Complex Environment Of The License Plate Located In The Character Segmentation Method To Achieve

With the rapid development of our country's road transportation, the intelligent traffic management system becomes a main problem and has been paid attention. As a part of the intelligent traffic management system, License Plate Recognition System plays an important role in automatic charging system at entrance, automatic parking lot monitor system and automatic record of peccancy system.Being a special computer vision system in the real-time case, the LPR (License Plate Recognition) system mainly includes the subsystem of license plate detection and character recognition. The LPR system covers many fields, such as Pattern Recognition and Artificial Intelligence, Computer Vision, Digital Image Processing etc. The detection of license plates is the key of LPR system. Because of the complex of image background, the uncertainty of plate position and image quality, the location of plates is not satisfied. Therefore, the study on the algorithm of license plate location is always focused.This paper has studied on license plate localization and character segmentation in LPR system. Based on the discussion of current study, the following four works have been mainly done in this paper:1. We proposed an effective preprocessing method for car image. The method firstly uses color edge detector to get the horizontal edge image, secondly it makes use of the global threshold method binarilize the grad image, and finally dilates the binary grad image from horizontal direction. After preprocessing, the area of license plate has become outstanding.2. At the license plate location phase, we propose a new rough localization method which based on connected area horizontal clustering. The method sufficiently considers the horizon related feature of characters in grad image, which gets a better recall ratio in the experiment. Moreover, this paper uses the grad shape and color feature of license plate to carry out precise location, which has approved localization result.3. At the character segmentation phase, a new approach for license plate character segmentation based on multiscale template matching is presented. The method takes into account of the structure feature of our country's license plate, which has achieved an outstanding veracity and robust, and acquire a satisfying character segmentation result. Besides that, this paper uses a method based on biggest connected area mean borderline to remove the noise in character images.4. At the experiment phase, a large number of image and video files which captured from different environment have been tested repeatedly, and gave the result of different testing case, and analysed the influence of license plate's different character to the location result.Experiments show that the license plate location and character segmentation algorithm in this paper has a less limit in background and illumination; furthermore license plate location and segmentation are accurate and rapid. So it meets the requirements of actual system.