The Design And Implementation Of License Plate Recognition Technology Based On FPGA

License plate recognition technology has been widely used in highway vehicle management, parking management, traffic violation management et al. The vehicle license would be extracted from a complex background when the vehicle is in motion, the number of license plate will be identified, recorded and displayed by plate extraction, image preprocessing, feature extraction, license plate number recognition et al. The great problem facing the license plate recognition technology is that the ability to identify declined significantly, and the accuracy of identify become low in low visibility weather or when the weather conditions have great changes, that will affect the normal vehicle management.To solve this problem, this thesis proposes a new algorithm of the dimensionality reduction threshold algorithm based on Otsu nature and guidelines. The algorithm does not need to calculate a lot of information about the two-dimensional image, but reduced the two-dimensional histogram to two one-dimensional histogram according to criteria and properties of the Otsu, then makes quick calculations according to the nature and guidelines of Otsu, get a two-dimensional threshold linear point that is thresholding method required to obtain a threshold line. According to this line, the image of license plate would be turned into binary image. This algorithm has better performance in anti-noise aspect, less time-consuming and faster speed compared to the two dimensional Otsu histogram method and the two-dimensional linear thresholding method simultaneously.Based on the above rationale,design FPGA-based license plate recognition system. Hardware and software system platform to build, design-related module and the bus, license plate recognition algorithm to verify the effectiveness of the proposed upgrade license plate recognition accuracy and processing speed of the entire system.According to the above theoretical basis of image recognition and algorithm analysis and the new algorithm proposed in this thesis, a new FPGA-based license plate recognition system is designed. With Quartus II software platform to build software and hardware systems, design each relevant module and bus about license plate recognition algorithm, design and implement the data exchange by linking the various modules to the Avalon bus.To verify the proposed dimensionality threshold algorithm should be effective which based the guidelinesses and properties of Otsu, and prove the recognition accuracy and processing speed of the entire system is upgraded.The experimental results show that the algorithm proposed in this thesis is effective and feasible according to FPGA technology, this identity system can have the license plate be binarized more better, and get more accurate positioning and identity of license plate in complex weather conditions.