An improved fast intra-prediction algorithm in H.264/AVC based on edge feature

H.264/AVC is the newest video coding standard with high compression efficiency. But it can not achieve real-time application because it takes a long time to code with extremely complex algorithm. In this paper, according to the characteristic of intra-prediction algorithm, a fast algorithm based on Edge feature is proposed. To reduce the number of intra-prediction mode, some candidate modes are selected based on the edge features or Key Edge Strength (KES) of the image, and then related RDO is calculated. We also simplify Chroma prediction based on the relativity between Luma and Chroma prediction to further reduce the computational complexity. Experimental results show that the proposed fast intra-prediction algorithm reduces about 80% computation with negligible PSNR degradation and slight bit-rate increment. Compared with the PAN's sobel operator based on edge detection algorithm, it also has obvious superiority.