Research On Plate Positioning Algorithm Based On Image Processing

License plate recognition has played a vital role in the modern traffic management, electronic toll collection and other intelligent transportation fields; therefore, it has a very broad application prospect. The license plate allocation is one of the key technologies in license plate recognition system. It is the premise of character segmentation and character recognition since it affects the accuracy of character segmentation and character recognition.In this thesis we research and analysis the algorithm of license plate location both at home and abroad, then take these recent algorithms as the basis for proposing an improved algorithm, and finally show experimental results. The main procesures that had been taken are as follows:1. Transform color images into gray images, then choose the histogram equalization to contrast enhancement by changing the histogram of license plate image.2. Compare three threshold technique, and use the Otsu threshold to binarization.3. In this thesis,we propose an algorithm for license plate localization based on rough-detail location. And this algorithm contains edge detection, projection method, binary, connected component labeling and least squares fitting. After edge detection, rough location, skew correction and detail location, the license plate area is determined.The experiment result shows that this algorithm that from coarse position to fine positioning finished the license plate localization. And it is better than similar algorithm in eliminating interference information such as border segmented. After this algorithm, the license plate character part is segmented, license plate character, the character for the subsequent part segment and character recognition creates powerful conditions, can achieve the desired goals.