Comparative Evaluation of Object Tracking with Background Subtraction Methods
Dennis Aprilla Christie, Topan Sukma · 2018 Third International Conference on Informatics and Computing (ICIC) · 2018
Object tracking is commonly used in a video surveillance system i.e. a home security cameras, smart city traffic cameras. They tracked motion in certain area of camera field of view. Motion can be detected by separating foreground from its background. Fundamentally, if background models are provided, then a foreground could be detected by subtracting an image with its background. This is called background subtraction, and there are several approaches to models the background. This paper specifically evaluates three of the approaches, namely, frame differencing (FD), running average gaussian (RAG), and eigen-background (EB). Experiments performed using CDNet dataset (see Fig. 2) with different sequences, and performance evaluation conducted using confusion matrix (sensitivity, precession, and F-score). The goal of this paper is to improve understanding about the theories behind, algorithms to implement, behaviour and performances of three of the approaches mentioned.