3D Seismic Data Segmentation Based on the Orientation Coherence Method

This paper proposed a new volumn segmenting algorithm. Traditional segmentation methods like minimum-variance-based region growing define homogeneity criteria to be value related. This method using 3D seismic data´s ridge and valley like structure extracts local orientation coherece first, then divides the homogeneity criteria into two parts. The first part uses a high threshold and the second one uses a low threshold and a coherence threshold. The two parts criteria makes the growing processing better. This new method has good performance in ununiform brightness condition, but the conherence value cube is time-consuming, and the user should define two extra threshold.