摘要
针对从移动测量车采集的图像中提取交通标志的需求,研究探索适合移动测量的交通标志检测识别方案。在交通标志检测阶段,提出一种检测多个圆形禁令标志连接的多圆度检测法,对于三角形警告标志的检测,在原有拐角检测法基础上提出基于候选区域的拐角检测法;在交通标志识别阶段,选用像素特征作为特征向量,采用概率神经网络实现交通标志的分类识别。最后设计实现移动测量后处理软件的交通标志自动提取模块。
Based on the demand of automatic extraction of traffic signs from photos collected by mobile mapping system, it presents a traffic sign detection and recognition solution suitable for mobile mapping system. In the detection phase, an improved round degree method is proposed to detect many connected round prohibitive signs. As for the triangular warning signs, a new regional corner detection method is proposed. In the recognition phase, the pixel feature is selected as eigenvectors; then a probabilistic neural network（PNN） is used to classify the different traffic signs. Finally a traffic sign automatic extraction module of mobile mapping system software is given.
出处
《测绘工程》
CSCD
2014年第5期13-17,20,共6页
Engineering of Surveying and Mapping
关键词
移动测量
交通标志自动提取
概率神经网络
mobile mapping
automatic extraction of traffic signs
probabilistic neural network
