A Systematic Algorithm for Moving Object Detection with Application in Real-Time Surveillance

Abstract

Moving object detection and tracking from video sequences are a relevant research field since it can be used in many applications. While detection allows to return object shapes discovered in the image,tracking aims to individually identify and estimate individual trajectories of detected objects over time. Hence, detection can have a crucial impact on the overall tracking process. This paper focuses on detection. Currently, one of the leading detection algorithms includes frame difference method (FD), background subtraction method (BS), and optical flow method. Here, we present a detection algorithm based on the first two approaches since it is very adequate for fast real-time treatments, whereas optical flow has higher computation cost due to a dense estimation. A combination of FD and BS with Laplace filters and edge detectors is a way to achieve sparse detection fast. Thus, a main proposed contribution is the achievement of a systematic detection algorithm for moving target detection with a more elaborated combination of basic procedures used in real-time surveillance. Experimental results show that the proposed method has higher detection accuracy and better noise suppression than the current methods for standard benchmark datasets.