Local average-based model of probabilities for JPEG2000 bitplane coder

Context-adaptive binary arithmetic coding (CABAC) is the most common strategy of current lossy, or lossy-to-lossless, image coding systems to diminish the statistical redundancy of symbols emitted by bitplane coding engines. Most coding schemes based on CABAC form contexts through the significance state of the neighbors of the currently coded coefficient, and adjust the probabilities of symbols as more data are coded. This work introduces a probabilities model for bitplane image coding that does not use context-adaptive coding. Modeling principles arise from the assumption that the magnitude of a transformed coefficient exhibits some correlation with the magnitude of its neighbors. Experimental results within the framework of JPEG2000 indicates 2% increment on coding efficiency.