基于改进模糊熵的车牌定位方法  

A License Plate Location Method Based on Improved Fuzzy Entropy

作　　者：黄鑫娟[1] 房岩[1] 周洁敏[1] 刘伯扬[1] 王占军[2] 陶思钰[1] 

机构地区：[1]南京航空航天大学民航学院,江苏南京210016 [2]南京林业大学林木遗传与生物技术省部共建教育部重点实验室,江苏南京210037

出　　处：《计算机技术与发展》2010年第1期193-196,200,共5页Computer Technology and Development

基　　金：2009年大学生创新基金

摘　　要：车牌定位是车牌自动识别系统中关键技术之一,定位的准确性直接影响到后期识别的效果。为了提高定位的准确性,提出了一种高效实时的车牌定位方法。该方法首先将彩色图像转化为灰度图像,并对图像进行灰度拉伸和灰度均衡,用中值滤波平滑图像,以去除噪声点;其次利用Sobel垂直算子检测出边缘,用模糊熵法求出二值化图像;最后结合水平扫描和垂直扫描定位出车牌。经过大量实验证明,该方法具有较好的自适应性,在不同外界环境下能够准确快速地定位出不同型号的车辆牌照。License plate location is a key technique in the automated license plate recognition system.In this system,the accuracy of location has a big influence on the effect of recognition.To enhance the accuracy of the location,present an efficient and a real-time method of license plate location.In the method,firstly,change the color image into the gray image,use gray level transform and histogram equalization method to enhance the image,in order to remove the noise,median filter is used to smooth the image;Secondly, extract out the vertical edges of the car image using Sobel vertical operator, display the threshold image using fuzzy entropy method; And finally combine the method of horizontal scan and vertical scan on the threshold image so as to segment the license plate out from the original car image. The experiments show that this method has preferable adaptive performance,and it can locate the different types of vehicle license plate quickly in different environment.

关 键 词：模糊熵 车牌定位 图像预处理 SOBEL算子 准确性 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]