Research Of License Plate Recognition Algorithm Based On Phase Congruency

License plate recognition (LPR) is the technology of license plate image acquired from road monitoring system to be detected and output license plate information automatically. LPR play an important role in the promotion of the current system of things. Because it gives the life of the car as the main means of transport great convenience, so the technology has been widely used, such as toll highways, public parking and when detection of cases. Widely used to make the LPR has become a hot topic in current science.The LPR is put forward in 1988, people have done wide study about it. But in the natural environment, due to the change of illumination angle and insufficient light make the license plate image quality to become poor.To the poor image, LPR obtain low accuracy. According to these problems, the author puts forward the license plate recognition method based on phase congruency. This method uses phase congruency to detecte the image edges. Due to the characteristics of the phase is not affected by contrast and illumination change, so the edge of the detected is clear and complete. On the edge of the phase plate texture features positioning license plate application, then using test phase consistency of license plate image segmentation and recognition. The method improves the license plate positioning accuracy under insufficient light and illumination change environment.This paper introduces the common license plate location technologies, and present license plate location algorithm based on phase consistency. Through the comparison of the experimental simulation, demonstrate the superiority of the algorithm in this paper. Based on the phase consistency of the license plate complete characters segmentation. By comparing several common character segmentation methods, present combining vertical projection with connected domain character segmentation algorithm. The algorithm improves the character segmentation accuracy and speed. In the split phase consistency matrix of a single character, application with wavelet analysis and support vector machine (SVM) complete the first character recognition. For confusing characters, characters were detected using Harris corner, according to the characters of the angle point location to realize the whole process of the license plate recognition.In this paper using Matlab7.1 simulation experiment, the results show that phase consistency of application improves the license plate recognition accuracy in illumination changes and insufficient light environment.