An Improved Block Truncation Coding Using Rectangular Non-symmetry and Anti-packing Model

Abstract

AbstractBlock truncated coding (BTC) is an image compression algorithm with a simple coding process and a high coding speed, which can be used in the field of military communications with high real-time requirements. As the price of pursuing simplicity and high speed, the compression ratio and the quality of the decoded image are sacrificed to some extent. Although some strategies have been proposed to improve the compression ratio and the quality of the decoded images, the effect is not obvious. Inspired by the quadtree-based block truncation coding (QEDBTC) and the non-symmetry and anti-packing model (NAM), in this paper, we propose a novel rectangular NAM-based block truncation algorithm (RNAMEDBTC), which uses rectangular NAM strategy to divide the initial blocks into rectangular homogeneous blocks. The spatial frequency measurement (SFM) is used as a measurement parameter to subdivide the initial blocks. For each homogeneous block, we replace the high and low quantization values and the binary bitmaps in the traditional block truncation coding with the average value of the pixels in the block, thereby a great improvement of the compression rate of the algorithm is achieved. In order to further improve the compression rate, we have increased the area of the smaller homogeneous blocks, and thus reducing the number of homogeneous blocks. These expanded blocks are called non-homogeneous blocks. For each non-homogeneous block, we need to do error diffusion block truncation coding (EDBTC) processing. The experimental results in this paper show that without degrading the quality of the decoded image, the proposed algorithm improves the compression rate significantly by 158.3% and 30.8% higher than the traditional BTC and QEDBTC algorithms, respectively.KeywordsBlock truncation codingNon-symmetry and Anti-packing modelSFM