TY  - JOUR
T1  - Iot Surveillance System Based on Moving Average Subtraction Technique

JO  - Journal of Engineering and Applied Sciences
VL  - 17
IS  - 3
SP  - 44
EP  - 51
PY  - 2022
DA  - 2001/08/19
SN  - 1816-949x
DO  - jeas.2022.44.51
UR  - https://makhillpublications.co/view-article.php?doi=jeas.2022.44.51
KW  - Video surveillance
KW  - IOT
KW  - computer vision
KW  - object tracking
KW  - remote monitoring
KW  - running average
KW  - gaussian kernel
AB  - Most algorithms for computer vision-based surveillance systems involve high computation overload, making them unsuitable for real-time application. Also, some of these algorithms cannot handle abrupt illumination changes in the image without causing estimation errors. This work employs a background subtraction algorithm model, which is robust against illumination variations and less computation intensive. The design consists of a robot surveillance agent and a control center. The robot was equipped with a camera and Raspberry-pi board to continually capture environment scenes. The video frames were pre-processed to enhance image quality and  a moving average algorithm was applied to detect motion. The frames containing motion are stored locally and also transmitted via the internet to a web application where further analysis can take place and necessary action can be taken. The web application was built on the flask framework, using Python language. The control center for monitoring the robot was a personal computer but could be a mobile device too. The implemented algorithm was able to detect small movements and it can continuously track an object's motion. The algorithm run time is less than the Gaussian mixture models’ algorithm. Analysis of video frames to detect motion enables savings in storage and transmission requirements to be made.
ER  -