An image coder for lossless and near lossless compression

In this paper, we propose a new image coder (DACLIC) for lossless and near lossless image compression. The redundancy removal in DACLIC (Direction and context-based lossless/near lossless image coder) is achieved by block direction prediction and context-based error modeling. A quadtree coder and a postprocessing technique in DACLIC are also described. Experiments show that DACLIC has higher compression efficiency than the ISO standard: LOCO-I (Low complexity lossless compression for images). For example, DACLIC is superior to LOCO-I by 0.12bpp, 0.13bpp and 0.21bpp when the maximum absolute tolerant error n = 0, 5 and 10 for 512 x 512 image "Lena". In term of computational complexity, DACLIC has marginally higher encoding complexity than LOCO-I but is comparable to LOCO-I in decoding complexity.