next up previous print clean
Next: My investigations Up: Detection of seismic discontinuities Previous: Discontinuity, picking, interpolation, and

Future

Interpretation of seismic subsurface images can and ought to be aided by computer processing. The delineation of faults is a step towards general automatic information extraction from seismic images. Automatic seismic image analysis would ease the tedious and time-consuming extraction of geological information from seismic image volumes by human experts. Additionally, automatic analysis could objectively and reproducibly compare the effect of competing seismic image processes on the interpretation of seismic data. However, artificial intelligence experiences great difficulties in the development of visual scene recognition and interpretation ().

For continued research into automatic seismic image analysis, I advocate (1) a classification framework for seismic features and (2) a sophisticated segmentation scheme. I wonder if basic statistical measures - mean, mode, variance, autocorrelation, and grey level histograms - can classify seismic image features and their boundaries - layered sediments, salt, base rock, off-lap surfaces, on-lap surfaces. If such a classification is feasible, the subsurface could be separated into amorphous segments of similar, stationary statistics and, consequently, regions of identical geology. Interpreters of remote sensing data successfully use similar classification and segmentation algorithms ().


next up previous print clean
Next: My investigations Up: Detection of seismic discontinuities Previous: Discontinuity, picking, interpolation, and
Stanford Exploration Project
3/8/1999