Aero-engine blade surface micro-crack segmentation detection method
By combining differential image processing and multi-scale, multi-directional transformation with frequency domain filtering and crack segmentation neural networks, the problem of distinguishing microcracks from background textures on the surface of aero-engine blades has been solved, achieving high-precision microcrack detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHENGDU AERONAUTIC POLYTECHNIC
- Filing Date
- 2026-06-16
- Publication Date
- 2026-07-17
AI Technical Summary
Existing image detection technologies cannot effectively distinguish between real microcracks on the surface of aero-engine blades and interference from processing textures, resulting in low segmentation accuracy.
Normalized images and Gaussian smoothed images are differencing, and dynamic low-pass masks are constructed using multi-scale, multi-directional black hat transform and Fourier transform for frequency domain filtering. Edge information is extracted using multi-scale Gaussian difference operators, and features are fused through crack segmentation neural networks to distinguish real cracks from background interference.
It significantly improves the accuracy and robustness of microcrack segmentation, effectively avoids false detection and missed detection, and enhances the sensitivity and completeness of detection.
Smart Images
Figure CN122415658A_ABST