针对探地雷达图像中自动提取地下管线的问题,提出了一种基于广义Hough变换的地下管线提取算法。根据地下管线在图像上呈现双曲线且具有一定宽度和成对出现的图像特征,首先对图像进行预处理,接着采用全向跟踪法进行边缘跟踪;然后利用双曲线方程线性化后的结果进行投票计算,确定出一对双曲线的描述参数,从而提取出图像中的地下管线特征。实验结果验证了该算法对地下管线提取的有效性。
To extract underground pipeline from Ground Penetrating Radar(GPR) images,a method was proposed by applying generalized Hough transform algorithm.According to the image features that the underground pipelines are the hyperbolic and have the certain width in pairs in image,the image is pretreated at first,and then the omnidirectional tracking method is used in edge tracking.Based on the hyperbolic feature model,the iterative calculation is carried out using the linear equations,and the description parameters of the hyperbola are determined.Then,the hyperbola in image is extracted.Experimental results indicated that the proposed approach was effective in extracting the underground pipeline and had robustness with varies conditions.