Normalized Edge Detectors Using Full Gradient Tensors of Potential Field.

Edge detection is a widely popular data processing tool in magnetic and gravity exploration which plays an important role in defining shapes and horizontal locations of underground geological structures such as faults, rock boundaries, and minor lineaments. Conventional edge detectors, which are mostly derived from the total field anomaly, have long been applied in field exploration. However, the development of full gradient tensor measurement, which improves detection resolution and provides greater geophysical detail for researchers, has given greater importance to edge detection for full gradient tensors. We propose the use of multiple combinations of gradient tensor components and their eigenvalues to construct balanced boundary recognition filters, taking full advantage of information from the gradient tensor matrix. Different synthetic tests demonstrated that model boundaries were definitely detected, and our methods demonstrated great robustness to noise and oblique magnetization. Field processing of magnetic and gravity data using our methods also yielded clear lineaments and structure locations. Compared with other edge detectors derived from gradient tensors, boundary detection accuracy was considerably improved and false edges were nearly eliminated. [ABSTRACT FROM AUTHOR]