Object Tracking Using Image Segmentation

Abstract

Tracking objects in real time is an increasingly common application in computer vision. In this work, the author analyses different methods for tracking an object using conventional image processing algorithms. Different image segmentation methods were developed and their efficacies in detecting objects were analysed on sample images. The image segmentation method selected was subsequently used in video analysis for tracking a particular object across frames. To improve the detection processing, a windowing technique was used to reduce the segmentation computation time. The method developed was able to detect and track objects in real time without significant loss of frame rate.