Motion Detection and Tracking of Classified Objects with Intelligent Systems

AbstractIn this paper, detecting and tracking of a car is aimed using a stationary camera system. Background subtraction is used to detect motion and Kalman filter is used for tracking of a moving car. Classification of the car is accomplished utilizing Support Vector Machines (SVMs) and Artificial Neural Networks (ANNs). In using SVMs and ANNs, some features should be extracted from Region of Interest (RoI). Before the extraction, image enhancement methods are used and then by using Discrete Wavelet Transform (DWT), the features are represented in different frequency scales. The current work compares the performances of ANNs trained via Levenberg-Marquardt optimization technique, Least Squares Support Vector Classification (LS-SVC) and v Support Vector Classification (v - SVC).