The Research Of License Plate Location Technology On License Plate Recognition System

License plate recognition system is an important part of modern intelligent traffic management system,which plays an important role in the highway electronic toll,urban intersection "electronic police",district vehicle management system and other fields.License plate location technology which acts as the core technology of license plate recognition system has been one of the important topic in the field of pattern recognition.In this thesis,the license plate location problem is studied in depth.The main work of the thesis includes:(1)The algorithm of the initial location of the license plate is proposed by combining the image binarization of HSV image with mathematical morphology.Initial location uses the HSV image to extract the color feature because of its integrity,the author selects color interval according to the background color of the license plate and scan each pixel line by line,then determine this condition of realizing the image binarization by judging whether the HSV color value of the pixel is satisfied with the condition of the background color selection.Then the author uses mathematical morphology theory to regional connectivity or eliminate noise of binary image according to the characters of vehicle plate area and border texture features.Finally,the author exracts feature of the ratio of the length and width of the license plate to find the license plate area to be recognized,the initial location of the license plate is completed.(2)The thesis proposed a method to realize the precise location of license plate by combining the light and dark jump method and the edge detection technology.In binarization image,the author uses the license plate area arrangement features of the characters,through to determine whether a scan line appear similar to the license plate character jump characteristics of light and shade,the purpose is to achieve the positioning of the top and bottom boundary of the license plate area.The result is not ideal for the localization of the left and right boundary by the method of the gray jump,the author uses the special edge detection algorithm to achieve the edge of the license plate extraction,and analyz edge pixels by vertical histogram,and set the left and right boundary thresholds,then the author completes the positioning of the license platet of its left and right boundary.(3)The thesis uses the Radon transform to detect the tilted license plate image,and uses geometric algorithm to correct its tilted angle and geometric deformation of the license plate image.(4)The thesis prepared a simulation program which uses the MATLAB language.Through the simulation program,the author completed the location test of 153 license plate images which were taked in the entrances and parking lots,and the effect is good.