Guide file for the database
In this database we provide full-spectrum (200 – 1000 nm) scans from natural scenes measured with a custom built hyperspectral scanner, as presented in our accompanying publication Nevala & Baden 2018 bioRxiv, and as partially used in Zimmermann, Nevala, Yoshimatsu et al. 2017 bioRxiv. The scanned images are taken from different parts of the world, including a variety of natural, urban and aquatic scenes. All data is provided under Creative Commons Attribution (CC BY) license.
Each scan consists of a 1000 point array (“pixels”) where each point comprises the full spectrum (200-1000 nm) of light as recorded with a Thorlabs CCS200/M, 200-1,000 nm spectrometer. Each folder comprises additional pre-processed data files, a photograph of the scene with (usually) the scan area indicated as well as details on the scan and spectrometer settings used.
Data file structure
Alldata_formatted: 1000x800 array, comprising 1000 800 point spectra, each starting at 200 nm in steps of 1 nm. All data is log10 compressed. Note that some arrays are missing 1 or more entries due to rare scan-failures. These can be interpolated in the Igor-scripts provided. Alongside each raw data array, ScanX and ScanY give the corresponding coordinates (in degrees) of each scan position.
Full_reconstruction: provides a series of 800 spatially interpolated images computed from the raw data. Each frame corresponds to a 1 nm window, starting from 200 nm.
Full_reconstruction_ZNormed: as above, but with each spectrum z-normalised within itself.