The Research And Implementation Of License Plate Recognition System Is The Key Technology

With the rapid development of image processing, computer vision, communications and other high-tech, Automatic License Plate Recognition Technology LPR(Licens Plate Recognition) has been very widely used in traffic detection.The in-depth study of key technology for the entire system, by contrast, slightly superior to summarize the various methods of existing technical solutions, and ultimately the recognition of license plate. In this paper, automatic license plate recognition system is divided into the following modules:image preprocessing, license plate location, character segmentation, character recognition.In the license plate image preprocessing, inroduced image pre-processing technology, including:graying of the color image, image enhancement, edge detection and image binarization.In license plate location, the license plate location method in-depth study, the characteristics of the vehicle license and the introduction of common license plate location method, license plate location method based on a combination of edge detection and image projection to locate the vehicle license, make up for the shortcomings of the two methods when used alone, the eventual adoption of the advantages of these two methods combined to achieve the precise positioning of the license plate.In plate character segmentation, were introduced the segmentation algorithm of the traditional character of the vertical projection method and template matching method, analyzed in the two methods are slightly superior, proposed combination of character segmentation method based on vertical projection-template matching. Segmented character correction, removal of border, remove the rivets and characters normalization, a series of processing, fully prepared for character recognition.In the license plate character recognition, the first overview of the development of license plate character recognition described, were analyzed the characteristics and principles of the character recognition. Analysis of the character recognition method, including: character recognition method based on template matching character recognition method, based on the structural model, based on neural network character recognition methods, conduct researched character feature extraction methods, and proposed a classification network characteristics of characters to extract and identify, the entire network is divided into subnets of Chinese characters, numbers, subnet and letter, the purpose of this classification for different characters use different neural network, so that can reduce the complexity of the entire process, and improve the speed of character recognition, which can significantly improve the efficiency of character recognition. This article focuses on the character recognition method based on BP neural network, and analyze its learning and training process, and using this method to achieve license plate character recognition.