Abstract
We propose a lossless video frame compression algorithm employing a dictionary coding method, the Huffman coding method and three schemes to achieve high compression ratio. We observe the smaller the absolute value of the differentials between the current pixel and its neighbors the higher the probability is by analyzing the distribution of this differentials. According to this distribution, we compute the data reduction ratio (DRR) for cases using different numbers of code words and find the more code words used the higher the DRR which approached a plateau. Considering memory usage, we choose a suitable number of code words for Huffman encoding. We employ a two-staged classification (TC) scheme consisting of the dictionary coding method and a longest prefix match (LPM) method. The LPM method we choose for each pixel group a best truncation length (BTL) using an adaptive prefix bit truncation (APBT) scheme. We further compress the code words by a head code compression (HCC) scheme. Due to large numbers of code words used, we can achieve about 0.5% more bit rate reduction compared to previous proposed algorithm and only 0.96% bit rate reduction less than using the maximum dictionary size.