Method and system for tracking human skeleton points in two-dimensional video stream

Abstract

The invention discloses a method and system for tracking human skeleton points in a two-dimensional video stream. The method includes acquiring a two-dimensional video stream by a camera, acquiring a foreground image through a foreground extraction module, and acquiring head points and neck points through a face detection module. Coordinates, the system judges whether the head point is on the screen, otherwise proceed to the face detection module; if it is, the human body is divided into left part ROI and right part ROI to detect other key points respectively, and the left shoulder point is obtained through the shoulder point detection module and right shoulder point coordinates; through the hand point detection module to obtain the left-hand point and right-hand point coordinates; through the elbow point detection module to obtain the coordinates of the left elbow point and right elbow point, and finally count the credibility of each point and display the credible points.

Description

technical field [0001] The invention relates to the research field of image processing, in particular to a method and system for tracking human skeleton points in two-dimensional video streams. Background technique [0002] Human-computer interaction technology refers to the technology that realizes effective communication between humans and computers in a way that is convenient for people to use through computer input and output devices. Human skeleton point tracking technology is an important technology in the field of human-computer interaction. It can identify the movement of the human body with the help of infrared rays, and can track multiple parts of the human body in real time without using any external equipment for motion capture. It has broad application prospects in computer-computer interaction environment. The human skeleton tracking technology in the prior art is generally based on the framework of Kinect and PC host system. Kinect is mainly responsible for c...

AI Technical Summary

Problems solved by technology

However, this method is sensitive to the surrounding lighting environment, and poor lighting conditions may affect the accuracy of tracking; occluders such as ornaments on human body parts will reduce some local features of the human body, which will affect the tracking of bones, or even fail to track them. The recognition accuracy is not high, which reduces the efficiency and naturalness of human-computer interaction

Images