基于纹理特征和神经网络牌照定位方法.pdf

文档介绍：þ ü
2
2
2
《现代电子技术》2010 年第 4 期总第 315 期计算机应用技术
基于纹理特征和神经网络牌照定位方法
王晓芳1 ,赵勇2 ,方宗德2
(1. 陕西职业技术学院陕西西安 710100 ;2. 西北工业大学机电学院陕西西安 710068)
摘要:汽车牌照的定位是一个较难解决的图像分割问题,神经网络为此问题的解决提供了一个有力的解决工具。这
里提出一种基于纹理特征和神经网络的牌照定位方法,首先采用基于纹理特征的方法对车牌图像进行行定位和列定位,确
定车牌候选区域,然后采用基于神经网络的方法对候选区域的特征进行分析判断,确定车牌区域。实验结果表明,该方法可
以实现汽车牌照的快速定位,并且定位准确率较高,鲁棒性较好。
关键词:牌照定位;纹理特征;神经网络;图像分割
中图分类号: TP391. 4 文献标识码:A 文章编号:1004 - 373X(2010) 04 - 049 - 03
License Plate Locating Method Based on Texture Features and work
WAN G Xiaofang1 ,ZHAO Yong2 ,FAN G Zongde2
(1. Shaanxi Vocational and Technical College ,Xi′an ,710100 ,China ;
2. College of Mechanical and Electronic Engineering ,Nothwestern Polytechnical University ,Xi′an ,710068 ,China)
Abstract :A fast license plate locating method is problem to be solved in the field of license plate recognition in the back
ground. A license plate locating method based on texture features and work is proposed. Firstly a method based on
texture features can be used to determine row location and column location ,the candidate regions can be found. Then the fea
tures of candidate regions is analyzed by determining the license plate region using method based on work. The ex
perimental results show that this method is excellent in the accuracy and the robustness.
Keywords :license plate location ;texture features ;work ;image segmentation
位,确定候选区域。
0