A new texture-based operator for edge enhancement of potential-field data

• Edge enhancement is necessary in potential-field data processing and interpretation.
• Edges differ from other anomalies not only in amplitude but also in texture.
• We propose a new texture-based operator for edge enhancement.
• The operator is based on the modified co-occurrence matrix and weighted dissimilarity.
• The operator overshadows other common methods in accurately capturing more details.

Edges of potential-field anomalies are assumed to result from sharp discontinuities or interfaces between contrasting materials, such as faults, unconformities, or intrusive contacts. The superposition effects of potential fields make edge-enhancement techniques necessary to reduce edge mislocation. Distinguishable amplitude difference is the edge feature most widely utilized in common enhancement methods. But edges differ from other anomalies not only in amplitude but also in texture, and in some cases they reveal as weak difference in amplitude but of meaningful contrast in texture. Accordingly, we propose a new texture-based operator for edge enhancement. The operator is a modified version of dissimilarity extracted from modified calculation of the gray-level co-occurrence matrix. The matrix is tailored to fit in with the characteristics of potential-field data and it differs from the common co-occurrence matrix mainly in three aspects: (1) for the modified matrix only the pairs between the central point of a computing window and other points are counted; (2) except from commonly-used eight directions, other directions are added for the windows with direction number depending on the window size; (3) co-occurring pairs are given weight to differ the contribution of joint occurrence in different distance. Moreover, we set standard-deviation-based weight to the common dissimilarity in each window in order to locally enhance edges. We evaluate the utility of the operator based on synthetic models and real data examples of the Huanghua Depression in the Bohaiwan Basin, Northern China, and of the Haijiao area in the East China Sea Basin, Eastern China. The modified dissimilarity is proved advantageous to the total horizontal derivative and common dissimilarity in capturing more detailed features, and provides more straightforward and accurate illustration of edges than the tilt angle does. The edges enhanced in the gravity data of the Huanghua Depression accord well with the faults deduced from an integrated geophysical interpretation. The edges detected in the magnetic data of the Haijiao area illustrate the distribution of rocks. The results show the potential of the texture-based operator to facilitate the interpretation of potential-field data in locating faults, rocks, or contacts, and thus to promote the application efficiency of the data in exploration and engineering.