Old people indoor fall-down detection method based on computer vision

Abstract

The invention discloses an old people indoor fall-down detection method based on computer vision. According to the method, first, a GMM model is utilized to extract a human body original contour, andhuman body coordinates are positioned by use of an ellipse fitting method; second, an ellipse is divided into different regions, centroids of all the regions are tracked through a closest distance method, and multi-centroid offset vectors are extracted to serve as motion features; third, a K-means clustering algorithm and a TF-IDF algorithm are used to establish a human body motion vision statement; and last, the human body motion vision statement is classified through a DAG-SVM to detect whether old people fall down. The method has the advantages that a fall-down action, a similar fall-down action and a lying action can be distinguished, a fall-down action parallel to a camera principal axis can also be detected, and the fall-down detection rate is high; and the method can be implementedjust through simple hardware equipment and is low in cost and easy to implement.

Description

technical field [0001] The invention relates to the technical field of image vision motion detection methods, in particular to a computer vision-based indoor fall detection method for the elderly. Background technique [0002] With the intensification of population aging, the medical health of the elderly has become an important social medical problem, and falls are one of the main health threats faced by the elderly. Because the falling process is often accompanied by violent impact, it will cause direct damage to the human body, such as fractures, soft tissue damage and brain trauma. At the same time, with the aging of physiological structure and the degeneration of body functions, the elderly often lose their ability to move after falling. If they cannot be found in time and timely rescue is provided to them, the injury caused by the fall will be aggravated. [0003] In the human fall detection method based on computer vision, there are the following difficulties: 1. How...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to analyze the image sequence change information of the monitoring environment, effectively extract the motion characteristics of the human body, solve the existing problems of high false alarm rate and high missed detection rate, and effectively distinguish the falling action from the lying action, so as to monitor the movement of the human body in real time. Activities, to directly and effectively detect the fall of the human body, and propose a computer vision-based indoor fall detection method for the elderly

Images