机场场面超分辨率视频的快速编码算法研究

With the rapid development of China's civil aviation industry, the security of air port scene becomes more urgent, especially when the small and medium-sized airports lack the necessary airport apron surveillance equipments. Therefore it is necessary to develop a airport scenes monitoring system with high efficiency to meet the needs of small and medium-sized airports. Enhanced Vision System(EVS) is developed by the CAAC. The solution aims at solving the problems of airport surface surveillance based on video processing technology. The programs compress and save high-resolution video sequences in real time. Because of the feature of high-resolution and the complexity of vedio coding , fast coding algorithms in airport apron environment are in urgen needs. This project uses the current HEVC coding standards. In view of specific characteristics of airport scenes, encoding is divided into three phases as stationary, inside movement and inner coding. Inside movement phase contains two states of leaving and arrival. By studying the distribution of units,modes and partitions in these regional and state units, and combined with temporal degree and spatial degree to predict the possibilities of units,modes and partitions. At the end of the project, we study the conditions needed for early termination corresponding to various regions and states. Results show the algorithmn can improve encoding speed which would promote the applicion of the system in the small and medium-sized airports..

随着我国民用航空的快速发展，机场场面安全保障的要求愈加迫切，尤其是中小型机场还缺乏必要的机场场面监视设备。因此有必要开发一套能满足中小机场场面监视需求，同时具有高性价比性能的场面监视系统。增强视景监视系统（EVS）是民航二所研发的用于解决机场场面监视问题的基于视频处理技术的专业解决方案，该方案对超分辨率视频序列进行实时压缩和保存，由于视频序列分辨率很大并且视频编码复杂，迫切需要研究基于机场环境的超分辨率视频的快速编码算法。本项目采用HEVC编码标准，针对机场场景的功能特性，编码时分割为静止、运动内部和内部编码三种区域，运动内部又包含起飞和达到两种状态。研究这些区域和状态的单元、模式和划分的分布规律，然后对各个区域和状态结合时间和空间相关性程度预测单元、模式和划分的可能性程度，最后对各个区域研究相应的提前终止判断条件。该算法的研究成果可以较大地提高编码速度，而推动整个系统在我国中小机场的应用

随着我国民用航空的快速发展，机场场面安全保障的要求愈为迫切，尤其是中小型机场还缺乏必要的机场场面监视设备。因此有必要开发一套能满 足中小机场场面监视需求，同时具有高性价比性能的场面监视系统。增强视频监视系统 （EVS）是民航二所研发的用于解决机场场面监视问题的基于视频处理技术的专业解决方案，该方案对超分辨率视频序列进行实时压缩和保存。而项目主申请单位清华大学深圳研究生院和民航第二研究所共同协作，主要研究内容是针对视频内容进行如下处理：一、高效视频编码即HEVC下视频编码技术，2、 机场场面的场景识别，如飞机起飞、降落和静止区域等，从而对不同场景的视频进行处 理和高效视频压缩；三、对各区域的目标进行识别和监控。