Performance Evaluation of Optical Motion-Based Object Detection and Tracking Using Background Subtraction ın High-Resolution Video

Abstract

The technique for automatic detection and tracking of an object is an emerging area of research for video surveillance and traffic surveillance. For background subtraction, foreground detection and optical flow algorithm for object detection and tracking, the proposed technique uses a Gaussian mixture model (GMM). The initial step in this technique is to differentiate the foreground object i.e. the region or object of interest from background image which is obtained by GMM. For detection of objects in motion does not need a stored database, only it requires successive frames. The most widely used context subtraction is the simplest method for detecting moving objects from a video string. Now day's high-resolution imaging sensors/cameras are being used in areas of video surveillance like security at public places, vehicle traffic monitoring systems, military applications, and satellite imaging systems. Evaluating the quality of the object tracking system in high-resolution video for the simple tracking system is important. The proposed system is assessed as to the execution time required for a single frame. We performed a performance analysis of the proposed technique for different resolution video in this paper. The technique proposed is tested using i5 core CPU 1.80 GHz processor and 4 GB RAM