bit stream by dividing the code tree into clusters, and by avoiding memory sparsity. The experimental results show that the new Huffman coding method performs significantly better. The compression ratio is 5 times higher than that of the JPEG standard, the computation speed is 4 times faster than that of classical Huffman decoding and the entire memory space requires only 84 words.