面向视频转码的视频去隔行研究

项目名称： 面向视频转码的视频去隔行研究

项目编号： No.61300122

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 自动化技术、计算机技术

项目作者： 黄倩

作者单位： 河海大学

项目金额： 22万元

中文摘要： 视频去隔行通常用于在视频解码后端消除隔行效应，以增强主观视觉效果。本项目将面向视频转码应用背景，着重研究视频编码前端的去隔行理论与方法。首先，研究运动信息的可靠性度量方法，重点考虑基于最大后验概率模型的运动信息可靠性度量方法，以便合理选用码流中的运动信息。其次，面向实时转码需求，基于区域特性研究细粒度的编码端快速去隔行理论与方法，并着眼于提高对突变运动场景的鲁棒性。最后，针对H.264/AVC标准、AVS标准以及最新的HEVC国际标准的特性，研究压缩域的去隔行理论与方法，尤其是部分可逆的去隔行理论与方法。预期上述理论和方法不仅有利于视频压缩和视频转码的研究，还能为视频检索等相近领域提供一种新的研究思路。

中文关键词： 视频压缩/编码；去隔行；智能监控；低成本；高性能

英文摘要： Video deinterlacing is often used to eliminate interlaced artifacts at the video decoding side for better subjective performance. This project aims at video transcoding applications, especially focusing on deinterlacing theories and methodologies at the video encoding side. Firstly, we will study the reliability, especially the maximum-a-posterior-based reliability, of motion information to reasonably select the motion information in bit-streams. Secondly, fine-grained fast deinterlacing theories and methodologies will be studied at the video encoding side based on region characteristics. The robustness to sudden changes like scene changes will be specially addressed in this part. Finally, frequency-domain deinterlacing theories and methodologies, especially the partially invertible deinterlacing theories and methodologies, will be studied based on characteristics of the H.264/AVC standard, the AVS standard, and the latest High Efficiency Video Coding (HEVC) international standard. It is expected that the above theories and methodologies will not only be beneficial to video compression and video transcoding, but can also open a new door for related research fields such as video retrieval.

英文关键词： video compression/coding；deinterlacing；intelligent surveillance；low-cost；high-performance