Research On License Plate Image Location In Car Wash Centre

To achieve the automatic identification of car wash center membership vehicle, it is first need to locate the image of vehicle license plate. But due to the denser fog, the poor lighting as well as the dramatic change of brightness, a larger plate tilting angle in the image and a low pixel number, the direct application of the existing license plate localization method will cause very low plate location accuracy rate, this dissertation focuses on the localization method in this particular application environments.To improve image quality in car wash center, this dissertation first makes use of the dark channel priori algorithm to achieve the image defogging and a multi-scale Retinex algorithm with color restoration to achieve the image enhancement processing. On this basis,this dissertation proposes an enhanced color image localization algorithm of license plate based on improving the color image binarization license plate localization algorithm, which achieve the obvious improvement for the accuracy rate of license plate location in car wash center. Since the license plate localization accuracy rate will significantly reduced when the license plate is similar with vehicle body in color, this dissertation proposes a license plate localization algorithm based on image rotation by improving the license plate localization algorithm based on threshold segmentation to solve this problem. Although it results in a lower accuracy rate of license plate localization for license plate localization in car wash center,this new localization algorithm enables the obviously higher accuracy rate than the license plate localization algorithm based on color image enhancement when the license plate is similar with vehicle body in color. Finally, from the perspective of the complementary advantage of license plate localization algorithm based both on the image rotation and on the color image enhancement, this dissertation proposes a plate localization algorithm based on the threshold selection processing, which basically solves the problem of plate image localization in car wash center.Experimental results show that the two improved algorithms achieve higher accuracy rate than the original algorithms for the license plate localization in car wash center. The license plate localization algorithm based on the threshold selection processing can solve the problem of license plate localization and meet the needs of the practical applications.