Bangla automatic number plate recognition system using artificial neural network

Abstract

Bangla automatic number plate recognition (ANPR) system using artificial neural network for number plate inscribing in Bangla is presented in this paper. This system splits into three major parts- number plate detection, plate character segmentation and Bangla character recognition. In number plate detection there arises many problems such as vehicle motion, complex background, distance changes etc., for this reason edge analysis method is applied. As Bangla number plate consists of two words and seven characters, detected number plates are segmented into individual words and characters by using horizontal and vertical projection analysis. After that a robust feature extraction method is employed to extract the information from each Bangla words and characters which is non-sensitive to the rotation, scaling and size variations. Finally character recognition system takes this information as an input to recognize Bangla characters and words. The Bangla character recognition is implemented using multilayer feed-forward network. According to the experimental result, (The abstract needs some exact figures of findings (like success rates of recognition) and how much the performance is better than previous one.) the performance of the proposed system on different vehicle images is better in case of severe image conditions.