In recent years, the amount of vehicle in Indonesia has been increasing rapidly. This surely, if it is conducted conventionally, challenges the upholder in recognizing and detecting the lawbreakers vehicle. The objective of this research aims to create the system which can automatically recognize vehicles number plates. This is also expected to be able to assist the upholder to take an action against the lawbreaker. The method used are sliding concentric windows and connected component for detecting and segmenting each of character on vehicles number plates. Further, multi-layer perceptron neural network classification model is used to identify each of character on it.The system has been tested using variety of vehicles number plate images and succesfully recognize 180 of 224 characters images (80.35%). Based on the computation of each character, the accuracy of the system, throughout tested vehicles number plate images, can reach 95.69% (1509 of 1577 characters can be identified).The tested system has shown prospective results, thus the technique used on this research can be implemented through vehicles number plate recognition system in Indonesia.
Copyrights © 2015