基于边缘方向信息的高效帧内预测算法

AN EFFICIENT INTRA-FRAME PREDICTION ALGORITHM BASED ON EDGE DIRECTION INFORMATION

为了降低H.264/AVC中帧内预测算法的复杂度,以便满足实时性要求较高的场合,提出基于宏块的灰度直方图能量分布特性,对宏块预测方式intra_4×4和intra_16×16预判;然后对Pan算法所存在的缺陷加以改进,给出一种新的帧内预测算法。实验结果表明,在全I帧编码情况下,该算法与全搜索算法相比编码时间平均减少了68.9%,与Pan算法相比,编码时间减少最大超过37%,在保证图像质量与压缩性能的前提下,可以有效地降低计算复杂度,从而大大提高编码速度。

In order to reduce the complexity of H. 264/AVC intra-frame algorithm so as to fit the situations requiring higher real-time property, in this paper we propose that to anticipate the macro block prediction modes intra 4 ~ 4 mode and intra_16 x 16 according to the energy distribution characteristics of grey histogram of macro block ; and then to improve the defects existed in Pan＇ s algorithm, thus we work out a new intra-frame prediction algorithm. Experimental results demonstrate, under the condition of all-I-frame encoding, compared with the full-search algorithm, the proposed algorithm reduces the encoding time by 68.9% in average, while compared with Pan＇ s algorithm, more than 37% timesaving in regard to encoding can be archived. The algorithm is able to effectively reduce the computational complexity therefore to significantly increase the encoding speed without sacrificing the image quality and compression performance.