The Research And Design Of System For Vehicle License-plate Recognition

The Vehicle License Plate Recognition System (LPR) is one of the basic contents of the Intelligent Transport System(ITS). It not only finds wide application in ITS, but also is a hot point of research in computer vision, image processing and pattern recognition. On the above background, we made a deep and systematic research for LPR and designed a complete system for it, that includes five modules such as image acquisition, vehicle detection, license-plate location, character segment and character recognition.As the first module, the function of image acquisition is to acquire the image sequences. Considering the actual condition, we have tried a lot to choose proper equipments and presented some effect measures to buildup the dynamic ranges and eliminate the movement blurs. The method of vehicle detection is basing on changed area. We settled the difficulties such as the building and updating of the background and the segment of object, and also reduced the data and wiped off the useless background's interference by setting detecting area. For the module of license-plate location, we used the information about shapes and texture of the plate well and proposed an algorithm basing on mathematical morphology. The methods of Character Segment and Character Recognition are basing on projecting image and BP network respectively, which are both classical methods. We also did some improvements to these methods.Testing 541 frame static images and 2 segment videos including 80 vehicles, we can achieve the accuracy of 89.2% for the static images'recognition, 96.3% and 92.2% for the videos'vehicle detection and license-plate recognition respectively. The results show that, these methods are reasonable and the system has large practical values.