计算机科学
遥感
人工智能
计算机视觉
高光谱成像
光谱带
作者
Jiaxun Zhang,Jianxin Li
标识
DOI:10.1016/j.optlaseng.2021.106632
摘要
Spectral imaging systems using liquid crystal tunable filters (LCTFs) can be applied to color reproduction. However, the spatially-varying spectral shift caused by inherent LCTF characteristics affects the results’ accuracy. This paper proposes a spectral shift correction method employing high-chroma color patches as calibration targets, which reduces the color difference from 3.67 to 1.98 (CIEDE2000). Additionally, a method is developed for adaptively selecting wavebands for image acquisition according to the target’s pre-estimated spectral reflectance, whereby color differences are reduced by 21.2% over the conventional method with a limitation of 15 spectral images acquired. These methods avoid LCTF’s drawbacks and exploit its tunability, making it more suitable for color-relevant applications.
科研通智能强力驱动
Strongly Powered by AbleSci AI