Recognition of Abnormal Human Behavior in Surveillance Video Based on Machine Vision Technology and Multiscale Spatiotemporal Features
Monitoring is easily affected by the dynamic characteristics of optical flow and cannot accurately identify abnormal human behavior.To this end,a method for identifying abnormal human behavior in surveillance videos based on machine vision technology and multi-scale spatiotemporal features was proposed.A multi-scale spatiotemporal feature matrix for abnormal behavior was constructed,and useless interest points were eliminated;the multi-scale spatiotemporal feature matrices for continuous frames of human contour height and width,the previous and current frames of human posture changes,and the distance from human contour points to contour centers were extracted;by identifying abnormal behavior trajectories of jumping,squatting,and crawling,the recognition of human abnormal behavior was achieved.The experimental results show that the maximum errors between the jumping,squatting,and crawling abnormal behaviors identified by the proposed method and the actual values are 0.2 m,0.02 m,and 0.01 m,respectively,indicating a precise recognition effect.
machine vision technologymulti-scale spatiotemporal characteristicssurveillance videoabnormal human behaviorrecognition
万方数据
维普
