Real-Time Object Detection and Tracking Using Velocity Control

Abstract

Real-time object detection and tracking is a vast, vibrant yet inconclusive area of computer vision. Automatic object detection and tracking are useful in surveillance, tracking systems used in security, mobile robots, medical therapy, driver assistance systems, and analysis of sports. Algorithms proposed in existing literature use color segmentation, edge tracking, shape detection for detection, and tracking of an object. The challenges such as tracking in dynamic environment and difficult tracking of multiple objects in multiple-camera environment and expensive computation restrict the implementation of these systems for solving real-world problems. This motivates us to develop a system that is efficient in real-time object detection and tracking. In this paper, authors develop the real-time object detection and tracking system using velocity control. Experimental results prove its efficacy in detection and tracking of simple as well as complex objects in both simple and complex backgrounds. The system is effective in detecting and tracking the co-occurrence of two objects. It clearly shows the impact of color dominance or shape dominance, self-shadow, and image of an object in a mirror.