Borehole Data Filtering for False Sinusoid Fracture Detections

The patent segments the fracture identification process into multiple independent filtering stages. Each stage applies specific filters (sinusoidality filter, continuity filter, intensity filter, etc.) that evaluate different characteristics of potential fractures. This segmentation allows the system to systematically eliminate false detections by checking multiple criteria rather than relying on a single detection method, thereby improving reliability while reducing false positives.

The patent implements feedback mechanisms where the results from each filtering stage are used to refine subsequent filtering. The system evaluates detected sinusoids against multiple criteria and uses the outcomes of earlier filters to adjust and inform later filtering decisions. This feedback loop enables the system to learn from previous detection results and continuously improve identification accuracy, reducing false detections while maintaining comprehensive coverage.

The patent applies partial filtering by implementing a hierarchical structure where not all filters are applied to all data uniformly. Instead, the system applies simpler, faster filters first to eliminate obvious false detections, then applies more computationally intensive filters only to candidates that pass earlier stages. This partial application of filtering reduces overall computational resource consumption while maintaining high accuracy by focusing computational effort where it is most needed.

The patent segments the computational workload into distinct, independent filtering stages that can be executed sequentially. Each stage processes data more efficiently by reducing the data volume before passing it to the next stage. This segmentation of computational tasks allows the system to manage resource consumption effectively while achieving comprehensive filtering through multiple specialized filters rather than one monolithic computationally expensive filter.