Seismic data processing on GPU

by
Won-Ki Jeong

Advised by
Ross Whitaker

Analyzing seismic data of the earth^Òs subsurface is critical to find oil and gas, and image processing techniques play an important role in this process. In this study, we propose a new anisotropic diffusion algorithm based on the directional analysis to extract seismic attributes. Local geometric informations, like gradients, are robustly calculated using structure tensor analysis, and the diffusion is performed along the structural directions of the data. By doing this, we can remove noises on the data while enhancing the seismic features like faults and channels. We also show how the current graphics hardware can be used to solve the computationally demanding problems like non-linear image processing efficiently.