As seismic facies are of great importance to help understand the depositional structure and properties of a reservoir, the classification of seismic facies is becoming increasingly essential when using seismic data especially facing complex reservoir targets. Whereas different from normal classification problems which just use attributes, the seismic facies classification involves both spatial coordinates and seismic attributes. Most of time, seismic attributes have corresponding spatial locations. So when solving this kind of classification problem, we have to consider the spatial structure. However, the commonly use methods like SVM usually just use the features, i.e. the attributes which often leads to bad spatial structure in classification results. In this paper, we propose a new attribute-spatial classification method which aims to use attributes and spatial information at the same time to achieve better classification results. Different from normal classification techniques, we use segmentation way to do classification. In addition, we use MRF to estimate probability distribution before segmentation process. And we use probability distribution to improve the performance of our method a lot. We then apply our method into seismic facies classification. We testify the efficiency of our method by comparing our seismic facies classification results with manual classification results of experienced stratigraphic interpreters. According to experiment results, our method can ensure spatial continuity very well with more homogeneous parts and the classification accuracy is also much higher than that of SVM.
[1]
B. P. West,et al.
Interactive seismic facies classification using textural attributes and neural networks
,
2002
.
[2]
M. Matos,et al.
Unsupervised seismic facies analysis using wavelet transform and self-organizing maps
,
2007
.
[3]
Frank Y. Shih,et al.
Automatic seeded region growing for color image segmentation
,
2005,
Image Vis. Comput..
[4]
Rolf Adams,et al.
Seeded Region Growing
,
1994,
IEEE Trans. Pattern Anal. Mach. Intell..
[5]
Babak Nadjar,et al.
Seismic Facies Classification Using 2-D And 3-D Multi-attribute Hierarchical Clustering Algorithms
,
2011
.
[6]
Donald Geman,et al.
Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration of Images
,
1984,
IEEE Transactions on Pattern Analysis and Machine Intelligence.
[7]
Mrinal K. Sen,et al.
Artificial immune based self organizing maps for seismic facies analysis
,
2012
.
[8]
Zoltan Kato,et al.
A Markov random field image segmentation model for color textured images
,
2006,
Image Vis. Comput..
[9]
C. Cattani,et al.
Markov Models for Image Labeling
,
2012
.
[10]
M. Nafi Toksöz,et al.
Seismic facies classification and identification by competitive neural networks
,
2003
.