Seismic entangled patterns analyzed via multiresolution decomposition
Abstract. This article explores a method for distinguishing entangled coherent structures embedded in geophysical images. The original image is decomposed in a series of j-scale-images using multiresolution decomposition. To improve the image processing analysis each j-image is divided in l-spacial regions generating set of (j, l)-regions. At each (j, l)-region we apply a continuous wavelet transform to evaluate Eν, the spectrum of energy. Eν has two maxima in the original data. Otherwise, at each scale Eν hast typically one peak. The localization of the peaks changes according to the (j, l)-region. The intensity of the peaks is linked with the presence of coherent structures, or patterns, at the respective (j, l)-region. The method is successfully applied to distinguish, in scale and region, the ground roll noise from the relevant geologic information in the signal.
