Real Time Number Plate Recognition System

Abstract

In this era of the world the technology is growing fast, there is a highly demand through people for a easily  lifestyle and travelling. In these years, the number of vehicles on road is grown as fast as much. With these increases  in the traffic sector every day,so the tracking of the vehicle. This proposal of the project giving an idea of an automated way of tracking the fast moving vehicles at real time with the help of the surveillance cameras on the road. To overcome on this idea, we can use an deep learning framework particularly meant that we can extract the licence plate from an surveillance video image using some computer vision technique and then we can use OCR (Optical Character Recognition) to detect licence plate number. This proposal having four steps of process. Firstly,  convert the video into images and identify the the car from each of the frames. The second step is to  detect the license number  plate from the detected frames. Now final step is that Reading and extracting  the number plate characters from the detected number plates. To make the process of training the deep learning model easier, this system uses some use full modules and  library and CNN algorithm. It provides very  easy way  to use module’s methods  to perform image recognition tasks. These images are taken under different  conditions and  angles. The given proposal method is tested in real time situation  and achieved 97% accuracy for car detection, 98% accuracy for number plate detection and 90% accuracy for character recognition.

Keywords: cctv footage, veichle localization, image recognition, ocr (optical character recognition), CNN, grey scale, edge detection, Thresholding.