Detection of bright spots in seismic signals using tree classifiers

Kou-Yuan Huang*, King Sun Fu

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

20 Scopus citations

Abstract

A tree classification method is proposed for the detection of bright spots in seismic signals. Envelope, instantaneous frequency, and polarity are used as the features. From the analysis of zero-phase Ricker wavelets, a tree classification technique is adopted. Bright spots are detected if one of three hypotheses is satisfied. The advantage of using a tree classifier is that it is easy to design and computationally efficient. Because of the frequency attenuation effect in a seismogram, a partitioning-method is introduced in conjunction with the tree classification for the detection of bright spots. Results from simulated and relative-amplitude real stacked seismograms are presented. The classification results show that the proposed method can be used to improve seismic interpretation.

Original languageEnglish
Pages (from-to)121-145
Number of pages25
JournalGeoexploration
Volume23
Issue number1
DOIs
StatePublished - 1 Jan 1984

Fingerprint Dive into the research topics of 'Detection of bright spots in seismic signals using tree classifiers'. Together they form a unique fingerprint.

Cite this