Robust prestack seismic facies analysis using shearlet transform-based deep learning

Abstract One of the primary purposes seismic stratigraphy is to evaluate components layer relationships within a depositional chronology. Prestack images contain wealth information, such as variations in offset and azimuth event, naturally produce higher-resolution facies analysis results than poststack data. However, prestack data usually suffer from potential unreliability issues due low signal-to-noise ratios. As this often overlooked, present methods sometimes fail extract accurate features images, which inevitably influences results. To address issue, article provides robust data-driven technique for extracting offset-temporal via shearlet transform-based deep convolution autoencoders (STCAEs). Unlike time domain, STCAE can optimally represent at multiple scales directions through two-dimensional transform, preserves fine edges while suppressing noise images. Subsequently, are extracted manner contractive convolutional autoencoder network. We compare our method with other advanced demonstrate advantages proposed approach classifying layers