地表探测雷达图像ROI的人工神经网络提取方法  

ROI Extraction Method of GPR Images with an Artificial Neural Network

作　　者：陈德莉[1] 粟毅[1] 黄春琳[1] 

机构地区：[1]国防科技大学电子科学与工程学院,湖南长沙410073

出　　处：《雷达科学与技术》2005年第6期372-377,共6页Radar Science and Technology

基　　金：国家863高技术计划项目(No.2001AA132020)

摘　　要：双曲线特征是探地雷达二维回波图像中的小目标的一个先验知识,本文根据这一特征提出了一种利用人工神经网络(ANN)进行感兴趣区域(ROI)提取的方法。根据探地雷达(GPR)数据解译的特点,设计了相应的B-SCAN图像预处理方法和ANN的结构、训练及测试方法,进而用塑料管和其他目标的B-SCAN实测数据对ANN进行训练和测试,并将检测结果与人工判读结果加以比较,结果表明此法可行而有效。Based on a priori knowledge hyperbolic feature of subsurface objects in GPR B-SCAN ima- ges, a method of extracting the region of interesting （ROD is presented, which is on the basis of ANN. According to the characteristic of GPR interpretation, this paper has designed a method of the BSCAN images pre-processing, and corresponding structure of the neural network and its training and testing method. Then the real B-SCAN data of plastic-pipe and other objects collected by a GPR set are used to train and test the ANN and compared with the information extracted by a trained operator. The results indicate this method is feasible and effective.

关 键 词：人工神经网络 探地雷达 感兴趣区域 检测 

分 类 号：TN959.71[电子电信—信号与信息处理]