Detection and localization of manhole and joint covers in radar images by support vector machine and Hough transform

In this paper, a novel manhole and joint covers detection algorithm from radar images by Support Vector Machine (SVM) and Hough transform is proposed. Due to its dense and high-speed monitoring capabilities, Ground Penetrating Radar (GPR) is a promising tool. Furthermore, manhole and joint covers are apparent from surface reflections. An SVM model was developed utilizing Histogram of Oriented Gradient (HOG) feature and Laplacian filter. Classification accuracy of manhole, joint covers and pavement section was up to 98%. Hough transform was applied to the detected areas to visualize objects in a map. The algorithm detected manhole and joint covers accurately and fast by the combination of SVM and Hough transform.