Analysis of Object Detection Technique Using Particle Swarm Optimization

Abstract

A key challenge in a surveillance system is the object detection task. Object detection in general is a non-trivial problem. From many years, among the object detection problems, the researchers have mainly focused on the problem of face detection in the field of image processing. Object detection is necessary, for guidance of autonomous vehicles, efficient video compression, for smart tracking of moving objects, for automatic target recognition (ATR) systems and for many other applications. Image matching is very important technique for wide range of applications, such as in guidance, navigation, robot vision, automatic surveillance, and in mapping sciences. Numerous techniques have been proposed for object detection. Most biological vision systems have the talent to cope with changing world. Cross-correlation and related techniques have dominated the field since the early fifties. Conventional template matching algorithm based on cross-correlation requires complex calculation and large time for object detection, which makes difficult to use them in real time applications. In this thesis a different algorithmic approach having origin of biological principles is applied to detect the position of the selected object (part of an image). Applied algorithm performs better to detect the exact position of object when numbers of iteration are fixed but population size is limitedly increased.