Image processing-based water supply pipeline defect detection method and system
By combining image processing and fluid dynamics models, the problems of low efficiency and insufficient accuracy in water supply pipeline inspection have been solved, achieving high-precision defect identification and quantitative assessment, and providing a scientific basis for water supply network renovation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA RAILWAY GUIZHOU ENG CORP LTD
- Filing Date
- 2026-05-06
- Publication Date
- 2026-07-21
AI Technical Summary
Existing methods for inspecting water supply pipelines are inefficient, subjective, difficult to quantify and evaluate, result in decreased image contrast, blurred details, large dynamic distortion, low detection accuracy, and inaccurate defect classification.
A defect detection method for water supply pipelines based on image processing is adopted, including preprocessing, multi-scale feature extraction, fluid dynamics inversion model and comprehensive quantitative evaluation. The method improves the YOLOv8 network for defect identification by using atmospheric scattering model to eliminate fog and optical flow field to correct the image. A comprehensive defect index is constructed by combining static geometric features and dynamic influence factors.
It improves image clarity and defect identification accuracy, quantifies defect levels, and provides a scientific basis for pipeline renovation decisions.
Smart Images

Figure CN122156193B_ABST