Seismic data interpretation using the Hough transform and principal component analysis

A genetic algorithm for filter design to enhance features in seismic images

“…To efficiently compare both the original and filtered images we decided to employ the method described in Orozco‐del‐Castillo et al . (), a PCA‐based automatic interpretation technique that yields quantifiable data. A statistical comparison using the Mann‐Whitney U ‐test showed that, while the interpretation was effective for both images, the filtered image allowed for a better interpretation than the original one.…”

“…7a), is better-suited for automatic interpretation. We decided to test both images by submitting them to the method followed in Orozco-del-Castillo et al (2011a) used for detecting salt bodies in seismic images using the principal component analysis approach. This method, originally proposed by Turk and Pentland for the detection of faces (Turk and Pentland 1991), basically consists in acquiring an initial set of images (training set) and calculating their principal components (eigenvectors of the covariance matrix formed by the data of the images in the training set) Figure 7 a) The resulting image of the convolution of Fig.…”

“…11(b) (filtered image). This means that the process described in Orozco-del-Castillo et al (2011a) for the original image responds well only when the obtained images resemble, in a very high measure, a salt body but small variations diminish the effectiveness, in contrast with the same procedure applied to the filtered image, which is less sensitive. In this sense, it can be concluded that this kernel also makes the automatic interpretation process more robust.…”

“…For example, the chosen cost function, some GA parameters, another optimization technique, or even the proposed convolution operator could be exchanged for a non-linear operation. Finally, an integral system could be constructed from both the procedure in this work and the one shown in Orozco-del-Castillo et al (2011a). The cost function could be adapted to the process used as validation, i.e., the objective could be to find those values of the convolution kernel W that maximize the contrast of the central tendencies between the sets of values that represent regions with and without the presence of salt bodies.…”

“…Results are then validated by the Principal Component Analysis (PCA) approach followed by Orozco‐del‐Castillo et al . (), whose methodology is described in the appropriate section and the Mann‐Whitney U ‐Test (Mann and Whitney ).…”