Authors: Chun-Ling Liu; Chun-Bao Huo
Addresses: College of Electrical Engineering, Liaoning University of Technology, Jinzhou, Liaoning, China ' College of Electrical Engineering, Liaoning University of Technology, Jinzhou, Liaoning, China
Abstract: In order to overcome the problems of low recognition accuracy and high-recall rate of traditional methods, an intelligent wristband human abnormal behaviour recognition method based on machine vision is proposed. Firstly, the human behaviour video in the smart wristband is obtained through intensive sampling, and the image feature descriptor is obtained according to regularisation processing; the human abnormal behaviour features are extracted by DT algorithm; secondly, softmax classifier is introduced to classify the extracted human abnormal line features of intelligent wristband; the dense trajectory algorithm is used to extract the abnormal behaviour characteristics of intelligent wristband human body, and the machine vision is introduced to recognise the abnormal behaviour of intelligent wristband human body. The experimental results show that the recognition accuracy of this method is as high as 98.2%, and the recognition recall rate is only 4%, which shows that this method can effectively improve the recognition effect.
Keywords: smartband; machine vision; softmax classifier; abnormal behaviour; weighted processing method.
International Journal of Product Development, 2022 Vol.26 No.1/2/3/4, pp.254 - 267
Received: 30 Dec 2021
Accepted: 15 Mar 2022
Published online: 07 Sep 2022 *