Research Of License Plate Recognition Technology

The License Plate Recognition System is one important part of the modern Intelligence Traffic Management, which is produced according to the requirement of the modern traffic development. In the view of the technology, the LPR research is integration involved the digital image processing, computer vision, pattern recognition and human intelligence and so on. The research work of the thesis includes the following four parts:1. The detection of the mobile vehicles is one part of the work of the thesis. The previous LPR processes the static images. The thesis aims at one video which includes the vehicles. For the aim of detecting the mobile vehicles quickly, we make use of the DirectShow development backage to grab the image from the video, then we minus the current image from the background image to detect the vehicles. In order to grab quickly, we only use one part of the gray value of the background image, then we can get the mobile vehicles from the video.2. The License Plate Location is one important part of the LPR. The thesis makes use of one new technology based on color, which makes use of the Hue information of the color model of HSV based on the traditional projection method. Then we use the information as the character of the plate and locate the plate correctly. The experimental results show that it is more efficient than the previous methods.3. Character segmentation affects the preciseness of the character recognition directly after LPL. The thesis adopts the projection method, scanning the plate which has been extracted, finding the lower part of the wave, making certain one right key value and the single character is picked out.4. Character Recognition is the kernel part of the whole system, whichincludes the recognition of the Chinese character, letter and digit. For Chinese character, the thesis extracts the characters of every character, and makes use of the detection of the similitude degree to recognize the character. For letter and digit, we make use of template matching technology and BP neural networks technology to finish the recognition. The experimental results show that the template matching works better. The mobile vehicle LPR system is designed and implemented based on the research work of the above. The system can detect most mobile vehicles and locate and segment the license plate satisfactorily. The recognition results show that digit and letter recognition has some space to improve.