光场
领域(数学)
计算机科学
计算机图形学(图像)
计算机视觉
结构光
人工智能
数学
纯数学
作者
Zhengren Zhang,Anjun Qu,Mengran Yang,Zile Li,Tian Huang,Shuqi Chen,Hua Cheng,Shaohua Yu,Guoxing Zheng
标识
DOI:10.1002/lpor.202401120
摘要
Abstract Three‐dimensional (3D) reconstruction based on structured light (SL) has wide applications in fields such as artificial intelligence, surveying and mapping, and precision manufacturing. An ideal 3D reconstruction system features both wide field‐of‐views (FOV), high‐accuracy and high‐speed reconstruction and ultracompact volume. However, these criteria have been achieved only partially with conventional SL techniques like optical projector and differentiate optical element, which will substantially limit their practical applications. In this study, an ultracompact SL and its 3D reconstruction system is designed and experimentally demonstrated, which exhibits super‐large FOV angles up to 210° × 210°, only 3.8‰ reconstruction errors, real‐time processing speeds up to 30 fps and SL dimensions as small as 0.3 mm × 0.3 mm. Specifically, the SL is realized with a resonant geometric metasurface, which can produce tens of thousands of random dots and project them into the whole 4π free space. The metasurface enabled SL is collaborated with a binocular stereoscopic camera and advanced 3D reconstruction algorism to form a practical 3D reconstruction system, which can provide robust hardware and software support for high‐accuracy and real‐time 3D scene reconstruction.
科研通智能强力驱动
Strongly Powered by AbleSci AI