Context-Based Lossless Compression of Mosaic Image with Bayer Pattern

In this paper, a predictive model for compression of mosaic image with Bayer pattern is proposed. It consists of TFNN neural network predictor and adaptive correction part based on context. As in JPEG-LS, the adaptive part of the predictor is context-based and it is used to ldquocancelrdquo the integer part of the offset due to the TFNN predictor. In the meantime, we propose a context quantization approach that achieves high coding efficiency. Compared with existing methods of CFA image lossless compression, the performance of proposed method is apparently the best.