结合全色度HEVC和有损字典算法的屏幕图像编码

发布时间：2018-04-15 04:27

本文选题：高效视频编码 + LD-Cost　；参考：《计算机科学》2014年03期

【摘要】：鉴于传统视频压缩算法对屏幕视频编码的效果不理想,提出了一种误差限自适应的有损字典编码方案,并和全色度HEVC有机融合,形成LDSC(Lossy Dual-coder Single Chroma-sampling-rate)算法。该算法根据屏幕内容中不同类型区域的编码特性,自适应地选择率失真性能较好的编码结果放入码流。为了评价有损字典编码的性能,提出了基于匹配长度和匹配失真的LD-Cost(Length Distortion Cost)模型,并使用拉格朗日乘子法对模型进行分析和计算。对于有损字典编码过程中可能出现的误差积累现象,给出了可行的解决方案。实验结果表明,与基于无损字典编码的双编码器方案相比,在不损伤主观图像质量的前提下,LDSC算法能明显提高编码性能,对连续色调内容居多的屏幕视频也表现出更好的编码适应性,全I帧配置下BD-rate性能比基于无损字典编码的方案提高了3%~15%,与单纯使用HEVC相比,BD-rate性能提高了7%~49%。
[Abstract]:In view of the unsatisfactory performance of the traditional video compression algorithm for screen video coding, a lossy dictionary coding scheme with adaptive error limits is proposed and combined with the full chroma HEVC to form a LDSC(Lossy Dual-coder Single Chroma-sampling-ratealgorithm.According to the coding characteristics of different types of regions in the screen content, the algorithm adaptively selects the coding results with better rate-distortion performance and puts them into the code stream.In order to evaluate the performance of lossy dictionary coding, a LD-Cost(Length Distortion cost model based on matching length and matching distortion is proposed, and the Lagrange multiplier method is used to analyze and calculate the model.A feasible solution to the possible error accumulation in the process of lossy dictionary coding is presented.Experimental results show that the LDSC algorithm can improve the coding performance without compromising the subjective image quality compared with the dual-encoder scheme based on lossless dictionary coding.It also shows better coding adaptability for screen video with more continuous tone content. The performance of BD-rate in full I frame configuration is improved by 3 / 15 compared with the scheme based on lossless dictionary coding. Compared with using HEVC alone, the performance of BD-rate is improved by 7% and 49%.
【作者单位】：同济大学超大规模集成电路研究所;
【基金】：国家自然科学基金(61201226,61271096) 上海市自然科学基金(12ZR1433800) 中央高校基本科研业务费专项资金(2810219002,2810219003)资助
【分类号】：TN919.81

【参考文献】