Fast Intraprediction Algorithm for HEVC Based on Machine Learning Classification Technique

An emerging video coding standard High Efficiency Video Coding (HEVC) accomplishes 50% reduction in encoding time respect to before standards H.264. In order to reduce the high computational complexity of 35 intra prediction modes in HEVC, the proposed scheme exploits the correlation between the reconstructed with current pixels horizontally and vertically. This work presents a fast intra prediction mode classification by using Machine Learning technique. Support Vector Machine (SVM) is used for decision making over the selected intra prediction mode classification which significantly reduces the number of modes. Simulation results shows that the suggested method reduced the encoding time on average with 32.01% with increase in Bjontegaard Delta bitrate (BD) at a rate of 0.87%.