Human body behavior spatio-temporal conjoint analysis method based on cellular automaton idea

Abstract

The invention discloses a cellular automaton idea-based human body behavior spatio-temporal conjoint analysis method. The method comprises the following steps of: discretizing obtained three-dimensional skeleton posture key points of a human body based on the cellular automaton idea; secondly, energy smooth values of the human body three-dimensional skeleton posture key points are calculated in combination with the energy values of the human body three-dimensional skeleton posture key points and position changes of the human body skeleton key points; and finally, constructing a human body spatio-temporal conjoint analysis deep network, and inputting the three-dimensional coordinate information extracted from the human body falling video frame by frame and the calculated smooth value of the energy change after the key change of the human body into the constructed human body behavior spatio-temporal conjoint analysis network for training so as to carry out human body posture detection. According to the method, a behavior analysis technology in the field of machine vision is simplified from complex environment human body image recognition processing to simple conditions of coordinates of human body three-dimensional skeleton key points and energy changes of the human body skeleton key points, computer computing power required by behavior analysis is reduced, and the method has a wide application scene.

Description

technical field [0001] The invention belongs to the field of attitude detection in the field of machine vision, and in particular relates to a spatiotemporal joint analysis method of human behavior based on the idea of ​​cellular automata. Background technique [0002] The research content of human pose recognition is a human pose restoration method composed of fixed joint points obtained from the observation of images. Human posture detection needs to locate the key nodes of the human body in a complex and changing environment. The focus of human body recognition includes human body detection and key point location. Deep learning and human pose estimation technology are developing rapidly today, and the research on human behavior analysis has also received extensive attention. Related researchers have proposed a large number of human pose detection methods. [0003] In today's target-specific behavior analysis, most researchers directly process the data of two-dimensional ...

AI Technical Summary

Problems solved by technology

But at the same time, the detection of the two-dimensional human skeleton model will have errors in many cases, and the extracted key points will also be blocked due to a specific action in the three-dimensional space where the target is located. Behavioral analysis of the results of the results has a considerable impact, making its judgments based on the continuity of the image error

Although there are 3D skeleton and bone key point extraction technologies nowadays, compared with the 2D skeleton and bone key point extraction technology, the environmental information in 3D space is geometrically increased compared with the environmental information in 2D space, and the current three-dimensional The more accurate methods of human body pose estimation have higher requirements on equipment, such as motion capture and other methods

This has led to the fact that although the accuracy of 3D human pose estimation is significantly higher than that of 2D, the disadvantages are that the research cost is high, the demand for computing power is large, and the timeliness of detection in complex environments is not good. Extraction occupies an important position in the human behavior recognition monitoring system, and its effectiveness has a significant impact on the results of subsequent behavior recognition

Images