Xiyuan Li
Research Subjects
Xiyuan Li has a total of 15 co-authors. Here are the top co-authors by number of shared publications:
- Shiling Wang
College of Metrology Measurement and Instrument, China Jiliang University, Hangzhou 310018, China
3 shared publications - Chongyang Zheng2 shared publications
- Dong Liu
Zhejiang University
2 shared publications - Lan Wu2 shared publications
- Shaojing Peng2 shared publications
- Zedong Wang2 shared publications
Latest Publications
- 4 August 2026
- Published by Elsevier in Measurement
- Vol. 290
Reconstruction and stitching method of sub-aperture images for defect detection on curved optical surfaces
- 4 August 2026
- Published by Elsevier in Measurement
- Vol. 290
Reconstruction and stitching method of sub-aperture images for defect detection on curved optical surfaces
- Open Access
- 19 May 2026
- Published by Elsevier in SSRN Electronic Journal
Reconstruction and Stitching Method of Sub-aperture Images for Defect Detection on Curved Optical Surfaces
The surface quality of curved optical components impacts optical system performance, yet their complex geometries pose challenges for defect detection. This study presents a multi-degree-of-freedom defect detection system (MDDS) integrating a multi-axis scanning stage, a microscopic imaging unit, and a data processing module. Driven by the motion stage, the imaging unit scans curved component surface, utilizing ring illumination, a CMOS camera, and a bi-telecentric lens to capture high-clarity defect images under dark-field conditions. To generate a full-aperture result, acquired images are processed using a sub-aperture reprojection algorithm (SARA) and a global weighted stitching strategy (GWSS). Under the orthographic projection model, SARA defines a projection ray for each pixel and calculates its intersection with the component surface to establish a mapping relationship. The sub-aperture images are then reprojected onto a unified reference plane, reducing stitching errors caused by viewpoint variations. Using reprojected images, GWSS employs the phase correlation method (PCM) for local registration and constructs a weighted system of equations based on normalized cross-correlation (NCC) of overlapping region. Finally, global optimal stitching coordinates are obtained via least-squares optimization. This strategy maintains both accuracy and efficiency in multi-image full-aperture stitching, even in the presence of defect-sparse or defect-free overlapping regions. Based on the system, we carried out simulations and tested on planar, spherical and aspheric components. The results show that our method can accurately reproject sub-aperture images and eliminate stitching artifacts. Our research contributes to the development of efficient, automated full-aperture defect detection, supporting the high-quality manufacturing of curved optical components.
- 19 May 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessReconstruction and Stitching Method of Sub-aperture Images for Defect Detection on Curved Optical Surfaces
The surface quality of curved optical components impacts optical system performance, yet their complex geometries pose challenges for defect detection. This study presents a multi-degree-of-freedom defect detection system (MDDS) integrating a multi-axis scanning stage, a microscopic imaging unit, and a data processing module. Driven by the motion stage, the imaging unit scans curved component surface, utilizing ring illumination, a CMOS camera, and a bi-telecentric lens to capture high-clarity defect images under dark-field conditions. To generate a full-aperture result, acquired images are processed using a sub-aperture reprojection algorithm (SARA) and a global weighted stitching strategy (GWSS). Under the orthographic projection model, SARA defines a projection ray for each pixel and calculates its intersection with the component surface to establish a mapping relationship. The sub-aperture images are then reprojected onto a unified reference plane, reducing stitching errors caused by viewpoint variations. Using reprojected images, GWSS employs the phase correlation method (PCM) for local registration and constructs a weighted system of equations based on normalized cross-correlation (NCC) of overlapping region. Finally, global optimal stitching coordinates are obtained via least-squares optimization. This strategy maintains both accuracy and efficiency in multi-image full-aperture stitching, even in the presence of defect-sparse or defect-free overlapping regions. Based on the system, we carried out simulations and tested on planar, spherical and aspheric components. The results show that our method can accurately reproject sub-aperture images and eliminate stitching artifacts. Our research contributes to the development of efficient, automated full-aperture defect detection, supporting the high-quality manufacturing of curved optical components.
- Open Access
- 13 February 2026
- Published by Elsevier in SSRN Electronic Journal
Reconstruction and Stitching method of Sub-aperture Images for Defect Detection on Curved Optical Surfaces
The surface quality of curved optical components impacts optical system performance, yet their complex geometries pose challenges for defect detection. This paper introduces a multi-degree-of-freedom defect detection system (MDDS) based on scattering dark-field imaging, integrated with a sub-aperture reprojection algorithm (SARA) and a global weighted stitching strategy (GWSS) for full-aperture inspection. SARA maps sub-aperture images onto a unified reference plane, improving the normalized cross-correlation (NCC) with the ideal projection by 62% and reducing the mean squared error (MSE) by 53%. GWSS determines optimal stitching coordinates by constructing constraint equations based on phase correlation method (PCM) and NCC in overlapping regions. Tests on planar, spherical, and aspheric components show the elimination of stitching artifacts and significant accuracy improvements, with cumulative NCC values increasing from 0.2055, 0.2179, and 2.8889 to 11.3882, 7.1567, and 5.3426, respectively.This approach enables efficient, automated full-aperture defect detection, supporting high-quality and cost-effective manufacturing of curved optical components.
- 13 February 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessReconstruction and Stitching method of Sub-aperture Images for Defect Detection on Curved Optical Surfaces
The surface quality of curved optical components impacts optical system performance, yet their complex geometries pose challenges for defect detection. This paper introduces a multi-degree-of-freedom defect detection system (MDDS) based on scattering dark-field imaging, integrated with a sub-aperture reprojection algorithm (SARA) and a global weighted stitching strategy (GWSS) for full-aperture inspection. SARA maps sub-aperture images onto a unified reference plane, improving the normalized cross-correlation (NCC) with the ideal projection by 62% and reducing the mean squared error (MSE) by 53%. GWSS determines optimal stitching coordinates by constructing constraint equations based on phase correlation method (PCM) and NCC in overlapping regions. Tests on planar, spherical, and aspheric components show the elimination of stitching artifacts and significant accuracy improvements, with cumulative NCC values increasing from 0.2055, 0.2179, and 2.8889 to 11.3882, 7.1567, and 5.3426, respectively.This approach enables efficient, automated full-aperture defect detection, supporting high-quality and cost-effective manufacturing of curved optical components.