Abnormal behavior detection method in video based on target positioning and characteristic fusion

Abstract

The invention discloses an abnormal behavior detection method in a video based on target positioning and characteristic fusion. The method comprises motion area detection based on background difference and optical flow statistics, characteristic fusion based on a space-time characteristic and a depth characteristic, and decision based on multi-SVM model training. In the invention, for each dimension characteristic, a classifier is trained respectively, and finally an integration learning method is selected to carry out abnormal detection decision. The invention aims at detecting a motion target area through a rapid detection means, traditional traversing small block detection is improved, calculating efficiency is improved, and simultaneously through fusion of a characteristic layer surface and a model decision layer surface, detection accuracy is increased.

Description

technical field [0001] The invention relates to the field of intelligent video monitoring, in particular to a method for detecting abnormal behavior in video based on target positioning and feature fusion. Background technique [0002] Abnormal behavior detection, as a computer intelligent video analysis method, has potential application value in the field of intelligent monitoring, and has a great role in promoting public security, improving user experience, and reducing labor costs. In addition, because the actual video scene usually It is complex and changeable. Abnormal behavior detection will face difficulties such as occlusion, illumination changes, viewing angle changes, scale changes, crowd crowding, and variability of the same behavior. It is necessary to comprehensively use theoretical methods in the fields of image processing, computer vision, and machine learning. , which has great challenge and research value. At present, the mainstream abnormal behavior detect...

AI Technical Summary

Problems solved by technology

[0002] Abnormal behavior detection, as a computer intelligent video analysis method, has potential application value in the field of intelligent surveillance, and has a great role in promoting public security, improving user experience, and reducing labor costs; in addition, because the actual video scene is usually It is complex and changeable. Abnormal behavior detection will face difficulties such as occlusion, illumination changes, viewing angle changes, scale changes, crowd crowding, and variability of the same behavior. It is necessary to comprehensively use theoretical methods in the fields of image processing, computer vision, and machine learning. , with great challenge and research value

At present, the mainstream abnormal behavior detection methods in video mainly include detection based on target tracking and detection based on spatiotemporal features. The detection effect of the former algorithm depends on the tracking effect of the target and the extraction of motion trajectory features, while the detection effect of the latter algorithm Relying on the design of spatio-temporal features, at the same time, the latter algorithm usually needs to traverse every small area of ​​each frame of video to locate the abnormal position, which requires a large amount of calculation

Images