License Plate Recognition System Related Algorithm Research And Improvement

As the core of the intelligent transportation system, license plate recognition system has become an important issue in the field of the traffic system presently, which also has a broad prospect of application.License Plate Recognition System aims to use the computer to complete the whole process of the License Plate Recognition System. Vehicle Plate Recognition System is composed by license plate image pretreatment, license plate localization, license plate character segmentation, license plate character recognition. After reading many of Chinese and foreign documents, the author mainly do research on the three key technologies of license plate localization and character segment, and put forward the corresponding improvement algorithm, and to achieve it through the experiment on the algorithm simulation.(1) Coarse position of vehicle license plate. it find if the background of picture was complex that the current algorithm can’t locat well, it always made mistakes. In order to solve this problem, a regional growth location algorithm based on gray value is put forward. Firstly, according to pixel grayscale value, let image to execute regional growth, and find license plate candidate area preliminary. And then make comprehensive use of prior knowledge license plate to find out its coarse position; Secondly, look for rectangular sequence which can meet the requirements and calculate the horizontal deflection Angle of the license plate according to its upper left vertex coordinates of its inscribed rectangular, then perform horizontal correction in order to realize horizontal accurate location; Lastly, on the basis of external rectangle and internal parallelogram complete vertical correction.(2) Accurate license plate location. This paper adopt iteration of the least squares fitting in the slope of the vertex straight line in all rectangular upper left tops of rectangular sequence, and based on this slope to calculate the level of deflection of the license plates, and finally realizing horizontal positioning; According to the characteristics of the license plate, the author puts forward a new method of the license plate vertical correction. Firstly, it calculates the vertical deflection angle of the license plate according to the relationship of external rectangle and internal parallelograms which are formed by the widest positions of the up and down half part of the character, and then complete the vertical correction through pixel mobile.(3) character segmentation. Based on study of the existing license plate character segmentation algorithm and combined with the features of license plate characters, this thesis puts forward a kind of new character segmentation algorithm which is based on new L-operator to realize edge detection, at the same time, combine with the license plate ranks of the pixel gray value statistics to achieve license plate character segmentation. The idea of the algorithm is:firstly, uses mathematical morphology to highlight the license plate character area, and then use the Canny operator to complete edge extraction, and highlight the single character area through the ranks of the statistics license plate pixel gray value, and finally combine with priori knowledge the license plate to split out individual characters.(4) character recognition. according to the characteristics of license plate character, this paper use the improved Hausdorff distance template matching algorithm to realize license plate character recognition. The algorithm uses improved Hausdorff distance to calculations suitability of character to identify template. in order to further improve recognition rate, it excute the second segment to the similar characters.Experimental results show that this license plate recognition system have strong anti-jamming ability to bad quality pictures. For example, picture which is affected by light, or deformation, or the handwriting is damaged and so on. Vehicle Plate Recognition System based on this paper’s algorithm can position plate and split the plate characters accurately, and can satisfy the real-time requirement.