1 / 37

Sipuikinene Miguel Angelo Marcílio Castro de Matos marcilio@ou matos.eng.br

Integrated seismic texture segmentation and clustering analysis to improve delineation of reservoir geometry. Sipuikinene Miguel Angelo Marcílio Castro de Matos marcilio@ou.edu www.matos.eng.br Kurt J. Marfurt. Summary. Introduction Grey Level Co-occurrence Matrix – GLCM

feryal
Download Presentation

Sipuikinene Miguel Angelo Marcílio Castro de Matos marcilio@ou matos.eng.br

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Integrated seismic texture segmentation and clustering analysis to improve delineation of reservoir geometry Sipuikinene Miguel Angelo Marcílio Castro de Matos marcilio@ou.eduwww.matos.eng.br Kurt J. Marfurt

  2. Summary • Introduction • Grey Level Co-occurrence Matrix – GLCM • Application to a photographic of an outcrop image • Application to 3D seismic data • Conclusions

  3. Textures • Small scale repeated spatial patterns • Visually analogous to tactile sensations - rough, silky, bumpy • Routinely used in remote sensing 7x7 424x556 256x256 There are some ways to segment different textures in an image. We used Grey Level Co-ocurrence Matrix, GLCM, proposed by Haralick et al., 1973, in this work.

  4. Convergent facies • Chaotic facies Internal seismic facies configurations • Draping facies (Prather et al., 1998)

  5. Seismic profile across Neogene carbonate platform margin showing different seismic facies. Platform margins are characterized by mounded to prograding geometries. Basin center and platform interior shelf areas have continuous to discontinuous parallel reflectivity. Trajectory of the platform margin through time is shown by the dotted black line. (Sarg and Schuelke, 2003)

  6. Contrast Seismic amplitude Randomness Homogeniety 0.0 1.0 Textural attribute volumes (Gao, 2007)

  7. Summary • Introduction • Grey Level Co-occurrence Matrix – GLCM • Application to a photographic of an outcrop image • Application to 3D seismic data • Conclusions

  8. The Grey Level Co-occurrence Matrix, P GLCM is a tabulation of how often different combinations of pixel brightness values (grey levels) occur in an image We will use "second order" texture calculations. • dpq = quantized integer value of scaled seismic data at grid (p,q)

  9. Amplitude 1 5 9 5 4 To pixel To pixel How GLCM is evaluated 3 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 0o GLCM GLCM 3 90o From pixel From pixel 6 4 4 3 7 2 8 5 9 1 8 2 6 Contrast: 0 Correlation: NaN Energy: 1 Homogeneity: 1 Contrast: 0 Correlation: NaN Energy: 1 Homogeneity: 1 7 3 4 6 7 To pixel To pixel 8 9 GLCM 45o GLCM From pixel From pixel 135o 8 Contrast: 0 Correlation: NaN Energy: 1 Homogeneity: 1 Contrast: 0 Correlation: NaN Energy: 1 Homogeneity: 1 7 6 5 Samples

  10. GLCM (texture) attributes Contrast L2 norm of similarity Dissimilarity L1 norm of similarity Homogeneity Inverse norm of similarity Energy A measure of smoothness Returns a measure of the intensity contrast between a pixel and its neighbor over the whole image. Contrast is 0 for a constant image. Returns a value that measures the closeness of the distribution of elements in the GLCM to the GLCM diagonal. Homogeneity is 1 for a diagonal GLCM. Returns the sum of squared elements in the GLCM. Energy is 1 for a constant image. Entropy A measure of disorderliness (or roughness) Mean Correlation Returns a measure of how correlated a pixel is to its neighbor over the whole image. Correlation is 1 or -1 for a perfectly positively or negatively correlated image. Correlation is NaN for a constant image.

  11. Amplitude 1 5 9 5 How GLCM is evaluated 4 To pixel To pixel 5 5 5 3 5 1 5 5 5 2 5 2 5 5 0o GLCM GLCM 5 3 5 5 5 5 From pixel From pixel 90o 6 4 5 4 5 5 5 5 7 3 8 2 5 5 5 5 5 5 1 9 8 2 Contrast: 0 Correlation: NaN Energy: 1 Homogeneity: 1 Contrast: 0 Correlation: NaN Energy: 1 Homogeneity: 1 6 7 3 4 6 7 To pixel To pixel 8 9 GLCM 45o GLCM From pixel From pixel 135o 8 Contrast: 0 Correlation: NaN Energy: 1 Homogeneity: 1 Contrast: 0 Correlation: NaN Energy: 1 Homogeneity: 1 7 6 5 Samples

  12. How GLCM is evaluated To pixel To pixel 2 3 7 6 8 0o GLCM GLCM From pixel From pixel 3 9 2 5 2 90o 4 6 5 4 6 7 1 8 1 9 Contrast: 19.1500 Correlation: -0.4496 Energy: 0.0600 Homogeneity: 0.2849 Contrast: 17.9000 Correlation: -0.2873 Energy: 0.0500 Homogeneity: 0.2641 9 8 4 7 1 To pixel To pixel GLCM 45o GLCM From pixel From pixel 135o Contrast: 13.5000 Correlation: 0.0046 Energy: 0.0625 Homogeneity: 0.3872 Contrast: 10.7500 Correlation: 0.2122 Energy: 0.0625 Homogeneity: 0.3495

  13. How GLCM is evaluated To pixel To pixel 0o 5 6 3 4 6 GLCM GLCM From pixel From pixel 90o 7 3 4 5 7 4 6 5 6 7 6 3 5 7 5 Contrast: 4.2500 Correlation: -0.1046 Energy: 0.0850 Homogeneity: 0.3850 Contrast: 4.0500 Correlation: -0.0309 Energy: 0.0900 Homogeneity: 0.4333 3 4 7 4 3 To pixel To pixel GLCM 45o GLCM From pixel From pixel 135o Contrast: 0.9375 Correlation: 0.7622 Energy: 0.0938 Homogeneity: 0.6979 Contrast: 4.8750 Correlation: -0.3000 Energy: 0.0938 Homogeneity: 0.3635

  14. How GLCM is evaluated To pixel To pixel 0o GLCM GLCM From pixel From pixel 90o Contrast: 4 Correlation: -1 Energy: 0.5000 Homogeneity: 0.3333 Contrast: 0 Correlation: +1.0 Energy: 0.5200 Homogeneity: 1 4 6 4 6 4 To pixel 4 6 4 6 4 To pixel 4 6 4 6 4 GLCM 4 6 4 6 4 45o GLCM From pixel From pixel 135o 4 6 4 6 4 Contrast: 4 Correlation: -1 Energy: 0.5000 Homogeneity: 0.3333 Contrast: 4 Correlation: -1 Energy: 0.5000 Homogeneity: 0.3333

  15. How GLCM is evaluated To pixel To pixel 1 1 1 1 1 9 9 9 9 9 1 1 1 1 1 9 9 9 9 9 1 1 1 9 9 0o GLCM GLCM From pixel From pixel 90o Contrast: 64 Correlation: -1 Energy: 0.5000 Homogeneity: 0.1111 Contrast: 64 Correlation: -1 Energy: 0.500 Homogeneity: 0.111 To pixel To pixel GLCM 45o GLCM From pixel From pixel 135o Contrast: 0 Correlation: 1 Energy: 0.5000 Homogeneity: 1 Contrast: 0 Correlation: 1 Energy: 0.5000 Homogeneity: 1

  16. Summary • Introduction • Grey Level Co-occurrence Matrix – GLCM • Application to a photographic of an outcrop image • Application to 3D seismic data • Conclusions

  17. Example : GLCM Matrices and attributes from a photo 5x5 pixels 256 gray levels and 4 attributes Outcrop image of Monongahela Group, Pittsburgh Formation (www.geology.pitt.edu)

  18. GLCM contrastevaluated along different directions 00 450 a) b) 256 0 c) d) 1350 900

  19. GLCM contrastevaluated with different running windows a) b) c) 3X3 running window 11X11 running window 21X21 running window d) e) 31X31 running window 30X30 running window

  20. 3D GLCM computation workflow Variance 3D dip estimates Mean Correlation Extract dipping window of seismic data at every sample Homogeneity Compute GLCM Attributes Classify using SOM 2-D Color Table Construct GLCM Energy Dissimilarity Contrast 3D seismic amplitude or attributes Entropy

  21. 3D GLCM Workflow

  22. a) SOM analysis GLCM contrastevaluated with different running windows contrast attributes computed along 900 b) d) 256 0 homogeneity computed along 900 c) e) dissimilarity also computed along 900

  23. Summary • Introduction • Grey Level Co-occurrence Matrix – GLCM • Application to a photographic of an outcrop image • Application to 3D seismic data • Conclusions

  24. Osage County, OK, USA • The goal here is to map subtle, thin-bed channels that form economic gas and oil reservoirs. The study was applied to a reservoir interval just below the interpreted horizon.

  25. Oswego time-structure map Incised channel

  26. Energy-ratio coherence Incised channel

  27. Dip magnitude Incised channel

  28. GLCM: Homogeneity Incised channel

  29. GLCM: Contrast Incised channel

  30. GLCM: Mean Incised channel

  31. GLCM: Dissimilarity Incised channel

  32. GLCM: Entropy Incised channel

  33. GLCM: Energy Incised channel

  34. SOM 2D colormap 4000 ft A A’ SOM 2D colorbar

  35. Summary • Introduction • Grey Level Co-occurrence Matrix – GLCM • Application to a photographic of an outcrop image • Application to 3D seismic data • Conclusions

  36. Conclusions • 2D seismic stratigraphy is based on human interpreter identification of subtle textures, such as onlap, offlap, unconformities, hummocky clinoforms, and parallelism. With the aid of attributes, 3D seismic geomorphology extends these concepts to volumetric data. • GLCM technology is a preliminary attempt at quantifying these relationships for further analysis using computer vision. • Texture attributes hold significant promise in quantifying geological features such as mass complex transport, amalgamated channels, and dewatering features that exhibit a distinct lateral pattern beyond simple edges.

  37. Acknowledgements Thanks to the Osage Nation for providing a license to the seismic data volume used in this paper. Thanks also to the industry sponsors of the OU Attribute-Assisted Seismic Processing and Interpretation (AASPI) consortium. Attribute-Assisted Seismic Processing and Interpretation http://geology.ou.edu/aaspi/ Thank you for your attention!!! marcilio@ou.eduwww.matos.eng.br

More Related