一种新的H.264/AVC快速帧内预测模式选择判决算法

An Improved Intra-frame Prediction Mode Selection Algorithm for H.264/AVC

针对H.264视频编码标准帧内预测模式选择部分计算量大的问题,本文提出了一种快速帧内预测模式选择判决算法。该算法首先提出了一个新的算子,用来描述图像灰度分布信息,作为判断图像复杂程度的依据,进而提出两个对量化系数(QP)自适应的阈值,将图像分为平滑、变化剧烈和特征模糊3种。平滑图像宏块使用Intra_16×16模式预测,变化剧烈的宏块使用Intra_4×4模式预测,特征模糊的宏块使用标准算法预测。实验结果表明,该优化算法能在保证很好的图像质量的同时,使帧内模式选择计算次数减少30%以上,而且对传输码率基本没有影响。

In this paper, to simplify the computational complexity of intra-frame prediction mode selection for H.264, an improved Intra-frame prediction mode selection algorithm is proposed. Firstly, a new operator is proposed to describe the gray-level info of an image which could be the image complexity degree criterion. Then two QP （Quality Parameter）-based adaptive thresholds are proposed and images can be divided into three kinds： smooth, variability and in fuzzy feature. Intra _ 16× 16 mode is used when the macroblock is smooth, Intra _4×4 mode is used when the macroblock is variability and the original algorithm is used when the macroblock is in fuzzy feature. Experimental results show that the improved algorithm almost has 30% decrement on intra-frame mode selection times, meanwhile the image quality is still good enough with a neglectable rate increase.