A Pedestrian Fall Recognition Method Based on Skeleton Detection

Abstract

A pedestrian fall recognition method based on skeleton detection, comprising the following steps: S1, using a camera to obtain an image of a monitoring area; S2, segmenting the image to obtain an image of a pedestrian human body area, and detecting the distribution information of pedestrian skeleton feature points; S3, analyzing pedestrian Skeleton feature point distribution information, extract the joint point coordinates of the key parts of the pedestrian's human body, and obtain the spatial position characteristics of the joint point of the pedestrian's human body and the geometric quantity of the pedestrian's posture according to the joint point coordinates. The key parts are the preset parts of the human body; Point spatial position features and pedestrian posture geometric quantities to establish a fall detection model; S5, use the fall detection model to judge the pedestrian posture, the pedestrian posture includes normal walking and falling, and realize the detection and recognition of pedestrian posture according to the judgment results. The invention can actively detect abnormal situations such as pedestrians falling in the monitoring video, and can be combined with an early warning system to improve the monitoring ability of pedestrian safety events.

Description

technical field [0001] The invention relates to the field of machine vision, in particular to a pedestrian fall recognition method based on skeleton detection through multi-feature comprehensive judgment. Background technique [0002] In recent years, the problem of population aging in my country has become increasingly serious. The number of elderly people is increasing year by year, and falls have always been one of the major problems that plague the elderly. The incidence of falls in the elderly is high, and the problem is often serious. This has become a serious medical and social problem in today's era. Through scientific and effective methods to detect the fall of the elderly, monitor the video to locate the location of the fall so as to reduce the time for the elderly to receive treatment. Use the characteristics of falling behavior to distinguish it from common behaviors in life. However, the current research on the fall behavior of the elderly cannot give an accu...

AI Technical Summary

Problems solved by technology

The fall detection technology based on wearable sensors has high accuracy, but the cost is too high and it is uncomfortable to wear

The video-based fall detection technology uses the motion features of the human body contour extracted by one or more cameras for recognition, but it is easily affected by the light intensity and viewing angle, so the recognition rate is very low

Images