Human body posture estimation method and human body posture estimation system

Abstract

The invention provides a human body posture estimation method and a human body posture estimation system. The method includes the following steps: A, building a human body structuralized model, and combining with information of a sensor to calculate posture of human body motion and position information in a geodetic coordinate system; B, building a Bayesian network model of human body motion, combining with the posture of human body motion and the position information in the geodetic coordinate system to solve the geodetic coordinate system, and predicting human body posture of a later frame by a belief propagation algorithm in the Bayesian network model through human body posture data of a former frame. The human body posture estimation method and the human body posture estimation system have the advantages that human body posture can be predicted, operation efficiency of the whole system is improved remarkably, and predicting instantaneity of the system is guaranteed.

Description

technical field [0001] The invention relates to the field of data processing, in particular to a method and a system for estimating human posture. Background technique [0002] At present, the existing methods for estimating human body posture are mainly based on digital image processing technology, which involves the field of computer vision. [0003] One of them is to use an ordinary camera to collect human body motion video sequences, and then input the human body motion video images into the computer, and use digital image processing related technologies on the computer to process the human body images (obtain the human silhouette through the background difference, extract the human body contour, and analyze the human body. The outline is refined to obtain the skeleton line of the human body, and then combined with some algorithms to find the positions of each joint point of the human body, so as to construct a human body model on the computer), and then combine the real...

AI Technical Summary

Problems solved by technology

[0005] The second vision-based motion analysis technology can be called the optical method, which is similar to the above method, except that when the front-end data is acquired, a high-speed camera is used, and luminescent marks are attached to various parts of the human body, and the high-speed camera records the human body at a higher frequency. Movement, and then process and calculate the spatial position of the luminous point on the human body in the video image on the computer, so as to purely restore the movement process of the human body, but this method does not involve the estimation of the human body posture

[0007] Problems of the prior art: (ordinary camera) 1. The prior art mainly uses the camera to collect the motion information of the human target, which determines that it can only estimate the posture of the moving human target within the observation range of the camera. The existing technology can only be limited to the laboratory or room, and the scope of use is limited

2. The existing technology uses digital image processing technology to extract human body motion information from image frames on the basis of video sequences collected by cameras. Due to the problem of the shooting angle of the camera, certain human body images will inevitably occur in the collected human body images. Overlapping and occlusion of some parts, which increases the difficulty of extracting human motion information

Whether it is an optical type or an inertial sensor type, the existing technology mainly focuses on reconstructing the motion state of the human body, but in terms of attitude estimation, most of the current research results still have the characteristics of poor accuracy and real-time performance, and most of the The method requires manual intervention and the assistance of some hardware devices, and there is still a certain gap between the goal of accurate tracking and attitude estimation under simple conditions.

Images