Research And Implementation Of Algorithm Of Vehicle License Plate Recognition System

With the development of technology and improvement of society, the vehicleappears explosive increasing; this not only burdens the society, but also lowers themanagement efficiency. How to manage this problem becomes a focus in the wholesociety. The intelligent transport system emerges under this background. The licenseplate recognition system mainly uses in the crossroads, parking lot, high-wayintersection and so on, enormous improving the management efficiency.The license plate recognition includes three parts: license plate location, characterssegmentation, characters recognition. In this paper’s research, it includes the three partsand proposes the new algorithm and improvement algorithm on every part.(1) License plate location algorithm. In this paper, the location algorithm combinesthe plate’s rich texture and fixed color collocation between the characters and thebackground. Using the sliding window slides the edge point in the edge image andjudges the point in the sliding window whether its color coincidence the plate’s colors,then finding plate’s areas. Using SVM to classify the real plates and non-plates, andthen tilt revising and precise location for the real plates.(2) Characters segmentation algorithm. In this paper, the segmentation algorithmimproves the projection curve segmentation algorithm and combines the sliding window.Using the sliding window to find the maximum gap in the plate firstly, and thencombines the width of the character and peak and valley of the curve to find the rightplace to segment the character.(3) Characters recognition algorithm. In this paper, the recognition algorithm usesthe improvement and multiple-step SVM. The model is trained according to platecharacters’ features and specific model for the similar characters for second recognition.The algorithms in this paper have tested by mass pictures and achieved better resultthan traditional algorithms.