Improved fault segmentation using a dip guided and modified 3D Sobel filter

Automatic fault detection and extraction is still considered to be a major challenge in the industry. Faults are usually detected with the use of edge detection attributes like variance, coherence, chaos, and semblance. This abstract presents the results of a modification of a classical edgg detector, namely Sobel. The detector is modified with weighting, normalization and dip guiding to achieve the presented results. The data set used for testing is Solsikke, which is located off the Norwegian coast. To validate the results, ant tracking is used on the attribute results to extract detected faults. The results are also volume rendered to show some fault patches in the selected sub‐volume. As a reference attribute for comparison, we are using variance. The comparison results are consistently in favor of the modified Sobel.