Human body motion capture and posture analysis system

Abstract

The invention discloses a human body motion capture and posture analysis system and a method for carrying out foreground detection and automatic extraction of joint points through the human body motion capture and posture analysis system. According to the human body motion capture and posture analysis system, a human body motion image is captured through a camera, the human body foreground and the human body joint points are extracted from the human body motion image, and then human body motions are analyzed through a tracking and recognition algorithm. Due to the fact that a human body target in a video is extracted through an algorithm combining a background difference method and a frame difference method, the defect that background noise interference easily exists in the process of foreground detection based on the background difference method is overcome; due to the fact that the method with combination of a plurality of algorithms is adopted, automatic initialization of the human body joint points is achieved, and the defects that optical marks or manual marking is needed by a traditional joint point extraction method, and an existing automatic extraction method is large in calculated amount and poor in accuracy are overcome. The human body motion capture and posture analysis system is low in complexity and high in accuracy, and the intelligence and the efficiency of the human body motion capture and posture analysis system are improved.

Description

technical field [0001] The invention relates to a human body motion capture and posture analysis system, especially a system that can be applied to the field of computer vision. Background technique [0002] At present, in the known human motion capture and posture analysis systems, the background difference method is usually used for foreground detection, and the Gaussian model is often used for background modeling. This method has a large amount of calculation, slow speed, and is easily affected by shadows. The detected target image has noise. , and objects that pause in motion are absorbed as part of the background, resulting in holes inside the objects. Moreover, optical marking or manual calibration is usually used for joint point extraction, which has a large human error and a low degree of automation, and the current automatic extraction method usually uses a curve fitting method, which has a large amount of calculation and low accuracy, and is not universally applica...

AI Technical Summary

Problems solved by technology

[0003] In order to overcome the defects of low accuracy of foreground detection results and low degree of automation in the joint point extraction process of the existing human motion capture and posture analysis system, the present invention provides a human motion capture and posture analysis system, which can improve the foreground detection effect and automatically Initialize human body joints

Images