VPx帧内压缩的快速算法  

Fast algorithm for VPx intra-frame compression

作　　者：胡金艳[1] 宋绍京[1] Hu Jinyan;Song Shaojing(School of Computer and Information Engineering,Shanghai Polytechnic University,Shanghai 201209,China)

机构地区：[1]上海第二工业大学计算机与信息工程学院,上海201209

出　　处：《电子技术应用》2020年第4期75-78,88,共5页Application of Electronic Technique

基　　金：校企合作汽车电子联合实验中心建设项目(A11NH190704);校重点学科建设项目(XXKZD1605)。

摘　　要：针对VPx视频的帧内压缩处理复杂度高的问题,给出一种快速算法。该方法首先根据图像中亮度子块在宏块中的空间位置确定最邻近子块,再通过最邻近子块的最佳预测模式、子块重构和率失真比较判断当前子块的可能预测模式,并结合率失真阈值得到用于该子块帧内压缩的最终预测模式。实验结果表明,该方法能够有效地减少VPx帧内压缩中由重构处理带来的包括频域变换、量化、反量化、频域反变换的大量运算,从而在保证压缩质量的情况下,提高帧内压缩速度。Aiming at the high complexity of VPx video compression,a fast algorithm is proposed.The nearest sub-block is firstly determined according to the spatial position of the luma sub-block in the macro-block in an image.Then the possible prediction mode of the current sub-block is judged by the best prediction mode of its nearest sub-block,sub-block reconstruction and rate distortion comparison.The final prediction mode for intra-frame compression of the current sub-block can be obtained by introducing the rate-distortion threshold.The experimental results show that this method can effectively reduce the computational complexity including frequency domain transform,quantization,inverse quantization and inverse frequency domain transform caused by the reconstructions in VPx intra-frame compression.Thus,the performance of intra-frame compression is improved while the compression quality is guaranteed.

关 键 词：VPX 帧内压缩 子块重构 预测模式 

分 类 号：TN919.81[电子电信—通信与信息系统]