基于纹理的牌照二值化算法(国外英文资料).docVIP

基于纹理的牌照二值化算法 A texture based two valued method for license plate images Ye Chenzhou. Liao Jinzhou. In the vehicle license plate recognition system, in order to facilitate the character recognition, the gray image of the license plate which is cut from the preprocessing part should be treated with two values so that the characters can be extracted Because of the uncertainty of the characters in the original image and the large gray scale range of the character pixels, it is difficult to realize the two values. In this paper, a texture based two valued method is proposed Firstly, the texture analysis of images is made, then the two minimum threshold is used to obtain the two valued threshold. The method is also applicable The two value of a bill or text image. Subject headings, texture analysis, pattern recognition, two valued I. Introduction In the vehicle license plate recognition system, preprocessing is necessary to facilitate character recognition Partially cut down the license gray image (see Figure 1) for two value processing The characters are black and the background is white. The position of the characters in the original license image, Of variable size, and especially special, is the unpredictable gray level of the characters in the license plate The level is relative high or low relative to the background, which makes it difficult for two values. Figure 1 In general, the solution to this problem is to use grayscale histogram, that is, the right histogram The image of the license plate is taken as a gray level statistic, and the characters in the image are distinguished according to the statistical results The gray range of the symbol. For example, a gray histogram of an ideal license plate image is shown in Figure 2 As shown (abscissa represents gray level, ordinate represents number of pixels). Exist in graph Two distinct peaks; ideally, a higher peak indicates the background of the license plate, The lower reaches of the crest represents the characters i