Accession Number : ADA181094

Title :   An Artificial Intelligence Technique for Automating Seismic Stratigraphic Interpretation,

Corporate Author : RICE UNIV HOUSTON TX DEPT OF ELECTRICAL AND COMPUTER ENGINEERING

Personal Author(s) : Shaw,Scott W ; DeFigueiredo,Rui J

PDF Url : ADA181094

Report Date : 25 Nov 1986

Pagination or Media Count : 27

Abstract : Studying the character of reflected seismic wavelets may reveal facts about the stratigraphy of the reflector. Computers can aid humans in this task by revealing structural similarities between various wavelets. The relational tree is a good way to represent a waveform's global structure. By representing a waveform as a relational tree, processing it symbolically, and clustering the processed trees, a seismic wave form recognition system can be constructed. The symbolic processing is based on a tree transformation. An objective function, which measures the effectiveness of such a transformation, utilizes the ratio of between-cluster to within-cluster scatter. The action of a tree transformation applied to tree spaces is the same as linear discriminants applied to feature spaces. When tested on simulated seismic data, the relational tree waveform recognition system performs well at high signal-to-noise ratios.

Descriptors :   *WAVEFORMS, *PATTERN RECOGNITION, *SEISMIC WAVES, ARTIFICIAL INTELLIGENCE, REFLECTORS, SEISMIC DATA, STRATIGRAPHY, VARIATIONS, COMPUTER AIDED DIAGNOSIS, LITHOLOGY

Subject Categories : Seismology
      Cybernetics

Distribution Statement : APPROVED FOR PUBLIC RELEASE