基于数学形态学与Radon变换的车牌定位法  

A Mathematical Morphology-Based Radon-Transfered Method for License Plate Location

作　　者：胡红萍[1] 白艳萍[1] 

机构地区：[1]中北大学理学院,山西太原030051

出　　处：《数学的实践与认识》2012年第24期109-113,共5页Mathematics in Practice and Theory

基　　金：中国第48批博士后科学研究基金(20100481307);国家自然科学基金(61275120);山西省自然科学研究基金(2009011018-3)

摘　　要：从粗定位和精定位方面考虑车牌定位.利用腐蚀、膨胀和差运算将RGB车牌图像转换为车牌图像位置为彩色的而其余位置是灰色的图像,将所获得的图像转化为灰度图像,利用Robert算子进行边缘检测,再进行腐蚀和膨胀运算,根据阈值获得连通区域,利用颜色特征实现车牌的粗定位.将粗定位后的图像进行二值化、差运算、Radon变换,再进行腐蚀运算,图像的填充等步骤,再考虑图像的上下左右边界实现车牌的精定位,对185幅车牌图像进行定位.通过实验,车牌定位准确率达到91.89%.说明所提出的方法简单易行,具有一定的实效性.In this paper, the license plate location is considered from the wide location and the refined location. Turn the inputted RGB image with license plate into the image with license plate into the image in which the license plate region is colorful and the remains restrain gray by use of erosion, dilation and difference. Change the obtained image into gray image and detect the edge of the gray image by Robert operator. Erode and dilation the obtained edge image and obtained the connect regions with the threshold. The wide location of license plate is obtained by using the color feature. The refined location is achieved by considering the up-down-left-right bound of the obtained image by binaryzation, difference operation, Radom transfer, erosion operation, filling etc. of the wide image. Take 185 license plate images located in this paper. In the experiments, the accurate rate of the license plate location is 91.89%. So the above proposed method with a certain effectiveness is easy to achieve.

关 键 词：数学形态学 RADON变换 车牌定位 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]