Seismic Dip Decimation for Efficient, Accurate RGT Image Computation

The patent applies segmentation by dividing the seismic image processing into two distinct stages: (1) computing seismic dips at full resolution to capture accurate local gradient information, and (2) processing the decimated image with iterative schemes to reduce computational burden. This segmentation allows the computationally intensive iterative processing to operate on reduced data while still achieving accurate RGT images through the guidance of full-resolution dip information.

The patent implements preliminary action by computing the seismic dip image at full resolution before decimation. This preliminary computation of local gradient information provides a foundation that guides the subsequent iterative processing on decimated data, ensuring that even though the iterative scheme operates on reduced-resolution data, it benefits from the accurate dip constraints established in advance.

The patent extracts the most critical information from the full-resolution seismic image by computing the seismic dip image, which captures the local gradient information necessary for accurate horizon picking. This extracted dip information is then used to guide the iterative processing on decimated data, allowing the system to work with reduced data volumes while maintaining accuracy through the extracted gradient constraints.

The processing is segmented into full-resolution dip computation followed by decimated iterative processing. This segmentation allows the system to retain full-resolution information where it matters most (in the dip constraints) while using reduced-resolution data for the computationally intensive iterative horizon picking, thereby managing data volume for existing GPUs.