A method for reconstructing a rail profile based on machine vision
By using a machine vision-based rail profile reconstruction method, dynamically correcting the conversion coefficients and data stitching, the problems of low efficiency and low accuracy in traditional manual reconstruction are solved, achieving efficient and accurate rail damage detection and reconstruction.
Patent Information
- Application Number
- CN202610483272.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-04-14
- Publication Date
- 2026-07-17
AI Technical Summary
Traditional manual rail profile reconstruction is inefficient and inaccurate, and cannot effectively and promptly address railway rail damage and defects, affecting the smoothness and safety of train operation.
A machine vision-based rail profile reconstruction method is adopted. By identifying defect areas, an initial transformation relationship between the profiler and the global coordinate system is established. The statistical deviation of the double profiler data in the overlapping area is calculated in real time, the transformation coefficient is dynamically corrected, and the rail surface data is reconstructed based on the profile data. A representative profile curve is fitted using a multi-curve fitting algorithm.
It improves the efficiency and accuracy of rail inspection, reduces human intervention errors, and can correct conversion parameters through dual-instrument data consistency verification even in cases of severe rail wear, ensuring reconstruction accuracy.
Smart Images
Figure CN122415472A_ABST