An unmanned aerial vehicle bridge crack detection method based on structure perception cross-modal fusion, electronic equipment, computer readable storage medium and program

By employing cross-modal fusion and multi-scale aggregation methods, the robustness and accuracy issues of bridge crack detection under complex lighting and viewing angle changes were addressed, achieving high-precision and stable crack detection.

CN122134725BActive Publication Date: 2026-07-24CHANGSHU INSTITUTE OF TECHNOLOGY
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
CHANGSHU INSTITUTE OF TECHNOLOGY
Filing Date
2026-05-06
Publication Date
2026-07-24

AI Technical Summary

Technical Problem

Existing bridge crack detection methods are not robust under complex lighting and viewing angle changes, making it difficult to maintain the topological continuity of cracks, and it is difficult to balance accuracy and real-time performance.

Method used

A structure-aware cross-modal fusion method is adopted. Visible light and infrared images are acquired, spatially and temporally aligned, feature representations are extracted, and illumination perception weights, structural consistency weights, and difference compensation weights are generated for weighted fusion. Combined with bidirectional long-range dependency modeling and multi-scale aggregation, crack detection is achieved.

Benefits of technology

It improves robustness under complex lighting conditions, maintains the continuous structure of cracks, adapts to changes in the drone's perspective, and increases the detection rate and accuracy of small cracks.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122134725B_ABST
    Figure CN122134725B_ABST
Patent Text Reader

Abstract

The application discloses a kind of unmanned aerial vehicle bridge crack detection methods based on structure perception cross-modal fusion, electronic equipment, computer readable storage medium and program, including obtaining bridge surface visible light image and infrared image, form space, time alignment multi-modal data pair, extract features to obtain enhanced multi-modal feature map, calculate illumination perception weight, structure consistency weight and difference compensation weight and be fused to obtain fusion feature, based on state updating sequence modeling mode bidirectional long-range dependence modeling is carried out to the fusion feature converted into time sequence representation, reconstruct to obtain global modeling feature, then hierarchical refining weighted aggregation obtains multi-scale aggregation feature, finally complete crack positioning, structure segmentation, center line extraction and width calculation, output result.The application can reduce the fracture of crack feature, improve small crack detection rate, complex illumination robustness and adapt to unmanned aerial vehicle view angle change, and detection stability is good.
Need to check novelty before this filing date? Find Prior Art