倾斜车牌图像边缘模糊特征识别方法研究  

Tilt License Plate Image Edge Fuzzy Feature Recognition Method Research

作　　者：张晓娟[1] 樊东燕[1] 

机构地区：[1]山西大学商务学院信息学院,山西太原030006

出　　处：《计算机仿真》2017年第1期372-375,共4页Computer Simulation

基　　金：山西省科技厅基础研究项目(2014011018-1)

摘　　要：针对倾斜车牌图像识别过程中存在无法进行水平坐标采集,无法抵御外界因素干扰,倾斜角和识别区域不连贯,采用当前算法对倾斜车牌图像进行识别时误差大等问题,提出一种基于边缘模糊特征的倾斜车牌图像识别方法。先利用旋转投影法建立车牌图像彩色边缘的笛卡尔坐标系,并对车牌边缘图进行坐标变换,将旋转后车牌图像的垂直边缘向水平方向投影,确定车牌的倾斜角度,将图像分割为不同的区域,对倾斜车牌图像进行校正,采用模糊隶属度建立车牌图像边缘模糊特征集,将特征集内的各个特征进行相似度对比,有效地完成了对倾斜车牌图像的识别。仿真证明,所提算法对倾斜车牌图像识别率有较大提高。In this paper, we proposed an image recognition method of tilted license plate based on edge fuzzy feature. Firstly, we built Cartesian coordinate system of colorized edge in license plate image using revolve projection method and did coordinate transformation. Then we projected the vertical edge of revolved license plate image along horizontal direction to confirm incline angle. The image was divided into different regions to correct. We also used fuzzy membership to build fuzzy feature set of license plate image. Finally, we compared similarity of each feature in feature set to complete recognition effectively. The simulation results show that the algorithm mentioned above can improve image recognition rate apparently.

关 键 词：倾斜车牌图像 边缘模糊特征 识别 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]