A method, device, equipment and medium for detecting defects of an insulator
By analyzing the edge contours of insulator images and the edge structure data of the skirts, a defect level numerical value is constructed, which solves the problem of unstable defect region boundaries in traditional detection methods and realizes the coherence and consistency of defect detection assessment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- PINGXIANG HUARUI ELECTRIC PORCELAIN APPLIANCE CO LTD
- Filing Date
- 2026-04-21
- Publication Date
- 2026-07-17
AI Technical Summary
Traditional insulator defect detection methods are susceptible to fluctuations in acquisition height, deviations in observation angle, and component obstruction, which can lead to distortion of defect area boundaries, difficulty in consistently judging the degree of defects, insufficient consistency in inspection results, and easy discrepancies in judgments during the review process, thereby weakening the reliability of equipment condition assessment and the pertinence of early warning.
By extracting the number of visible pixels on the edge contour of the insulator image, calculating the gradient change sequence of the occlusion rate, analyzing the edge structure data of the umbrella skirt, and combining the pixel gradient direction vector and the direction perturbation path, a closed missing region is constructed, and the defect level value is calculated.
It enhances the consistency of defect localization, strengthens the stability of defect boundary extraction and the objectivity of grade classification, and improves the consistency of condition assessment.
Smart Images
Figure CN122415501A_ABST