Title:

Salt dome delineation using edge- and texture-based attributes

Author(s):

Muhammad Amir Shafiq*, Yazeed Alaudah, and Ghassan AlRegib

Center for Energy and Geo Processing (CeGP),

School of Electrical and Computer Engineering,

Georgia Institute of Technology,

Atlanta, GA, 30332 *amirshafiq@gatech.edu

Summary:

A lot of research has been done in the past to capture seismic features based on different

edge- and texture-based attributes. In this paper, we apply a phase-based edge detection

algorithm, namely phase congruency (PC), and a texture-based algorithm, namely

gradient of texture (GoT), to localize a salt dome within SEAM dataset. Phase congruency

(PC) can highlight small discontinuities in images with varying illumination and contrast

using the congruency of phase in Fourier components. PC can not only detect the subtle

variations in the image intensity but can also highlight the anomalous values to develop

a deeper understanding of post-migrated seismic data. In contrast, GoT measures the

perceptual dissimilarity of texture between two neighboring windows at each point in a

seismic image along time or depth, and crossline directions, respectively. The GoT can

effectively detect subtle variations characterized by changes in the texture of seismic data

even in the absence of strong seismic reflections. We propose an interpreter-assisted

workflow based on an attribute map obtained using either PC or GoT for computational

seismic interpretation with an application to subsurface structures delineation within

migrated seismic volumes. Experimental results show the effectiveness of PC and GoT

for salt dome delineation on the SEAM dataset.

Introduction

Salt formations beneath the Earth’s surface are impermeable and form stratigraphic traps for petroleumand gas reservoirs. Furthermore, geomodeling and exploration studies require accurate delineation ofsalt bodies. Experienced geophysicists interpret the migrated data by observing the intensity and tex-ture variations of seismic traces near different structures. The modern wide-azimuth and high-densityseismic acquisitions have aided seismic interpreters by yielding seismic data with higher quality andbetter resolution. However, these techniques have resulted in the striking growth of acquired seismicdata, which in turn are causing manual interpretation extremely time consuming and labor intensive.Therefore, there is an increased interest among researchers to automate the process of seismic interpre-tation. Over the last few decades, researchers have proposed several methods for salt dome delineation,which include edge-based detection methods by Zhou et al. (2007), Aqrawi et al. (2011), and Amin andDeriche (2015b), texture-based methods by Berthelot et al. (2013), Hegazy and AlRegib (2014), Shafiqet al. (2015b), Wang et al. (2015), and Shafiq et al. (2016b), active-contour- and level-set-based methodsby Haukas et al. (2013) and Shafiq et al. (2015a), saliency-based methods by Drissi et al. (2008) andShafiq et al. (2016a), machine-learning-based methods by Guillen et al. (2015), Qi et al. (2015), andAmin and Deriche (2015a), and other different image processing techniques by Lomask et al. (2004),Lomask et al. (2007), Halpert et al. (2009), and Wu (2016).

Phase Congruency (PC) originally proposed by Morrone et al. (1986) and modified by Kovesi (1999) isan edge detection approach based on the observation that the pixels along edges have Fourier compo-nents that are maximally in phase. PC varies between 0 and 1 corresponding to no and perfect phasecongruency, respectively. PC is a dimensionless quantity that is not affected by the changes in imageillumination and contrast, thereby making it practicable for images with dominating and inconspicuousedges. In seismic interpretation, Russell et al. (2010) and Kovesi et al. (2012) showed the efficacy of PCby detecting seismic discontinuities and velocity anomalies in migrated seismic volumes, respectively.Shafiq et al. (2017) applied PC for segmenting geophysical images that have chaotic structures and vary-ing textures such as salt domes and showed its efficacy on seismic dataset from the North Sea, F3 block.In contrast, Gradient of Texture (GoT), a recently proposed method by Shafiq et al. (2016b), is a texture-based method, which defines the perceptual dissimilarity of the texture between two neighboring cubesthat share a square face centered around each voxel in a migrated 3D volume. The GoT is computed inthe spatial domain, whereas the texture dissimilarity function, which is calculated along time, crossline,and inline directions of given seismic volume V is calculated in frequency domain. Shafiq et al. (2016b)showed the efficacy of GoT by segmenting salt domes from a real dataset acquired from the NetherlandsF3 block in the North Sea.

In this paper, we present an approach for salt dome delineation based on different seismic attributes suchas phase congruency and GoT. The proposed workflow leads to improved salt dome delineation by ac-curately and efficiently detecting the presence of strong and weak seismic reflections. The experimentalresults on the SEAM dataset (Fehler and Keliher, 2011) show the effectiveness of the proposed workflowfor salt dome delineation.

Proposed Workflow

We propose an interpreter-assisted workflow for the computational seismic interpretation of salt domesas shown in Figure 1. Given a 3D seismic volume V , we apply pre-processing operations such as noiseremoval and image enhancement to yield a 3D seismic volume, Vp, for better feature detection. We thencompute the attribute map (PC or GoT), which highlights the salt-dome edges and different geologicalfeatures in a seismic image. In the next step, we determine an adaptive threshold Th using Otsu’s methodto obtain a binary map of highlighted salt-dome boundaries, B, in which the white regions highlight thesalt-dome boundaries. Next, we apply morphological processing operations, if required, to close the saltbody and disconnect the noisy and non-salt regions, which have either low congruency in phase or lowtexture dissimilarity. To get rid of noise and detect a salt body S from the binary map B, we apply aregion growing method by manually selecting an initial seed point, ps, and growing it pixelwise until ithits the salt-dome boundary. The region growing method yields a binary map of the salt body, S, which

79th EAGE Conference & Exhibition 2017Paris, France, 12 – 15 June 2017

Pre-Processing

Seismic Volume

Phase CongruencyOr

Gradient of Texture

Thresholding(Otsu’s) Salt Dome

Region Growing

Interactive Editing (Optional)

V PC B SpV

Seed Point

spiB

GoT

Figure 1: The block diagram of the proposed salt-dome delineation workflow.

is superimposed on the original seismic inline section to highlight the salt dome.

Experimental Results

To show the efficacy of the proposed workflow, the results of salt dome delineation on different seismicinline sections from the SEAM dataset (Fehler and Keliher, 2011), which contains a challenging salt-dome structure, are shown in Figure 2. The results of salt dome delineation obtained using PC attribute(Kovesi, 1999) are shown in Figure 2a-c, whereas the results of GoT (Shafiq et al., 2016b), are shown inFigure 2d-f. PC yields very good delineation even in the absence of strong seismic reflection as observedin the bottom left sides and the right sides of Figure 2. Similarly, GoT also yields good delineationresults. However, GoT digresses from the observed salt-dome boundary in the absence of strong texturecontrast. The careful examination of the results show that there are very minute differences in salt domedelineation obtained by PC and GoT. However, these areas can be interactively edited by the seismicinterpreters to yield accurate results for the salt dome delineation. Finally, the delineation time requiredper slice by PC is 1.60 seconds as opposed to 84.75 seconds required by GoT. Subjective evaluationshows that the PC yields better results for salt dome delineation and is also computationally inexpensiveas compared to the GoT.

Conclusions

In this paper, we have proposed a workflow for delineating salt domes within migrated seismic volumesusing an edge-based (PC) and a texture-based (GoT) attribute. Each attribute yields a feature map, whichis adaptively thresholded to produce a binary map. A software tool interactively guided by an interpreterand the region growing method are then used to accurately detect the salt bodies within seismic volumes.Experimental results on the SEAM dataset show the effectiveness of edge- and texture-based attributemaps for salt dome delineation. The proposed workflow is suitable for segmenting seismic volumes hav-ing weak seismic reflections, varying texture, illumination, and contrast. The workflow based on the PCmap is computationally inexpensive and is expected to not only reduce the time for seismic interpretationbut also become a handy tool in the interpreter’s toolbox for delineating geological structures.

Acknowledgements

This work is supported by the Center for Energy and Geo Processing (CeGP) at Georgia Tech and KingFahd University of Petroleum and Minerals.

References

Amin, A. and Deriche, M. [2015a] A hybrid approach for salt dome detection in 2D and 3D seismicdata. In: Image Processing (ICIP), 2015 IEEE International Conference on. 2537–2541.

Amin, A. and Deriche, M. [2015b] A new approach for salt dome detection using a 3D multidirectionaledge detector. Applied Geophysics, 12(3).

Aqrawi, A.A., Boe, T.H. and Barros, S. [2011] Detecting salt domes using a dip guided 3D Sobelseismic attribute. In: Expanded Abstracts of the SEG 81st Annual Meeting. Society of ExplorationGeophysicists, 1014–1018.

Berthelot, A., Solberg, A.H. and Gelius, L.J. [2013] Texture attributes for detection of salt. Journal ofApplied Geophysics, 88, 52–69.

79th EAGE Conference & Exhibition 2017Paris, France, 12 – 15 June 2017

(a) PC: Seismic inline 1793 (d) GoT: Seismic inline 1793

(b) PC: Seismic inline 2393 (e) GoT: Seismic inline 2393

(c) PC: Seismic inline 3293 (f) GoT: Seismic inline 3293

Figure 2: The results of salt dome delineation using phase congruency (PC) and gradient of texture(GoT). Left column shows the results of PC, whereas right column shows the results of GoT.

79th EAGE Conference & Exhibition 2017Paris, France, 12 – 15 June 2017

Drissi, N., Chonavel, T. and Boucher, J. [2008] Salient features in seismic images. In: OCEANS 2008 -MTS/IEEE Kobe Techno-Ocean. 1–4.

Fehler, M. and Keliher, P. [2011] SEAM Phase 1: Challenges of Subsalt Imaging in Tertiary Basins,with Emphasis on Deepwater Gulf of Mexico. Society of Exploration Geophysicists.

Guillen, P., Larrazabal, G., Gonzalez, G., Boumber, D. and Vilalta, R. [2015] Supervised learning todetect salt body. In: SEG Technical Program Expanded Abstracts. 1826–1829.

Halpert, A.D., Clapp, R.G. and Biondi, B. [2009] Seismic image segmentation with multiple attributes.In: Expanded Abstracts of the SEG 79th Annual Meeting. Society of Exploration Geophysicists, 3700–3704.

Haukas, J., Ravndal, O.R., Fotland, B.H., Bounaim, A. and Sonneland, L. [2013] Automated salt bodyextraction from seismic data using level set method. First Break, EAGE, 31.

Hegazy, T. and AlRegib, G. [2014] Texture attributes for detecting salt bodies in seismic data. In:Expanded Abstracts of the SEG 84th Annual Meeting. Society of Exploration Geophysicists, 1455–1459.

Kovesi, P. [1999] Image Features From Phase Congruency. Videre: A Journal of Computer VisionResearch. MIT Press, 1(3).

Kovesi, P., Ben, R., Eun-Jung, H. and Jeffrey, S. [2012] Phase-Based Image Analysis of 3D SeismicData. ASEG Extended Abstracts, 1–4.

Lomask, J., Biondi, B. and Shragge, J. [2004] Image segmentation for tracking salt boundaries. In:Expanded Abstracts of the SEG 74th Annual Meeting. 2443–2446.

Lomask, J., Clapp, R.G. and Biondi, B. [2007] Application of image segmentation to tracking 3D saltboundaries. Geophysics, 72(4), P47–P56.

Morrone, M., Ross, J., Burr, D. and Owens, R. [1986] Mach bands are phase dependent. Nature 324,250–253.

Qi, J., Cahoj, M., AlAli, A., Li, L. and Marfurt, K. [2015] Segmentation of salt domes, mass transportcomplexes on 3D seismic data volumes using Kuwahara windows and multiattribute cluster analysis.In: SEG Technical Program Expanded Abstracts. 1821–1825.

Russell, B., Hampson, D. and Logel, J. [2010] Applying the phase congruency algorithm to seismic dataslices: a carbonate case study. Reservoir Geoscience and Engineering, first break, 28, 83–90.

Shafiq, M.A., Alaudah, Y., AlRegib, G. and Deriche, M. [2017] Phase Congruency for Image under-standing with applications in Computational Seismic Interpretation. In: The 42nd IEEE InternationalConference on Acoustics, Speech and Signal Processing (ICASSP), New Orleans, USA. IEEE.

Shafiq, M.A., Alshawi, T., Long, Z. and AlRegib, G. [2016a] SALSI: A new seismic attribute for saltdome detection. In: The 41st IEEE International Conference on Acoustics, Speech and Signal Pro-cessing (ICASSP).

Shafiq, M.A., Wang, Z. and Alregib, G. [2015a] Seismic interpretation of migrated data Using edge-based geodesic active contours. In: Proc. IEEE Global Conf. on Signal and Information Processing(GlobalSIP), Orlando, Florida, Dec. 14-16.

Shafiq, M.A., Wang, Z., AlRegib, G., Amin, A. and Deriche, M. [2016b] A texture-based interpretationworkflow with application to delineating salt domes. Accepted in Interpretation.

Shafiq, M.A., Wang, Z., Amin, A., Hegazy, T., Deriche, M. and AlRegib, G. [2015b] Detection of Salt-dome Boundary Surfaces in Migrated Seismic Volumes Using Gradient of Textures. In: ExpandedAbstracts of the SEG 85th Annual Meeting, New Orleans, Louisiana. 1811–1815.

Wang, Z., Hegazy, T., Long, Z. and AlRegib, G. [2015] Noise-robust detection and tracking of saltdomes in postmigrated volumes using texture, tensors, and subspace learning. Geophysics, vol. 80(6).

Wu, X. [2016] Methods to compute salt likelihoods and extract salt boundaries from 3D seismic images.GEOPHYSICS, 81(6), IM119–IM126.

Zhou, J., Zhang, Y., Chen, Z. and Li, J. [2007] Detecting boundary of salt dome in seismic data with edgedetection technique. In: Expanded Abstracts of the SEG 77th Annual Meeting. Society of ExplorationGeophysicists, 1392–1396.

79th EAGE Conference & Exhibition 2017Paris, France, 12 – 15 June 2017