Document image compression by nonlinear binary subband decomposition and concatenated arithmetic coding

This paper proposes a new subband coding approach to compression of documen t images, which is based on nonlinear binary subband decomposition followed by the concatenated arithmetic coding. We choose to use the sampling-exclu sive on (XOR) subband decomposition to exploit its beneficial characteristi cs to conserve the alphabet size of symbols and provide a small region of s upport while providing the perfect reconstruction property. We propose a co ncatenated arithmetic coding scheme to alleviate the degradation of predict ability caused by subband decomposition, where three high-pass subband coef ficients at the same location are concatenated and then encoded by an octav e arithmetic coder. The proposed concatenated arithmetic coding is performe d based on a conditioning context properly selected by exploiting a nature of the sampling-XOR subband filter bank as well as taking the advantage of noncausal prediction capability of subband coding. We also introduce a unic olor map to efficiently represent large uniform regions frequently appearin g in document images. Simulation results show that each of the functional b locks proposed in the paper performs very well, and consequently, the propo sed subband coder provides good compression of document images.