摘要：【中文】本发明公开了一种探地雷达资料中管线目标的自动识别方法及系统，本发明使用区域卷积神经网络识别出图像中管线特征(双曲线)区域的定位准确、快速，模型的训练时间短；使用互相关运算和一维参数空间的霍夫变换从管线特征区域提取管线参数相较于传统的霍夫变换运算时间更短、对硬件要求低；使用对全图线进行区域识别，再从区域中提取参数的方式相较于传统的全图搜索、计算的方式避免了对非特征区域的计算，大大提高了计算效率。 【EN】The invention discloses an automatic identification method and a system for a pipeline target in ground penetrating radar data, wherein the method uses a regional convolution neural network to identify that the location of a pipeline characteristic (hyperbolic curve) region in an image is accurate and rapid, and the training time of a model is short; compared with the traditional Hough transform operation, the time for extracting the pipeline parameters from the pipeline characteristic region by using the cross-correlation operation and the Hough transform of the one-dimensional parameter space is shorter, and the requirement on hardware is low; compared with the traditional full-graph searching and calculating mode, the mode of identifying the area of the full graph and extracting the parameters from the area avoids the calculation of a non-characteristic area, and greatly improves the calculation efficiency.