Vehicle Recognition System Using Singular Value Decomposition and Extreme Learning Machine

The purpose of this research is to develop a system that is able to recognize and classify a variety of vehicles using image processing and artificial neural network. In order to perform the recognition, first, all the images containing the vehicles are required to go through several images processing technique such as thresholding, histogram equalization and edge detection before obtaining the desired dataset for classification process. Then, the vehicle images are converted into data using singular value decomposition (SVD) extraction method and the data are used as an input for training process in the classification phase. A Single Layer Feedforward (SLFN) network trained by Extreme Learning Machine (ELM) algorithm is chosen to perform the recognition and classification. The network is evaluated in terms of classification accuracy, training time and optimum structure of the network. Then, the recognition performance using the ELM training algorithm is compared with the standard Levenberg Marquardt (LM) algorithm.