一种基于金相分析的压缩机铸件质量评估方法及系统
By adaptively extracting seed points through Gaussian filtering and distance transformation, and combining dual-channel screening and adaptive replanting mechanisms, the problems of seed point errors and missed pore detection in metallographic image segmentation of compressor castings are solved, achieving high accuracy and reliability in casting quality assessment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- XIAN ISE MACHINERY CO LTD
- Filing Date
- 2026-05-15
- Publication Date
- 2026-07-17
AI Technical Summary
In existing technologies for metallographic image segmentation of compressor castings, incorrect seed point marking leads to misalignment of segmentation boundaries and missed detection of pores. Fixed thresholds cannot distinguish between graphite flakes and pores, resulting in misjudgments and noise interference, which affects the accuracy of casting quality assessment.
Candidate seed points are adaptively extracted using Gaussian filtering and distance transformation, combined with dual-channel screening based on extreme value significance and boundary gradient density, and an adaptive reseeding mechanism. Accurate phase classification is achieved through watershed segmentation, eliminating noise interference and missed pore detection.
This improves the accuracy of microstructural defect segmentation in compressor castings, reduces the misjudgment rate of qualified products, and ensures the credibility and objectivity of casting quality assessment.
Smart Images
Figure CN122200204B_ABST