Fast Intra- and Inter-Prediction Mode Decision of H.264/AVC for Medical Frame Compression based on Region of Interest

This paper aims at applying H.264/AVC in medical video compression applications and improving its compression performance with higher perceptual quality and lower coding complexity. We propose a new method that uses lossless compression in the region of interest (ROI) and very high rate lossy compression in other regions. The propose method achieves a fast intra- and inter- prediction mode decision that is based on encountering coarse MBs for intra- and inter- prediction mode decision of the background region and fine MBs for the ROI region. The MBs of the background region are encoded with the maximum quantization parameter allowed by H.264/AVC in order to maximize the number of null coefficients. Also, in order to further reduce the computational complexity, a two-adaptive search range decision method (proposed previously by the authors) is enhanced using the ROI concept. Experimental results show that the proposed algorithm achieves a higher compression rate on medical videos with a higher quality of ROI with low coding complexity when compared to other standard algorithms reported in the literature and our previous algorithm.