Falling Detection System based on Webcam with Raspberry Pi B+

Chiam, Evon Sock Theng (2016) Falling Detection System based on Webcam with Raspberry Pi B+. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Evon Chiam Sock Theng Full Text.pdf Restricted to Registered users only Download (3MB)

Abstract

Anyone both young and old can have a fall anytime and anywhere. The risk of falling increases with age which means that older people are of higher probability to fall. A fall from old people can cause grave injuries if they are not given the medical attention needed. In today’s society, older people are left at home alone while the younger generation will be out for work or studies leaving no one to watch over the older people and help them during the event of a fall. In this project, a robot prototype with falling detection system is proposed to help to overcome this problem. The falling detection in this project will be based on image processing from the image acquired from the webcams using Raspberry Pi B+. Identification and tracking of the human is based on colour feature matching where the human have to wear a blue colour shirt to allow tracking. The servo motor that has the webcam mounted on it will rotate accordingly to ensure that the image tracked will be at the center. Based on the tracking using colour feature, the contour of the human detected will be determined which is then used to extract the angle of rotation of the human tracked. Once the angle is above the allowed threshold indicating a fall could have occur, the program will check if the size and area in the second webcam that has a view closer to the ground is the same as the size and area detected in the first webcam that performs tracking. In the event of fall, an email alert will be sent out using the Python library smptplib to send email using SMTP server. Gmail SMTP server will be used to send out email and a buzzer will sound when fall is detected.