Human body fall detection method based on three-dimensional camera

Abstract

The invention discloses a human body fall detection method based on a three-dimensional camera. The human body fall detection method comprises the steps of: acquiring training data, wherein the acquisition is implemented by acquiring a plurality of groups of falling and non-falling action segments of a plurality of experimental subjects of different heights and different ages; preprocessing data; extracting action features, wherein the extraction is implemented by extracting bone node distribution features, height features and inertia features of each kind of action, and forming behavior feature vectors; training a classifier for acquiring a motion model, wherein the training is implemented by inputting the behavior feature vectors of a training set into a hidden Markov model classifier, so as to obtain a parameter optimal model; and finally, carrying out feature extraction on data in successive frames of each test sample, inputting the processed data into the hidden Markov model for testing, and making judgment. The human body fall detection method realizes human body falling behavior detection by adopting a machine learning method, and has high identification rate. The human body fall detection method belongs to the field of computer vision and machine learning, can effectively identify the occurrence of indoor falling behaviors, sends out an alarm in time, and saves the labor force of nursing staff.

Description

technical field [0001] The invention relates to the research on human body behavior recognition in the field of human-computer interaction, and adopts a machine learning method to propose a human body fall detection method based on a three-dimensional camera. Background technique [0002] In recent years, the analysis of human behavior has received widespread attention in the fields of medical monitoring, intelligent transportation, and security monitoring. important. Data show that falls are the main cause of injury for the elderly, and advanced image processing technology and sensor technology can automatically detect the falls of the elderly and patients and give timely alarms. It avoids the aggravation of the injury after the fall, the induction of the disease, and the missed rescue opportunity; at the same time, the cost of nursing care is reduced and the labor force is saved. [0003] Human fall detection research can be roughly divided into two categories, one is ba...

AI Technical Summary

Problems solved by technology

This type of method utilizes the bone node features captured by the 3D camera, human body contour features, and human inertial features. The discrepancy with the actual position leads to misjudgment, and the recognition rate is not high

Images