next up previous print clean
Next: Segmentation Methodology Up: Lomask et al.: Improved Previous: Lomask et al.: Improved

Introduction

An accurate velocity model is needed to image beneath a salt body. This velocity model is typically created by manually picking the top of the salt. Improvements are then made to the velocity model and the data is remigrated causing the salt boundary to move and refocus. This process may be repeated several times and the manual picking of the salt boundary can be time consuming. Using image segmentation to pick the boundary could make this process easier. The interpreter is then only required to check the results and make some minor modifications.

Hale and Emanuel (2002, 2003) apply the normalized cut image segmentation method developed by Shi and Malik (2000) to paint a 3D coherency-based reservoir model. Our approach for tracking salt boundaries is similar Lomask (2003). This image segmentation technique creates a matrix containing weights relating each pixel to every other pixel in a local neighborhood. The matrix is then used to cut the image where the normalized sum of weights cut is minimized. We have modified the weight calculation to be dependent on the negative absolute value of the complex trace (instantaneous amplitude) of the seismic. This makes the weights very weak at salt boundaries, causing the segmentation algorithm to cut along the boundary.

In this paper, we give a very general review of the normalized cut segmentation technique. We then describe how we modified it for application to salt dome seismic data. We test this technique on synthetic seismic sections to illustrate its efficacy with discontinuous salt boundaries. Approaches to make this method more robust and cost effective are also presented.


next up previous print clean
Next: Segmentation Methodology Up: Lomask et al.: Improved Previous: Lomask et al.: Improved
Stanford Exploration Project
5/23/2004