远程全景实时监控系统设计与实现

针对现有实时视频通信系统因视频编解码模块和传输协议之间协作不紧密影响视频通信的传输性能和视觉质量,以及常用的远程实时监控系统监控范围和角度有限,多画面独立显示不够直观的问题,采用新型紧耦合式的视频通信架构,通过逐帧联合控制视频的编解码和传输,优化了每一帧的编码长度,提高了实时监控视频传输的端到端时延性能和视频图像质量;设计了多路远程视频源间的帧同步算法,并基于OpenCV的图像拼接模块定制了图像拼接流程;通过在特征点提取和匹配、图像变形和图像融合阶段使用GPU加速,并在逐帧拼接时设置合适的拼接缝更新间隔,实现了监控视频的近实时拼接。实验证明,系统能够获得良好的监控视频质量和更广的监控范围,一定程度上解决了视频监控中的盲区问题,同时具有较好的实时性和视觉效果。

Design and Implementation of Remote Real-time Panoramic Monitoring System

Aiming at the problem that the collaboration between the video codec and the transmission protocol in existing real-time video communication systems is not close,which affects the transmission performance and visual quality of video communication,the commonly used remote monitoring system has a limited monitoring range and angle,and the video is not intuitive enough with multi videos displayed independently,a new kind of tightly coupled video communication architecture is adopted to jointly control the video encoding,decoding and transmission frame by frame,which optimizes the encoding length of each frame and improves the end-to-end delay performance and image quality of real-time monitoring video transmission.A frame synchronization algorithm between multiple channels of remote video is designed,and an image stitching process is customized based on OpenCV image stitching module.By using GPU acceleration at the stage of feature extraction and matching,image deformation and blending,setting appropriate stitching seam update interval,near real-time stitching is achieved.Experiment shows that this system can obtain better monitoring picture quality and wider monitoring range,which solves the problem of blind spots in video monitoring to some extent,with well real-time capability and visual effects at the same time.