摘要
在用探地雷达对地下管道进行探测时,地下管道的雷达图像呈现明显的双曲线特征,但地下管道的管径大小却无法在雷达图中直接反映出来。为了能够同时测量地下管道管径,本文首先采用一种改进的小波包阈值去噪法对雷达图进行去噪预处理,然后采用Hough变换对特征双曲线进行拟合,确定管道的位置,从而实现管道的检测,最后采用支持向量机的方法对拟合的双曲线进行分类,将双曲线分类结果对应到管径分类上,从而间接地实现地下管道管径的识别。实验结果表明:该方法对地下管道管径大小的正确识别率达到90%。该方法能够满足工程应用的实际需要。
When detecting underground pipe diameter size using Ground Penetrating Radar（GPR）, the radar maps of underground pipe appear obvious hyperbola feature, and the diameter size of underground pipes can not be reflected directly in radar maps. In order to measure the pipe diameter synchronously, this paper first uses an improved wavelet packet threshold denoising method to preprocess the radar maps, then use Hough transform to simulate hyperbola and determine the positions of pipes. Thus, the detection of pipes can be finished. At last, an SVM method is adopted to classify the simulated hyperbolas and then correspond the classification results of hyperbola to pipe diameter classification, so as to realize the recognition of underground pipes diameter indirectly. The experimental results show that the proposed method has higher recognition rate up to 90%. Therefore, this method not only can fit the requirement of engineering application, but also brings convenience to municipal engineering construction.
出处
《信息与电子工程》
2011年第4期403-408,共6页
information and electronic engineering
基金
国家自然科学基金资助项目(60940017)
关键词
小波包
阈值去噪
霍夫变换
支持向量机
wavelet packet
denoising
Hough transform
Support Vector Machine
作者简介
廉飞字（1970-），男，郑州市人，硕士，副教授，近年来发表论文40多篇，主要从事电磁波检测技术、计算机应用等研究．email：lfywork@163．com．
李青（1962-），男，武汉市人，教授，博士生导师，主要研究方向为高性能计算．
