Spectral Imaging Using Basis Lights

The spectral reflectance (SR) of objects provides innate information about material properties that have proven useful in applications such as classification, synthetic relighting, and medical imaging to name a few. General approaches to hyperspectral (HS) imaging include brute force capture of narrowband image stacks at consecutive wavelengths, use of conventional cameras with computational techniques for estimation of SR, use of specialized optics or filters, and active lighting. In particular, active lighting methods are fast and accurate. However, to our knowledge, past active lighting methods have not shown their illuminants to be optimal for SR recovery. In this paper, we propose a new HS imaging method that utilizes optimally derived active illuminants in conjunction with a standard monochrome camera. We begin by showing mathematically, what kinds of illuminants would provide optimal SR recovery using the well-known observation that SR can be compactly represented using basis functions. We then demonstrate that such optimal illuminants can be well approximated in reality and show they can be used for accurate SR imaging of real scenes. In addition, we show that our method experiences little degradation even in the presence of unknown ambient light.