一种改进的越南车牌定位方法AnImprovedMethodforVietnam.PDFVIP

Scientific Journal of Information Engineering 一种改进的越南车牌定位方法* 1, 2 1 1 1 梅荣誉 ，王睿智 ，张红云 ，高灿 1. 同济大学 计算机科学与技术系，上海 201804 2. 越南交通大学 电力与电子系，越南 河内 10000 摘 要：车牌定位（License Plate Location - LPL）是车牌自动识别 （Automatic License Plate Recognition - ALPR ）的关键步骤。本 文基于灰度图像的形态操作及图像减法、二值化图像和边缘检测（Canny 操作），提出了一种改进的越南车牌定位方法。算法使 用Radon 变换和双线性插值定位车牌的角度并进行旋转，然后根据越南车牌地区属性测量提取车牌的区域。在不同环境（白天和 黑夜）、角度、光照、规模、类型和颜色的500 幅越南车辆照片上的实验分析表明，本文提出的算法较已有算法对越南车牌定位 问题更为有效，达到97.27 ％准确率，说明了算法的有效性。 关键词：车牌定位；图像转换；边缘检测；图像减法；数学形态学；Radon 变换 An Improved Method for Vietnam License Plate Location 1, 2 * 1 1 1 Mai Vinh Du , Ruizhi Wang , Hongyun Zhang , Can Gao 1. Department of Computer Science and Technology, Tongji University, Shanghai, China, 201804 2. Department of Electrical Electronic, University of Transport and Communications, Hanoi, Vietnam, 10000 Email: maivinhdu@ Abstract: The License Plate Location (LPL) is the key step in the Automatic License Plate Recognition (ALPR). In this paper we proposed a new LPL algorithm for Vietnam license plates (LP), which included preprocessing, morphology operation and image subtraction on grayscale image, image binarization based on threshold, edge detection using Canny operator, morphology operation on binary image, finding the LP angle rotating LP based Radon transform and bilinear interpolation, and then cutting exactly LP region based on measuring properties of Vietnam LP