基于决策树的HEVC SCC帧内编码快速算法

为减少HEVC屏幕内容编码的编码时间,提高编码 效率,本文提出了一种基于决策树的HEVC屏幕内容帧内编码快速 CU划分和简单PU模式选择的算法。对视频序列特性分析,提取有效的特征值,生成决策树模 型。使用方差、梯度信息熵和 像素种类数用于生成CU划分决策树,使用平均非零梯度、像素信息熵等用于生成PU模式分类 决策树。在一定深度的决策 树模型中,通过对相应深度的CU的特征值的计算快速决策当前CU的划分与PU模式的类型。这 种利用决策树做判决的算法 通过减少CU深度和PU的模式遍历而降低编码复杂度,达到快速帧内编码的效果。实验结果表 明,与HEVC屏幕内容的标 准算法相比,该算法在峰值信噪比(PSNR)平均下降0. 05 dB和码率 平均增加1.15%的情况下,能平均减少30.81% 的编码时间。

A fast intra coding algorithm for HEVC SCC based on decision tree

In order to reduce the encoding time of HEVC screen content coding and improve coding efficiency,this paper proposes a fast intra coding algorithm based on decision tree to achieve the impr ovement in CU partitioning and simple PU mode selection.Through the analysis of video sequence characteristics,effective features are proposed,and the decision tree model is trained,in which the variance,gradient information entropy and pixel type number are used for CU partition,and the average non-zer o gradient,pixel entropy information,etc.are used for PU mode classification.In a decision tree model,the current CU partition and the PU mo de type are quickly determined by calculating the features of the CUs of the depth.The algorithm for decision tree is used to mak e the decision to reduce the cycle process about CU depth and PU modes in intra coding,thereby achieving fast intra c oding.The experimental results show that compared with the standard algorithm of HEVC screen content coding,the proposed algorithm can reduce the coding time by 30.81% on average when the peak signal-to-noise ratio (PSNR) decreases by 0.05dB and the code rate increases by 1.15%.