Pengenalan Karakter pada Plat Nomor Kendaraan dengan Metode Backpropagation

Abstract

Character recognition is a way to get information from an image that contains characters such as letters and numbers. Thus in need of a process to obtain information from a pattern. Problem in this study is how to recognize the characters on defective license plates. In this study we assume the process of capturing image as well as the determination of the position of the plate has been done before. In this thesis described how the recognition of the characters on the plate of the vehicle is done by using back propagation neural network, where the will go through the stages of image processing to obtain the input data, segmentation and character recognition stage. Test results demonstrated network architecture that provides the most optimal results are 112 neurons in the hidden layer, learning rate 0.1, without momentum. The results of memorization on the network architecture is 97.77%. Character recognition results on vehicle license plate is 91.78%.