인문학
사회과학
자연과학
공학
의약학
농수해양학
예술체육학
복합학
DBpia AI
AI 에이전트
AI 아이디어
AI 리더
AI 뷰어
Citeasy
크롬 서지관리 다운로드
지원사업
학술연구/단체지원/교육 등 연구자 활동을 지속하도록 DBpia가 지원하고 있어요.
커뮤니티
연구자들이 자신의 연구와 전문성을 널리 알리고, 새로운 협력의 기회를 만들 수 있는 네트워킹 공간이에요.
논문 기본 정보
- 자료유형
- 학술저널
- 저자정보
- 발행연도
- 2026.6
- 수록면
- 873 - 879 (7page)
- DOI
- 10.5302/J.ICROS.2026.26.0062
이용수
초록· 키워드
Corrosion detection based on red - green - blue (RGB) images is highly sensitive to illumination variations and color ambiguities, limiting its reliability in practical environments. Although multispectral imaging provides material-dependent spectral cues, conventional multispectral systems are typically expensive and unsuitable for real-time deployment. This study presents the design of a low-cost multispectral camera system for real-time corrosion detection. The proposed system employs a depth-aware multicamera configuration that simultaneously captures four narrow-band spectral images. Through multicamera calibration and depth-guided viewpoint alignment, the captured images are accurately registered into a unified multispectral representation. By leveraging relative spectral response differences across wavelengths instead of absolute reflectance estimation, the system achieves robust corrosion discrimination under varying illumination conditions. To validate the effectiveness of the proposed camera design, a lightweight semantic segmentation model based on SegFormer was trained using the aligned four-channel multispectral images to perform pixel-wise corrosion classification into three severity levels.
Experiments on a real-world multispectral corrosion dataset demonstrate stable segmentation performance in near-range scenes and improved robustness against illumination changes and color similarity compared with the performance of RGB-based approaches. The results confirm that the proposed multispectral camera design enables practical and real-time corrosion detection while maintaining low system cost and computational efficiency.
Experiments on a real-world multispectral corrosion dataset demonstrate stable segmentation performance in near-range scenes and improved robustness against illumination changes and color similarity compared with the performance of RGB-based approaches. The results confirm that the proposed multispectral camera design enables practical and real-time corrosion detection while maintaining low system cost and computational efficiency.
#corrosion detection
#multispectral imaging
#multi-camera registration
#semantic segmentation
#depth-based alignment
상세정보 수정요청해당 페이지 내 제목·저자·목차·페이지정보가 잘못된 경우 알려주세요!
목차
- Abstract
- I. 서론
- II. 다중 분광 영상 획득 및 처리 시스템 설계
- III. 부식 검출 데이터셋 구축 및 성능 평가
- IV. 결론
- REFERENCES
참고문헌
참고문헌 신청최근 본 자료
