Abstract: Targeted on unattended operation requirements of video surveillance in power systems, a research and implementation solution of Intelligent Video Analysis（IVA） system based on image processing techniques is proposed. HD real-time video streams are analyzed through image processing and feature extraction techniques such as connected component analysis, code mapping, Hough line detection, contour detection, Gaussian mixture foreground detection and cvBlob tracking. Several functionalities related to device state recognition and security abnormal identification are implemented in proposed IVA system. The experimental results verify the efficiency and accuracy of each functionality, and outperform state of the art methods in relevant literature to indicate the improvement.

Key words: substation HD video surveillance, Intelligent Video Analysis（IVA）, device state recognition, security abnormal identification

摘要： 针对电力系统在视频监视方面的无人值守需求，提出了一种基于图像处理技术的视频智能分析系统研究和实现方案。通过连通域分析、编码映射、Hough直线检测、轮廓检测、高斯混合前景检测、cvBlob目标跟踪等图像处理和特征提取技术对高清实时视频流进行分析，实现了几种设备状态识别和安防异常行为识别功能。实验结果验证了各分析功能的实时性和准确性，并对比部分功能在相关文献中的实验结论，以表明该系统方案的有效改进。

关键词: 变电站高清视频监控, 视频智能分析, 设备状态识别, 安防异常识别