Multi-frame stacked point cloud super-resolution system for three-dimensional reconstruction

CN121767571BActive Publication Date: 2026-06-09WUHAN RULER TECH CO LTD

Patent Information

Authority / Receiving Office: CN · China
Patent Type: Patents(China)
Current Assignee / Owner: WUHAN RULER TECH CO LTD
Filing Date: 2026-03-03
Publication Date: 2026-06-09

AI Technical Summary

Technical Problem

In 3D reconstruction, existing technologies struggle to effectively preserve complementary sampling points during multi-frame super-resolution of point clouds, resulting in insufficient resolution improvement and even the smoothing out of edges and microstructures.

Method used

By generating a multi-frame point cloud set and a point-level confidence set, pose registration and complementary sampling labels are performed, and complementary point sets, outlier point sets and sets of points to be corrected are output. These are then overlaid, fused, and verified for consistency to generate super-resolution point cloud results.

Benefits of technology

It achieves cleaner, denser point clouds with more continuous point distribution in detailed areas such as edges and grooves, reducing the number of rework scans. The output super-resolution point cloud results are suitable for direct use in dimensional inspection and defect observation.

✦ Generated by Eureka AI based on patent content.

Smart Images

Figure CN121767571B_ABST

Patent Text Reader

Abstract

The application discloses a multi-frame superimposed point cloud super-resolution system for three-dimensional reconstruction, and particularly relates to the field of structured light three-dimensional scanning, and is used for solving the problem of thick shell and pseudo detail caused by the mixing and fusion of abnormal points and geometric conflicts in repeated scanning superposition; a plurality of point cloud sets of a same target are generated by collecting a plurality of frames of structured light point clouds, and a point-level confidence set is synchronously generated; a pose transformation result is obtained by registering the plurality of point cloud sets, a unified coordinate point cloud set is generated, and a point-level registration residual set is generated; a complementary sampling label is generated according to a stripe phase recurrence degree parameter and a line-of-sight interlayer separation parameter on the unified coordinate point cloud set, and a complementary point set, an abnormal point set and a to-be-corrected point set are output; the super-resolution point cloud result is obtained by performing attachment and restricted fusion according to the complementary sampling label; and the point-level confidence set and the complementary sampling label are updated by back projection consistency verification, and the verified and constrained super-resolution point cloud result is output.

Need to check novelty before this filing date? Find Prior Art