Method and system for automatically identifying characteristic points in human body chain structure.

Abstract

The invention relates to a method and a system for automatically identifying characteristic points in a human body chain structure. The method comprises the following steps of: 1, reading a human body chain structure model of an observed object, and confirming rigid bodies in the model and the characteristic points of each rigid body; 2, acquiring frames of a motion image of the observed object and a characteristic point set of each frame through motion; 3, selecting a first frame of which the number of the characteristic points is more than or equal to that in the human body chain structure model as a first frame; and 4, enumerating a characteristic point assembly in the characteristic points of the first frame for each rigid body, selecting the characteristic point assembly which meets distance constraint conditions and corresponds to the rigid body as an enumeration result, selecting a characteristic point assembly from the enumeration result as a matched pair of the rigid body, and identifying characteristic points of the matched pair corresponding to the characteristic points in the rigid body. Due to the method and the system, the first frame can be determined and the characteristic points in the first frame can be labeled from a motion data sequence with lost or false characteristic points.

Description

technical field [0001] The invention relates to the fields of computer vision and pattern recognition, in particular to a method and system for automatically recognizing feature points in a chain structure of a human body. Background technique [0002] The chain structure model of the human body embodies the shape, size and connection relationship of each part of the human body, and it contains the prior knowledge of the physiological structure of the human body. In the human motion analysis with feature points, to describe the motion of the object, it is necessary to track the feature points of the moving object, calculate the position of the joint center, and drive the bone model. For different moving individuals, it is necessary to specify the names and positions of the reconstructed feature points according to the topological structure of the human body model, and establish a human chain model that meets the requirements, so as to continue to complete the tracking task a...

AI Technical Summary

Problems solved by technology

However, due to the complexity of human motion, in the process of motion reconstruction, there will inevitably be reconstruction error points, which are called noise points; in addition, due to the influence of occlusion and self-occlusion, there will be loss of marker points. situation, called lost point

It is undoubtedly a very tedious task to manually select the appropriate first frame and label the names of the feature points one by one, and if there is a wrong label, the accuracy and reliability of tracking will be significantly reduced.

Images