Fast Intra-Prediction Mode Decision For H.264

H.264 has been approved in May, 2003 by ITU-T as Recommendation H.264 and by ISO/IEC as International Standard 14496-10 (MPEG-4 part 10) Advanced Video Coding (AVC). The elements common to all video coding standards are present in the H.264. And some new techniques, such as spatial prediction in intracoding, adaptive block size motion compensation, 4x4 integer transformation, multiple reference pictures and content adaptive binaryarithmetic coding (CABAC), are used in this standard. The testing results of H.264 show that it greatly outperforms existing video coding standards in both peak signal-to-noiseratio (PSNR) and visual quality.In order to achieve this, a robust rate-distortion optimization (RDO) technique is employed to select the best coding mode and reference frame for each macorblock. As a result, the computational burden of this type of brute force-searching algorithm is far more demanding than any existing video coding algorithm.To reduce the complexity, this paper presents two fast intra-prediction mode decision algorithms for H.264. As an important part of H.264, the intra coding has many prediction modes. And these modes is not only calculated in Intra frames, but also needed in P and B frames, so that the fast algorithm can significantly reduce the overall complexity.From the extensive experiment results of various QCIF video test sequence, we observed that a strong correlationship of the best Macroblock (MB) type exists between the neighboring Intra MBs. And the coding results of two intra MB types in one MB also have strong correlationship. Based on the results, we proposed a pre-decision algorithm of Intra MB type. And the algorithm is implemented on H.264 test model JM6.1. Experimental results show that the pre-decision algorithm contributes about 12% time saving with negligible loss of the quality.But the experiment results also show that the pre-decision algorithm does not contribute too much time saving, so the fast mode decision algorithm for every intra MB types is necessary for farther amelioration. In this paper, we present an improved fast algorithm using local edge information, and this algorithm considerably reduces the amount of calculations needed for intraprediction with acceptable loss of coding quality.