基于计算机视觉的轨道表面缺陷检测系统
By analyzing continuous images of the track surface over time, water film disturbance and crack features are separated. Texture is enhanced by reverse convergence and edge information superposition, which solves the instability problem of track detection under water film coverage after rain and improves the accuracy and consistency of detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANGHAI CONTRON INFORMATION TECH CO LTD
- Filing Date
- 2026-04-28
- Publication Date
- 2026-07-17
AI Technical Summary
When residual water film covers the track surface after rain, computer vision-based track surface defect detection systems are prone to misjudging water film disturbances as structural cracks, leading to unstable detection results, ignoring real defects and delaying their treatment, thus increasing the risk of crack propagation.
By analyzing continuous images of the track surface over time, dynamic dark texture features are extracted and a hierarchical recognition path is formed. Water film disturbance and crack features are separated, and the texture is stabilized by using reverse convergence and edge information superposition to restore the true crack features.
It effectively avoids misjudgment due to water film disturbance, improves the stability and accuracy of defect detection, reduces the probability of missed detection, and ensures the consistency of identification results.
Smart Images
Figure CN122115437B_ABST