一种高清帧内预测单元快速选择算法
A kind of high definition intra-prediction unit fast selection algorithm

针对高效视频编码标准提出一种CU划分快速终止算法. 该算法选择图像纹理复杂度作为特征向量，将HEVC帧内预测与k-means聚类方法相结合，通过每种尺寸CU的聚类中心减少率失真优化的次数，提前终止CU块的划分，在达到降低编码复杂度的同时依然保持较好的编码效率. 测试结果表明，本算法与参考算法HM10.0相比较，视频图像质量基本不变，而编码时间大幅减少.
We proposed a CU fast selection algorithm for high efficiency video coding standard，we select the image texture complexity as the feature vector，combined the HEVC intra prediction with k-means clustering method，reduce times of the CU rate distortion optimization sizes and make early divided termination of CU blocks by each cluster centers，reduce coding complexity while still maintaining good coding efficiency. The test results show that the algorithm is compared with the reference algorithm HM10.0，video quality will remain basically unchanged，while significantly reducing encoding time