Fast Detection of GPR Objects with Cross Correlation and Hough Transform

A GPR object detection algorithm delivers a promising performance using the Hough transform through a high computational load. This paper presents a fast Hough-based algorithm. To reduce the parameter space of the Hough transform, first, two parameters for a reflection hyperbola were estimated using cross correlation between adjacent A-scans. Next, only a 1D Hough transform is necessary to detect an object compared with the 3D transform, which comprises the traditional Hough-based methods. Our method is compared with three other detection methods using field data. The results show that the proposed method has an encouraging detection ability and high computational efficiency.