基于固定背景的视频浓缩算法设计与实现

Design and Implementation of Video Concentration Algorithm Based on Fixed Background

视频浓缩是通过图像处理和机器视觉等手段来完成从长时间的监控视频中提取到有价值的内容信息。在视频监控领域中,仅通过人力进行视频的筛查会造成效率低下以及漏报和误报的问题。文章通过相邻帧内容检测方法,运动片段检测方法配合Harris特征点提取算法和SIFT特征点提取算法实现图像特征点配准。提出双模融合的背景建模与内容检测方法,并且采用基于2D距离的运动内容跟踪方法以及GrabCut抠图算法进行了对视频浓缩算法的设计与实现。

Surveillance video is to extract valuable content information from long-term surveillance videos through image processing and machine vision.In the field of video surveillance,only manual video screening will cause inefficiency and problems of under-reporting and false-reporting.The article uses the adjacent frame content detection method,the motion segment detection method to cooperate with the Harris feature point extraction algorithm and the SIFT feature point extraction algorithm to achieve image feature point registration.A dual-mode fusion background modeling and content detection method is proposed,and a 2D dis-tance-based motion content tracking method and GrabCut matting algorithm are used to design and implement the video concentration algorithm.