基于纹理特征的车牌定位方法

为了提高车辆牌照定位成功的概率以及定位精度,提出了一种基于纹理特征,采用自适应二值化的车牌定位方法。该方法首先利用小波分析对图像进行预处理,提取车牌图像字符区域的纵向纹理特征,然后利用边缘检测算子对图像纹理特征进行二次提取,再进行自适应二值化。该方法克服了直接对小波分析后图像进行二值化时,阈值选取非常困难的缺陷。实验结果表明,该方法能够达到提取有效车牌图像的目的,适用于各种复杂条件下拍摄的车牌图像定位。

Location Method of Vehicle-license-plate Images Based on Texture Feature

In order to improve the probability of the successful localization and the localization precision of vehicle license plate,an license plate localization method based on texture feature and adaptive binary is presented. The method emphasizes utilization of wavelet analyses to preprocess images and to extract vertical texture feature from them,and accomplishes second extraction with edge detection operators,eventually,adopts adaptive binary to detect license plate. Experimental results show that the proposed method is able to overcome thresholding sensitivity due to binary after directly using wavelet analyses,and to extract effectively license plate. This method is able to localize the vehicle license plate shot under various kinds of complicated conditions.