Edge Detection in Potential-Field Gradient Tensor Data by Use of Improved Horizontal Analytical Signal Methods

Abstract

Potential-field gradient tensor data contain nine signal components. They include higher-frequency signals than potential field data, which can help delineation of small-scale features of the sources. Edge-detection technology has been widely used to delineate the edges of the sources. We need to develop a new edge detector to process gradient tensor data. Many methods are used to recognize the edges of data. The analytical signal method is a widely used edge-detection filter. We make some improvements to the analytical signal method so it can process potential-field gradient tensor data. We define new filters based on the horizontal directional analytical signal and the second-order horizontal directional analytical signal. To display the large and small amplitude edges simultaneously, we present two normalization methods: use of the maxima of nearby values to normalize the center point in a moving window and use of different orders of vertical derivatives to normalize the new filters. The methods were tested on synthetic and real potential-field gradient tensor data to verify their feasibility. Compared with other balance filters, the normalized second-order horizontal directional analytical signal and true vertical derivatives of the directional analytical signal normalized by use of the vertical derivative of vertical gravity gradient furnish better results and reveal more detail.