Building structure defect detection system based on machine vision
By combining dual-modal detection with visible light and depth thermal imaging data, and using dynamic cognitive kernel modules for cross-validation and iterative refinement, the problem of false alarms and missed detections in building structure defect detection systems under complex textures and uneven lighting environments has been solved. This has resulted in a highly reliable and accurate defect identification system with self-optimization capabilities.
Patent Information
- Application Number
- CN202511871014.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-12
- Publication Date
- 2026-03-03
AI Technical Summary
Existing machine vision-based building structural defect detection systems struggle to accurately identify minute structural defects in complex textures and uneven lighting environments, resulting in high false alarm and false negative rates and failing to provide reliable structural safety assessments.
The system employs a visible light camera unit and a depth thermal imaging acquisition unit to acquire dual-modal data. It combines a dynamic cognitive kernel module for cross-validation and iterative refinement, optimizes feature extraction through adaptive attention weight maps and feature selection parameters, and optimizes the model by combining an offline training library, thereby achieving accurate differentiation between real defects and pseudo-features.
It effectively reduces false alarms and missed detections, improves the reliability and accuracy of detection in complex environments, and forms an intelligent detection system with self-optimization and closed-loop verification, providing reliable technical support for building structural health monitoring.
Smart Images

Figure CN121600407A_ABST