Research On Vehicle License Plate Location And Recognition

Intelligent traffic management is the main direction of traffic management at present and in future. In traffic management, Each motor vehicles have a unique license number that was used to identify a vehicle. Therefore, the license plate recognition system (LPRS) has naturally become an important part of intelligent transportation, and has a wide range of application and development in intelligent traffic management in all countries. Based on the previous study,this paper will study license plate location, character segmentation and character recognition.By analyzing, researching and concluding the problems above, this thesis has finished the following researches.(1) License plate location algorithm. In this paper, the license plate location method is based on the RGB color model and wavelet transform. The license plate is divided into the blue license plates of non-blue vehicles, the yellow license plates of non-yellow vehicles and the others,then, using different methods to locate. Using Three threshold extract the license plate of the blue license plates of non-blue vehicles and the yellow license plates of non-yellow vehicles,then, selecting the most optimal result in these three results. For other license plates, we use the image grayscale conversion, adaptive binarization, median filtering, wavelet transform means based on the license plate texture features to extract the candidate license plate regions. Finally, we use morphological operations and connected component labeling, and connected component analysis and other methods based on China’s license plate with the characteristics of positioning plate region.(2) Character segmentation algorithm. This paper gives a new license plate tilt correction method based on connected component analysis and a new character segmentation method based on Template matching. First, the license plate to have located was pretreated.then the license plate to have located was changed from sloping to horizontal and vertical by connected component analysis and the vertical projection,and finally completing the character segmentation by template matching method.(3) Character recognition algorithm. This paper gives a fuzzy sorting method of characters based on mathematical morphology. This classification result is determined by analyzing the image to be recognized,which has been pretreated with basic operations of mathematical morphology.Finally, the character recognition can be done with a modified template matching method. This paper presents a new method of license plate locationlocation, character segmentation and character recognition. Matlab Simulation results indicate that the algorithms in this paper can gain high probability of correct classification and low compute quantity,and can meet the practical demands in certain circumstances. Above, the proposed method has some practical value and research value in the intelligent traffic management.