| 道路标志图案识别方法研究 |
| |
| 引用本文： | 初秀民,严新平,毛喆.道路标志图案识别方法研究[J].汽车工程,2006,28(11):1051-1055. |
| |
| 作者姓名： | 初秀民 严新平 毛喆 |
| |
| 作者单位： | 武汉理工大学ITS研究中心,武汉,430063;吉林大学交通学院,长春,130025;武汉理工大学ITS研究中心,武汉,430063 |
| |
| 基金项目： | 国家自然科学基金项目(50578128)，高校博士点科研专项基金(20050497009)，吉林大学青年教师科研基金(419070100095)资助。 |
| |
| 摘 要： | 针对道路标志图案自动识别,以转弯道路标志图案为例,通过颜色空间变换,将道路标志图像的RGB量值转换为H(hue,色度)S(saturation,饱和度)I(Intensity,亮度)量值,利用H、I作为分类器的特征值,设计了欧式距离分类器,实现道路标志背景颜色的识别。采用固定阈值法分割道路标志图像的图案区域,并对分割后的区域进行Daubech ies-4正交小波变换,利用小波变换系数的能量值和方差比值作为道路标志图案的纹理特征。最后,设计了BP神经网络道路标志图案分类器,实现道路标志图案的自动识别。实验结果表明,文中提出的方法可行。
|
| 关 键 词： | 安全辅助驾驶 道路标志 图像识别 小波变换 |
| 收稿时间： | 2005-09-16 |
| 修稿时间： | 2005-09-162005-11-18 |
A Study on Traffic Signs Image Recognition |
| |
| Chu Xiumin,Yan Xinping,Mao Zhe.A Study on Traffic Signs Image Recognition[J].Automotive Engineering,2006,28(11):1051-1055. |
| |
| Authors: | Chu Xiumin Yan Xinping Mao Zhe |
| |
| Institution: | 1. ITS Center, Wuhan University of Technology, Wuhan 430063 ; 2. College of Transportation, Jilin University, Changehun 130025 |
| |
| Abstract: | Taking turning signs as example,a method for automatic recognition of traffic signs is presented in this paper.First,the RGB(red,green and blue)model is converted into the HIS(hue,saturation and intensity)color model.Then a Euclidean distance classifier is designed to recognize the background color of traffic sign with H and I as eigenvalues.And the area of interest of traffic signs is segmented using constant threshold technique.And then the Daubechies-4 wavelet transform is performed on segmented areas and the ratio of energy /variance of wavelet transform coefficients is used as texture features.Finally,a BP neural network classifier is designed to realize the automatic recognition of traffic signs.The experiment results show that the proposed method is feasible. |
| |
| Keywords: | Driving safety assistance Traffic signs Image recognition Wavelet transform |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
