Human behavior detection method and device

Abstract

The invention discloses a human behavior detection method and device. The method comprises the following steps: image frames of a video to be detected are obtained; a global optical flow field between the obtained two adjacent image frames is determined, and motion information between the two adjacent image frames is obtained; according to the obtained motion information, at least one motion region having a default motion direction is obtained, wherein the motion region includes at least one pixel for forming an image frame; the filtering process is performed on the obtained motion region, and the default human behavior shape matching process is performed on the filtered motion region; and according to the matching result, human behaviors included in the video to be detected can be determined. By adopting the technical scheme, the advantage of good application universality can be realized, the human behavior detection accuracy in a video monitoring system can be well improved, and the consumption of resources can be reduced.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a human behavior detection method and device. Background technique [0002] Intelligent video surveillance is based on digital and networked video surveillance, but it is different from ordinary networked video surveillance and belongs to a higher-end video surveillance application. The intelligent video surveillance system can identify different objects, find abnormal conditions in the surveillance screen, and can issue alarms and provide useful information in a faster and better way, so that it can better and effectively assist security personnel to deal with crises and maximize security. reduce false positives and false negatives. [0003] In the prior art, there are mainly two ways to detect human behavior through video surveillance. One way is to record human behavior through video surveillance, and then the network management personnel will manually monitor and re...

AI Technical Summary

Problems solved by technology

However, in the specific implementation of this method, it is necessary to collect a large amount of sample video information for labeling, and then propose relevant features from the collected large amount of video sample information for machine learning, and finally a human behavior detection model can be obtained. model to realize the analysis of the video to be detected, so the accuracy is low and it is a waste of resources

[0004] To sum up, the human behavior detection method proposed in the prior art has low accuracy and wastes resources

Images