Research On License Plate Recognition Based On HSV Space

License Plate Recognition System(License Plate Recognition System, LPRS) is a license plate information system, which is to detect the controlled road vehicles and extract the unique ?identity? on the behalf of vehicle. It is one of the indispensable part of modern intelligent transportation system, and based on the technology of digital image processing. Computer vision and pattern recognition. system is widely used in security monitoring, traffic investigation, electronic police without parking fees, and the closed area vehicle practices(organs, factories, residential area, parking) and so on. Due to the technology can make intelligent transportation system have been considerable development, so in order to maintain the traffic safety and city public security, prevent traffic congestion, and achieve the traffic automation management, the in-depth research of license plate recognition system can realize the double value of theory and economic.License plate recognition system generally consists of three parts: license plate location, character segmentation and character recognition. This paper aims at the the three component analysis and puting forward the improved algorithm in the key links. The experiment was carried out to achieve the desired results and further completed the license plate recognition system. In this paper, the main work is as follows:1. Research of license plate location algorithm. On the basis of the existing license plate location algorithm, and aiming at the difficult problems caused by the influence of various complex factors, this paper presents a general license plate location algorithm, which combines the color feature, the gray feature and the geometric shape of the license plate. First of all, the algorithm in the HSV color space uses the license plate color to preliminary positioning and improved Sobel edge detection operator of secondary location achieve plate texture feature extraction in multiple directions. Then using Otsu method thresholding and rejecting pseudo plate from binary image with mathematical morphology combined license plate features, the algorithm filters out qualified license plates.2. Research of license plate character segmentation algorithm. From the division method of projection and horizontal sampling line two character, and improving the instability of projection analysis and the threading method, this paper presents a character segmentation algorithm,which is based on vertical projection and template matching. In order to obtain the license plate characters, first, the algorithm requires threshold of the candidate plate in the positioning stage then the algorithm corrects Corrects inclined plate with linear fitting algorithm and algorithm removes the license plate frame finally, the algorithm get license plate charater through segments. The results show that in a complex environment, this algorithm has better adaptability and robustness.3. Research on license plate character recognition algorithm. throughout the license plate recognition system, the character recognition is the last link, but can reflect the critical success of the entire system. The basic principles of traditional statistical pattern recognition are empirical risk minimization. Its drawback is that it can not guarantee effectively recognize when the sample size is not large, coupled with the license plate character of the sample is limited, so the use of traditional approach can not reach the desired results. In spite of the small sample size, the support vector machine is still able to work well. The goal is that regardless of sample size more and less, it can get the optimal solution. Therefore, after the detailed research on support vector machine model, and comprehensive analysis of the pros and cons of various identification methods, this paper use SVM multiple classification method to recognize license plate.4. The realization of license plate recognition system. In VS2013 platform, combined with Open CV, and applying the improved algorithm in all above aspects of the article to finish the system. After testing, the license plate recognition performance can better meet the needs of users and have certain value.