A body conflict behavior detection method based on low-dimensional spatio-temporal feature extraction and topic modeling

Abstract

The invention belongs to the technical field of video surveillance, and relates to a method for detecting body conflict behaviors based on low-dimensional spatiotemporal feature extraction and theme modeling. The detection steps are to first define a word book, then quantify the pixel position of the object, and describe the scene in the scene. The size of the foreground target and the movement of the foreground pixels are determined. After the above steps, the establishment of a complete vocabulary and corpus is completed. The judgment of the physical conflict behavior is performed through the above calculation method. This method combines low-dimensional data feature representation and Model-based complex scene analysis, using the changes in human body position information in the action, learns an overall motion model that has nothing to do with body parts. By analyzing the overall motion model, the detected results are compared with the parameters in the model, and then judged. Compared with the prior art, the method of the present invention has ingenious design concept, scientific detection principle, simple detection method and high detection accuracy, and has great market prospect.

Description

Technical field: [0001] The present invention belongs to the field of video surveillance technology, involving a method of detecting limb conflict behavior, and more particularly to a method of detecting limb conflict behavior based on low-dimensional time and space characteristics and topic modeling. Background technique: [0002] In recent years, with the increase of various safety emergencies, the promotion of public safety awareness, accompanied by the penetration of artificial intelligence concept, intelligent monitoring is increasingly received by people. Traditional monitoring systems are mainly implemented by manual monitoring, lack of real-time and initiatives. In many cases, video surveillance has not been supervised by the operation of the video backup due to unmanned management. In addition, with the popularity and extensive surveillance camera, traditional artificial monitoring methods can no longer meet the needs of modern monitoring. To solve this problem, the publ...

AI Technical Summary

Problems solved by technology

Existing inventions for abnormal behavior detection have the inherent characteristics of failure to truly understand abnormal behavior, so the existing abnormal behavior detection model cannot fully reflect the essence of abnormal behavior, resulting in the detection accuracy obtained based on the existing abnormal behavior detection model. The ideal effect was not achieved. Therefore, a detection method for body conflict behavior based on low-dimensional spatiotemporal feature extraction and topic modeling was designed. The calculation method is accurate and the detection result is accurate.

Images