On Predictive-Based Lossless Compression of Images with Higher Bit Depths

Due to the rapidly increasing requirements for data transmission and storage, applications for fast and efficient compression of data have a very important role. Lossless compression must be applied when data acquisition is expensive. For example, lossless image compression must be applied in aerial, medical and space imaging. Besides the requirements for high compression ratios as much as it is possible, lossless image coding algorithms should be as fast as possible. During the late nineties of the previous century, many predictive-based algorithms for lossless compression of 8-bit images were introduced. These algorithms were usually expanded to enable processing of images with higher bit depths. All predictive based algorithms used more or less efficient predictors to remove spatial redundancy in images. This paper gives a comparative analysis of predictor efficiency with special emphasis on images with higher bit depths. A novel predictive-based, lossless image compression algorithm with a simple context-based entropy coder is presented, as well. A comparison with standardized lossless compression algorithms JPEG-LS and JPEG2000 is made on a large set of 12-bit medical images of different modalities and 12-bit and 16-bit natural images. It is shown that the proposed solution can achieve approximately the same bitrates as standardized algorithms even though it is much simpler.