基于内容分析的低复杂度高效视频编码方法

项目名称： 基于内容分析的低复杂度高效视频编码方法

项目编号： No.61501246

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 无线电电子学、电信技术

项目作者： 潘兆庆

作者单位： 南京信息工程大学

项目金额： 22万元

中文摘要： 高效视频编码是最新一代的视频编码标准，它与H.264/AVC相比，压缩效率提高了约两倍。但是，高效视频编码的高压缩效率是以巨大的计算复杂度为代价的，巨大的计算复杂度阻碍了高效视频编码的广泛应用。传统的高效视频编码计算复杂度优化方法主要利用时、空相关性，优化单一方面的计算复杂度，通过分析视频内容的特征和编码编码参数选择之间的关系，可以进一步优化高效视频编码的计算复杂度。本项目首先在超低计算复杂度的情形下分析了视频内容的特征，构建了视频内容编码计算密集区域模型和图像局部内容间编码关系模型。然后，以这两个模型为基础，结合高效视频视频编码中计算复杂度分布特征，研究基于视频内容特征分析的快速编码树单元四叉树深度选择、快速帧间预测单元模式选择、快速多参考帧运动估计，从整体上降低高效视频编码的计算复杂度。本项目的研究成果为高效视频编码的广泛应用提供理论和方法基础，促进视频信号处理领域的研究与发展。

中文关键词： 高效视频编码；计算复杂度；编码树单元；预测单元；多参考帧运动估计

英文摘要： The state-of-the-art video coding standard High efficiency Video Coding (HEVC) obtains the outstanding coding efficiency. Compared with the H.264/AVC, it achieves about 50% bit rate saving. However, the achieved coding efficiency is at the cost of high computational complexity. The high computational complexity limits the HEVC encoder to be widely used in multimedia applications. The traditional complexity optimization methods mainly use the spatial and temporal correlations, and the complexity can be further reduced by analyzing the correlation between the image content and the coding parameters. The project firstly proposed an image intensive computing area model and an image local content coding correlation model, which are based on the results of image content analyses. Then, in order to reduce the overall computational complexity of the HEVC encoder, based on these two proposed models and the computational complexity distribution in the HEVC encoder, we research on the video content analyses based fast coding tree unit depth decision, fast inter prediction unit mode decision and fast multiple reference frames based motion estimation. The results of this project provides theories and methods for the HEVC to be widely used in multimedia applications, which will promote the development of the video signal processing related research areas.

英文关键词： High Efficiency Video Coding;Computational Complexity;Coding Tree Unit;Prediction Unit;Multiple Reference Frames Based Motion Estimation