Fast License Plate Positioning System Design And Implementation

License plate is the only identification of vehicle, the license plate recognitiontechnology plays an important role in the modernization of our country traffic as thecore of ITS. License plate location is the largest amount of data processing in threemain modules of license plate recognition, it is the premise and foundation ofcharacter segmentation and character recognition. the positioning accuracy and speedof license plate location affects the entire license plate recognition system directly, soit is the critical and difficult of license plate recognition system. In this paper, weresearch image preprocessing detailedly, determined the algorithm and process forlicense plate location pretreatment on the basis of results and the aim of real-time, wepresent a fast algorithm for license plate location and given the experimental results.The main work is as follows:1、The approximate formula for the color image gray-scale conversion methodusing a weighted average of the best.2、 We compare effect and speed of the common binarization algorithm:empirical formula, Otsu algorithm and Bernsen algorithm, choose the empiricalformula with the best positioning effect and the fastest speed on the basis ofsummarizing the advantages and disadvantages of each algorithm.3、We compare detection effect and elements of speed for commonly usedseveral kinds of edge detection operator synthetically, This paper finally choseRoberts operator. It’s fastest, no more pixel width and edge connectivity.4、We research common filtering methods extensively, choose the median filterby comparing the advantages and disadvantages. Determining the filter fourparameters: the height of the filter, the width of the filter, X coordinates of the centerof the filter element, Y coordinates of the center of the filter element.5、We used projection method combined with the corresponding decisioncondition to determine the up, down, left and right boundary in the algorithm for license plate location. This method saves a lot of digital image processing technologyin coarse positioning and fine positioning analytical method on the basis of guaranteethe positioning accuracy, so the arithmetic speed is faster.In conclusion, we pay attention to the improvement of arithmetic speed in theimage preprocessing step, realized good positioning result and fast arithmetic speed inthe case that the license plate image is not tilted serious, not dirty is occluded andsingle license plate. Our research has a certain reference value to realize real-timelicense plate location.