A pedestrian target tracking method based on human skeleton structured features

Abstract

A pedestrian target tracking method based on human skeleton structured features comprises the following steps: recording a target motion video, performing equalization preprocessing on a target motionvideo image, extracting skeleton coordinate information of a pedestrian, manually selecting and tracking the target pedestrian, establishing a target initial feature template image, and tracking thetarget pedestrian; if the target pedestrian is normal, continuing tracking, and if the target pedestrian is lost, executing the next step; extracting a target structured feature image according to thevideo image before the target pedestrian is lost and the skeleton coordinate information of the target pedestrian, updating the target feature template image, extracting structured feature images ofall pedestrians, matching the structured feature images with the target feature template image one by one, and repositioning the position of the target pedestrian according to a matching result; realizing target tracking continuously. The method can achieve the stable and continuous tracking of the target pedestrian in a complex multi-person scene, and solves the problem that a current tracking algorithm is easy to lose a tracking target under the conditions of multi-person overlapping, shielding and the like, and is difficult to automatically find the tracking target.

Description

technical field [0001] The invention relates to the field of computer vision, in particular to a pedestrian target tracking method based on the structural features of human skeletons. Background technique [0002] With the continuous development of artificial intelligence technology, target tracking has become one of the important research topics in the field of computer vision, widely used in human-computer interaction, military, medical, transportation and other fields, and has far-reaching significance for social development. Target tracking is the process of locating and detecting the state and position of a moving target by using target feature information or background difference frame separation. Trajectory positioning and prediction technology to achieve the purpose of tracking. With the continuous research and development of tracking algorithms, from traditional tracking algorithms such as Meanshift, Kalman filter, and particle filter used before 2010 to TLD, Struc...

AI Technical Summary

Problems solved by technology

However, the current tracking technology is not stable enough in complex environments such as occlusion and viewing angle changes, and it is prone to problems such as target drift and loss, which makes it difficult to relocate the target, and the continuous performance of target tracking is poor.

Images