Hyperspectral environmental illumination maps for outdoor and indoor scenes

This repository contains a dataset of hyperspectral illumination maps collected from 6 outdoor and 4 indoor scenes.

If you use this dataset in your research, please cite:

Takuma Morimoto, João M. M. Linhares, Sérgio M. C. Nascimento, and Hannah E. Smithson, “How many surfaces can you distinguish by color? Real environmental lighting increases discriminability of surface colors,” Optics Express (in press).

Technical details about data acquisition are described in:

Takuma Morimoto, Sho Kishigami, João M.M. Linhares, Sérgio M.C. Nascimento, and Hannah E. Smithson, “Hyperspectral environmental illumination maps: characterizing directional spectral variation in natural environments,” Optics Express, 27, 22, 32277 - 32293. (2019). https://doi.org/10.1364/OE.27.032277

Each file includes the following formats:

1. png: RGB image for visualization.
   
   
2. mat: Hyperspectral image with wavelengths from 400 nm to 700 nm in 10 nm steps. Each pixel value represents spectral radiance in W m−2 sr−1 nm−1. The image and wavelength range are stored in the variables ‘radiance’ and ‘wls’, respectively.

The images have an average spatial resolution of 1019 (height) × 2035 (width) across 10 scenes.

File names follow the format X_sceneY, where X is the scene type (outdoor or indoor) and Y is the scene number.

For Python users, the mat file can be loaded using e.g. scipy.io.loadmat. More information: https://docs.scipy.org/doc/scipy/reference/generated/scipy.io.loadmat.html

Note: To use for hyperspectral renderings (e.g., Mitsuba), convert the hyperspectral image to the OpenEXR format.