Construction site image perspective alignment method

By building a multi-view rendering library and optimizing occlusion robust weights, the instability problem of view alignment between construction site images and BIM models was solved, achieving efficient and accurate view alignment and reliable alignment results, thus improving the level of automation in inspections.

CN122023490BActive Publication Date: 2026-06-26HANGZHOU HAOLINK INTELLIGENT TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
HANGZHOU HAOLINK INTELLIGENT TECHNOLOGY CO LTD
Filing Date
2026-04-13
Publication Date
2026-06-26

AI Technical Summary

Technical Problem

Existing methods for aligning construction site images with BIM model perspectives are inefficient, subjective, and lack error prevention. Furthermore, alignment is unstable under conditions of occlusion, lighting changes, and structural repetition, and reliability assessments are lacking.

Method used

By acquiring stable structural features, a multi-view rendering library is constructed to perform coarse pose estimation and optimization. Combining occlusion robust weights and robust loss functions, cross-domain reprojection errors are resolved, occlusion interference is shielded, and alignment accuracy is improved.

Benefits of technology

It achieves efficient and accurate view alignment in complex environments, avoids matching failures, provides reliable alignment results and a self-correction mechanism, and improves the credibility of inspections.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122023490B_ABST
    Figure CN122023490B_ABST
Patent Text Reader

Abstract

The application relates to the technical field of building information processing, and provides a construction site image view angle alignment method, which can construct a multi-view rendering library covering typical inspection view angles according to stable structure features, avoids matching failure caused by view angle omission, and solves the unstable cross-domain alignment problem of BIM and construction site images; coarse pose estimation is carried out according to two-dimensional structure features and the multi-view rendering library, and fine pose is obtained by optimizing the coarse pose according to a cross-domain re-projection error target function; in combination with a shielding robust weight and a robust loss function, abnormal feature interference caused by temporary object shielding is effectively shielded, and the accuracy of view angle alignment is improved.
Need to check novelty before this filing date? Find Prior Art