摘要
针对交通标志识别系统要求实时性高,处理的信息量大以及受多种情况干扰等问题,提出了一分块图象特征与BP神经网络相结合的识别方法。对33幅图象加入不同干扰后得到132幅图象,实验表明该特征提取方法具有良好的灰度畸变、旋转、平移以及尺度不变性,取得了良好的识别效果。
The road traffic sign recognition would be disturbed. So the recognition must be frequent and exact. A method using subblock image features with BP Neural Network was presented. Experiments performed on 132 images show that the method keeps good invariance under gray transformation and geometry distortion,the results was good.
出处
《微计算机信息》
北大核心
2006年第10S期303-304,共2页
Control & Automation
基金
天津市自然科学基金:023600111
关键词
交通标志
图像识别
BP神经网络
Road Traffic Sign,Image Recognition,BP Neural Network
