An indoor fall detection method for old people

Abstract

The invention discloses an indoor fall detection method for old people, and aims to provide a solution for indoor abnormal behavior detection and alarm for old people, and the scheme comprises the following steps of obtaining one frame in video data, and carrying out the preprocessing of noise reduction, contrast enhancement and the like; calculating image direction gradient histogram features, and using an SVM classifier to realize human body detection so as to determine whether a human body is contained or not; tracking a human body area by using a target tracking algorithm under the condition that the image contains the human body; representing a human body feature by tracking an inscribed ellipse in a rectangular area, wherein a central point and a lower vertex of the inscribed ellipserespectively represent the center and feet of a human body trunk, and calculating the acceleration, angle, height difference and residence time of the two feature points to judge whether an old person falls down or not. Compared with a traditional recognition method based on an acceleration sensor or a bone sensor Kinect, the method has the advantages that the installation cost is lower, and therecognition accuracy is higher compared with a traditional judgment method based on the human body contour length-width ratio.

Description

technical field [0001] The invention relates to the technical field of video analysis and recognition, in particular to a fall detection method for an indoor elderly person, and further relates to gesture detection under indoor ambient light. Background technique [0002] In today's population aging society, the number of empty nesters is increasing day by day, and their health security is facing serious threats. Accidental falls are extremely harmful to the elderly. If the elderly living alone can receive timely medical assistance after a fall, the risk of accidental casualties can be effectively reduced. Statistics show that: 1 / 3 of the elderly over the age of 65 have fallen at least once in a year. Therefore, it is of great practical significance to automatically detect and send an alarm message when the accidental fall of the elderly living alone occurs. [0003] At present, human fall detection at home and abroad is mainly divided into three categories: (1) Fall detec...

AI Technical Summary

Problems solved by technology

Due to the lack of overall information about human body movements, the system based on wearable sensors has a high false alarm rate; (3), the environment-based fall detection system models human body movements through the skeleton data provided by the Kinect sensor, and uses The spatial motion trajectory curve and motion curve are used as the expression of the human body action, and then the discrete Fréchet distance is used to measure the similarity of the action, and finally the fall action is recognized by the K nearest neighbor (KNN) classifier, but the The method is expensive to install and the Kinect sensor is expensive

Images