旋光法
遥感
图像分辨率
分辨率(逻辑)
环境科学
计算机科学
计算机视觉
地质学
光学
物理
人工智能
散射
作者
Junren Wen,Shuaibo Feng,Yipeng Chen,Weiming Shi,Haiqi Gao,Yu Shao,Yuchuan Shao,Yueguang Zhang,Weidong Shen,Chenying Yang
标识
DOI:10.1002/lpor.202401983
摘要
Abstract In recent years, intelligent optical sensing has seen rapid advancements driven by innovations in optoelectronics and AI‐based algorithms. Specifically, spectro‐polarimetric imaging (SPI) provides comprehensive insights by capturing both spectral and polarization information. In this paper, a spectro‐polarimetric camera is presented that combines full‐Stokes polarimetric imaging with multi‐spectral imaging, achieving full spatial resolution across the visible wavelength range. Utilizing a CMOS‐compatible micro‐filter array and a deep unfolding network (GAP‐net), the camera achieves a spatial resolution of 2016 × 2016 pixels, and successfully resolves 14.30 lines per millimeter, as verified by the USAF 1951 test chart. The camera achieves video‐rate snapshot spectral imaging for polarization‐insensitive scenarios and delivers exceptional spectral reconstruction fidelity for both static and dynamic scenes, with a remarkable fidelity of 99.53% for the standard color checker. For polarization‐sensitive scenes, the full‐Stokes parameters ( S 0 , S 1 , S 2 , S 3 ) are successfully restored, while maintaining the spectral precision. The proposed spectro‐polarimetric camera holds significant importance for high‐dimensional intelligent optical sensing and demonstrates considerable potential in remote sensing, autonomous driving, and surveillance.
科研通智能强力驱动
Strongly Powered by AbleSci AI