Presentation on theme: "Figures Cluster World Oct OpendTect article"— Presentation transcript:
1 Figures Cluster World Oct. 2004 OpendTect article For optimal printing quality do not copy Figures 3, 4 and 6a, 6b and 6c from this Powerpoint file. Instead use original tif images!The OpendTect logo is delivered in EPS format.Contact for other formats and/or questions.
2 Figure 1. Open Source model. Third PartydGB-GroupOwnership, M&S ResponsibilityLicense and M&S feesWaived fees (dGB plugins)Commercial UsersAcademic UsersCommercial Plug-insBaseFree Plug-insFigure 1. Open Source model.
4 Figure 3. Geometrically consistent tracking of horizons and faults.
5 Input Attributes: Meta-Attribute ANN O1EnergyInput Attributes:Energy, Frequency, Cube Similarity Continuity, Dip Var., Azimuth Var., Absorption, Curvature, ..Interpreter’sKnowledgeMeta-AttributeANNFigure 4. “Meta-attribute” concept. Multiple attributes and interpreter’s knowledge are combined by a neural network to give the optimal view of the object of interest. In this case a salt dome.
6 Object detection Inversion Filtering Pattern recognition Rock Chimneys& dGB pluginsSalt… and turbidites, 4D bodies, ….FaultsObjectdetectionRockpropertiesInversionBeforeAfter dip-steered median filterFilteringFacies / ChannelsPatternrecognitionFigure 5. Application examples of dip-steering and neural network plugins to OpendTect.
7 abcFigure 6. Generating TheChimneyCube®: a) Picked examples b) neural network performance graphs (RMS error vs. training cycles, percentage mis-classification and input attributes colored to indicate relative importance, with red being most important) and c) overlay of chimney “probability” on seismic data.