A license plate location method combining the license plate location method based on the HSV color space and the license plate location method based on the boundary detection

Abstract

The first step of the license plate automatic recognition is the license plate location,therefore,the license plate location method is very important to the license plate automatic recognition.A license plate location method combining the license plate location method based on the HSV color space and the license plate location method based on the boundary detection is presented in this paper. A total of 216 experimental images are tested in this paper. Through the analysis of experimental results, it’s found that this method can achieve the license plate location accurately, and the method’s accuracy is 98.6 percent. Introduction The license plate location method based on boundary detection[1,3] and the license plate location method based on the color space[4,5] are the most common license plate location methods, according to the license plate color, the second method carries out the subsequent operations , and it performs well, but it will be failed when the plate is dyed with other color. In this paper, the license plate location method based on the HSV color space and the license plate location method based on the boundary detection are combined, and its license plate location accuracy is good. A license plate location method combining the license plate location method based on the HSV color space and the license plate location method based on the boundary detection At first,the RGB color space will converted to HSV color space[6,7] ,then, the binarization processing is taken, in the HSV color space, the binarization processing can be done by judging three components values of HSV color space. The value of H components are different when colors are different, if the value of S component and the V component value is fixed, the color will be changed when the value of H component is changed, Looking over all pixels of the image, value of the pixel will be set to 1 if each component value of the pixel is in the component value range of the color specified, otherwise, the value of pixel will be set to 0, the Fig.1 is the original image, and the Fig.2 is the binarization processing result of Fig.1. Fig.1 Fig.2 4th International Conference on Machinery, Materials and Computing Technology (ICMMCT 2016) © 2016. The authors Published by Atlantis Press 1913 Then, the morphology operation is taken to obtain license plate candidate areas, the morphology operation used in this method includes morphological closing operation and morphological open operation, after that, peripheral contours of the connected domains can be drawn and minimum tangent rectangles of the drawn connected domains can be calculated, and the minimum tangent rectangle areas are candidate areas of the license plate. After the morphology operation, the peripheral contours can be drawn and displayed on the original image, the result of the morphology operation is shown below in Fig.3, The length and the width of the Chinese license plate are specified, so the ratio of the Chinese license plate length and the Chinese license plate width is almost a constant value, that constant value is assumed to be M, and x is set as a parameter of allowable error, then the license plate area will be found by verifying the ratio of the minimum tangent rectangle length and the minimum tangent rectangle width. The minimum tangent rectangle area is considered to be the license plate area if the ratio of its length and its width is a value in [(1x)M,(1+x)M], the result is shown below in image Fig.4. Fig.3 Fig.4 Fig.4 is a successful result of license plate location method base on HSV color space, but the license plate location method base on HSV color space will be failed when every ratio of the minimum tangent rectangle length and the minimum tangent rectangle width isn’t a value in [(1x)M,(1+x)M]. In the HSV color place, the binarization processing can be done by judging the component value of the HSV color space, the binarization processing result of Fig.5 is shown below in Fig.6,and then, the license plate location method based on the HSV color space is failed. Fig.5 Fig.6 When the license plate location method based on the HSV color space is failed, the license plate location method based on the boundary detection will be followed to finish the license plate location. After some image preprocessings being taken the grayscale image can be acquired. The method will detect the grayscale image edge by using Sobel operator, the result is shown below in the Fig.7, After that, the binarization processing is taken, then, the morphology operation is taken to obtain license plate candidate areas, the result is shown below in image Fig.8, then, the peripheral contours of the connected domains can be drawn and minimum tangent rectangles of the drawn connected domains can be calculated, after the peripheral contours being displayed on the original