Automated Indian Vehicle Number Plate Detection

Abstract

Many countries have standardized vehicle license plates and the constraints of plates like font, size, color, spacing between characters, and number of lines are strictly maintained. Even though standards for number plate are being decided by the Indian government and the process of standardization is in process, it can be seen that every 8 out of 10 vehicles have a variation from that of the standard number plate in terms of either location of plate, fonts used in the plate or various text and designs on the vehicle as well as on plate. Because of these variations, it has been challenging to develop an automated vehicle number plate detection system which can localize number plates or fonts of number plate correctly. This paper presents novel experiments at all the important phases of number plate detection like preprocessing, number plate localization, number plate extraction, segmentation, and character recognition at last. Proposed methodology detects number plate and the characters with high accuracy of 98.75% where neural network has been used for character recognition.