摘要
车牌字符分割是车牌识别系统中的核心步骤,而车牌预处理的效果直接关系到分割的准确率;针对传统基于灰度图的预处理方法难以消除由拍摄硬件和成像环境造成的干扰特征,提出一种基于R通道和灰度拉伸的车牌图像预处理方法;该算法将原始图像以R通道的数据表征,抑制车牌成像的干扰特征,提高了字符与背景底色的区分度;为了进一步增强图像的对比度,提出改进的灰度拉伸算法,有效分离字符和背景;为验证提出的预处理算法对字符分割的效果,引入一种基于投影和模板匹配的分割算法,实验表明,该算法不仅改善了污损车牌的成像效果,同时也有效提升了分割准确率。
Vehicle license plate character segmentation is the core step in license plate recognition system,and the effect of license plate preprocessing is directly related to the accuracy of segmentation.Aimed at the traditional gray image-based preprocessing method,it is difficult to eliminate the interference characteristics caused by the shooting hardware and the imaging environment.A preprocessing method of license plate images based on R-channel and grayscale stretching is proposed.The algorithm characterizes the original image with R channel data and restrain the interference characteristics of the license plate imaging and enhances the distinguishing degree between the background color and the character.In order to further enhance the contrast of the image,an improved gray stretch algorithm is proposed to effectively separate the characters and background.In order to verify the effect of the proposed preprocessing algorithm on character segmentation,a segmentation algorithm based on projection and template matching is introduced.Experiments show that this algorithm not only improves the imaging performance of the defaced license plate,but also improves the segmentation accuracy.
作者
赖道亮
赵平
钟昆
牛新征
Lai Daoliang1 , Zhao Ping2 , Zhong Kun2 , Niu Xinzheng3(1. College of Economics and Information Engineering, Southwestern University of Finance and Economics, Chengdu 611130, China;2. Zigong City Public Security Bureau Traffic Police Detachment, Zigong 643000, China; 3. College of Computer and Engineering, University of Electronic Science and Technology, Chengdu 611731, Chin)
出处
《计算机测量与控制》
2018年第9期250-254,共5页
Computer Measurement &Control
基金
四川省公安厅科研项目(2015SCYYCX06)
四川省科技计划项目(2017FZ0094)
关键词
图像对比度
RGB颜色空间
灰度拉伸
字符分割
image contrast
RGB color space
grayscale stretching
character segmentation
