Vehicle Recognition System Using Singular Value Decomposition (SVD) and Levenberg-Marquardt
Zuraidi Saad, Muhammad Khusairi Osman, Zuli Imran Zulkafli, Sopiah Ishak · 2009
The purpose of this research is to develop a system that used to recognize image of vehicle and classified it into their classes using image processing method and artificial neural network. In the research, all the selected images are required to go through image processing technique to obtained desired data. Images are converted into data using singular value decomposition extraction method and the data then are used as the input for the training purposes. The multilayered perceptron network trained by Levenberg-Marquardt algorithm was chosen in recognition and classification stage. The input variables were taken from 3 sets images of motorcycle, bus and lorry. The data inputs consist of 215 data. For training data set is 96 sets of data and used in training process and the other used in testing process. This training method can recognize the vehicle type in images successfully.