Abstract:There are a few algorithms commonly used for the edge extraction of potential data at present.By analyzing results of these edge extraction algorithms,we divide them into two types,viz.the step edge and the roof edge,and introduce the edge detection technique to data processing of gravity gradient tensor.Target boundaries are extracted directly by the step edge extraction algorithm and the roof edge extraction algorithm.The model experiments show that the target edge extracted by the edge detection algorithm is consistent with the real boundary very well.Comparing with previous boundary extraction algorithms,the new edge detection technique is more precise and continuous.By adding 5% Gaussian white noise to theoretical model,we prove that the edge detection algorithm has certain anti-interference ability.