DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING
NEURAL NETWORK BASED VEHICLE LICENSE PLATE RECOGNITION SYSTEM
III YEAR CSE
III YEAR CSE
This paper presents a neural network based system that is able to analyse the image of a car given by a camera to locate the registration plate, recognise and validate the registration number of the car. A novel image segmentation technique called Sliding Concentric Windows (SCW) is discussed. In the process of license plate recognition, the Region of Interest is detected by SCW segmentation, image masking, binarisation and connected component labeling arranged in sequence. Then, the detected region is segmented and forwarded to a Neural Network based Optical Character Recognition (OCR) engine. A three-layer Probabilistic Neural Network (PNN) of topology 108-180-36 with supervised training is used for the task of character recognition. After recognition, the characters are fed into a character validation module that determines whether the character is sufficiently valid based on the validation threshold provided as a parameter. The limitations such as image blurs, insufficient lighting conditions and arbitrary size of license plates and the scope of the system are also discussed.
Automatic vehicle license plate recognition systems are mass surveillance methods that use optical character recognition (OCR) on images to read the license plates on vehicles. License plate remains the principal vehicle identifier despite the fact that it can be deliberately altered in fraud situations or replaced with a counterfeit one (e.g., with a stolen plate). Recently, systems can scan number plates at around one per second on cars travelling up to 160 km/h. They use existing closed-circuit television (CCTV) or road-rule enforcement cameras, or ones specifically designed for the task. Systems commonly use infrared lighting to allow the camera to take the picture at any time of day. A powerful flash is included in at least one version of the intersection-monitoring cameras, serving to both illuminate the picture and make the offender aware of his mistake. This technology tends to be region specific, owing to plate variation from place to place. This paper discusses a computer based intelligent character recognition system for license plate recognition and validation. This is meant to be used as a core for intelligent infrastructure like electronic payment systems (toll payment, parking fee payment), highway traffic surveillance. Moreover, as increased security awareness has made the need for vehicle based authentication technologies extremely significant, the proposed system may be employed as access control system for monitoring of unauthorized and unregistered vehicles.
DESCRIPTION OF THE SYSTEM:
The system is composed of a camera, a frame grabber, a general purpose computing device and software for image analysis and character recognition. The system is triggered by an external signal, coming from a suitably positioned infra-red barrier (as shown in Fig.1.1) or other sensors, it acquires and stores the image of the car (which is presumably in front of the camera) and analyses the image with the purpose of finding and recognising the car number plate. The focus of this paper is on the integration of a novel segmentation technique implemented in a LPR system that is able to cope with outdoor conditions if parameterized properly. Specifically, the contributions are: • A novel segmentation technique used for faster detection of Regions, which is used in the segmentation and processing of the license plate.
Fig.1.1 showing the position of the camera, the car and the sensor
SCHEMATIC BLOCK DIAGRAM OF THE ENTIRE PROCESS:
This paper comprises of the following processing...