Pipeline magnetic flux leakage image super-resolution reconstruction and defect detection method and system
By combining differentiated normalization and regional adaptive loss function, the contradiction between signal fidelity and edge supervision in super-resolution reconstruction of pipeline magnetic flux leakage images is resolved, achieving high-precision defect detection results and adapting to industrial inspection needs.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENYANG UNIVERSITY OF TECHNOLOGY
- Filing Date
- 2026-04-17
- Publication Date
- 2026-07-17
AI Technical Summary
Existing super-resolution reconstruction schemes for pipeline magnetic flux leakage images cannot simultaneously ensure signal fidelity in real-world scenarios and meet the edge supervision requirements during the training phase. This results in insufficient image contrast or amplified background noise, loss of edge gradient information, and an inability to meet the high-precision requirements of industrial inspection.
A differentiated normalization strategy and a region-adaptive loss function are adopted. The low-resolution image is globally normalized and the high-resolution image is locally normalized through differentiated normalization processing. Combined with the physical prior characteristics of magnetic flux leakage, the image is automatically divided into background, defect subject and edge region. Differential weight constraints are applied to construct a linkage mechanism between super-resolution reconstruction and downstream defect detection.
It achieves image reconstruction with smooth background and no artifacts, and clear and sharp defect edges, which significantly improves the accuracy of pipeline defect detection and meets the high precision and high reliability requirements of industrial inspection.
Smart Images
Figure CN122415490A_ABST