Occluded Edge Detection in 3D Point Clouds

The patent applies preliminary action by performing automatic segmentation and filtering operations before the edge fitting process. The system pre-identifies and removes spurious points from the point cloud data, and pre-segments the data into relevant regions, so that when the edge fitting algorithm runs, it works with cleaned, pre-processed data rather than requiring manual intervention to remove errors.

The system implements self-service through automated algorithms that perform segmentation, filtering, and edge detection without user intervention. The computer automatically identifies spurious points, segments the point cloud into meaningful regions, and fits edges to the data, replacing the need for manual segmentation and scaling steps that users would otherwise have to perform.

The patent transitions from step-wise 2D representation methods to direct 3D point cloud processing. By working in the full three-dimensional space rather than projecting to 2D views, the system can accurately locate edge endpoints and follow irregular shapes without the dimensional limitations that cause overshooting or undershooting in traditional step-wise approaches.

The system replaces manual mechanical operations (user interaction to rotate views, draw shapes, and adjust selections) with automated computational algorithms. The computer automatically processes the 3D point data, identifies edge features, and fits geometric models without requiring users to manually manipulate views or draw reference shapes, thereby eliminating the precision errors associated with step-wise manual methods.

The patent applies segmentation by dividing the point cloud data into distinct regions or groups before processing. Instead of analyzing every single point individually at full density, the system segments the data into meaningful clusters or zones, allowing it to process representative samples from each segment rather than all points, thereby reducing computational energy while maintaining edge detection precision.

The system uses partial action by processing only the necessary portion of the data at full density. Rather than applying full data density processing uniformly across the entire point cloud, the system identifies and processes only the relevant regions containing edge information at high precision, while using lighter processing for other areas, thus reducing overall computational energy consumption.