A gradient intensity-adapted algorithm for the fast decision of H.264/AVC intra-prediction modes

H.264/AVC is widely applied for the encoding of videos due to its superiority of finding an optimal tradeoff between the bit-rate expense and the video quality. Nevertheless, the procedures of finding an optimal intra or inter-prediction mode can degrade the run-time performance of the coding process since considerable computations on rate-distortion cost evaluation are required for different prediction modes. For a macroblock, hundreds of rate-distortion costs have to be calculated before an optimal intra-prediction mode can be determined, which is really a heavy burden for the encoder. To alleviate the computational complexity of the coding process, we apply in this paper a simple yet effective gradient evaluation approach so that the texture orientation can be found efficiently. Moreover, we propose an adaptive selection strategy in this paper so that only a variable number of the most probable intra-prediction modes will be sent for the rate-distortion optimization process. Experimental results show that a noticeable speedup on the runtime performance can be obtained with only a minor degradation on the visual quality and bit-rate expense. When compared with existing state-of-the-art fast decision algorithms, a distinct improvement on a proposed cost-performance metric can be attained, which demonstrates the superiority of the proposed approach.