摘要
为简化场景图处理的计算量,针对场景图复杂的颜色信息,提出了一种快速的交通标志检测预处理方法——颜色标准化。将颜色信息复杂的场景图映射成简单的由8种标准颜色组成的图像,并提取5种与交通标志相关的感兴趣颜色,滤除冗余区域后得到标准颜色的交通标志。该方法大幅度简化了场景图颜色信息的复杂性,节省了RGB-HSI模型转换的计算时间。实验表明,该预处理方法具有很好的鲁棒性,快速准确地实现了交通标志检测。
To reduce the computational complexity in the scene image processing, this paper proposes a fast method of pre-processing for traffic signs detection based on the complex color information, i.e., color-standardization as defined in this paper.It uses the RGB color-components to threshold a complex scene image which is mapped to a simple benchmark image consisting of standard color of eight categories.Five colors relating to traffic signs are extracted to form the standardized traffic signs region.Compared with the generic method of color extraction based on the HSI model,the proposed method substantially attenuates the complexity of color information in the scene image, and is time saving in computing color model conversion.Experimental results show that the presented approach indicates better robustness, and achieves the traffic sign detection with high processing speed and efficiency.
出处
《计算机工程与应用》
CSCD
北大核心
2010年第29期229-231,244,共4页
Computer Engineering and Applications
基金
浙江省科技计划项目(重点)的资助(No.2007C21179)
关键词
交通标志检测
颜色标准化
分量阈值化
RGB模型
traffic sign detection
color-standardization
component thresholding
RGB model
