% -------------------------------------------------------------------------
27.Mar.2017
last modified 15.Dec.2017
Copyright (c) 2017, Arash Mirhashemi
% -------------------------------------------------------------------------

This folder contains the SpecTex database and the associated files. SpecTex is the Hyperspectral Texture Image Database of textile samples. The files in folder are:
SpecTex_cubes_05nm.zip = the zip file that contains the SpecTex image cubes (with 5nm sampling interval)
SpecTex_cubes_10nm.zip = the zip file that contains the SpecTex image cubes (with downsampled 10nm sampling interval)
SpecTex_spectra.zip  = the sampled spectra from SpecTex database
m_files.zip                = the provided utility Matlab codes

% -------------------------------------------------------------------------

1–SpecTex Database
The database contains sixty samples T01 to T60. Each sample is saved as a TIFF (Tagged Image File Format) image. The TIFF files are multi-image files, where the first image is an 8-bit simulated RGB presentation of the spectral cube under D65 illumination, and the rest are the bands of the spectral cube, saved as grayscale images. The data is in 32-bit floating point format and represents the reflectance values at each wavelength and spatial location.

The TIFF files are compliant with the standard TIFF specifications, however, for ease of access, reader/writer Matlab functions are provided for loading the data in Matlab's workspace, and saving spectral cubes with the proposed TIFF-stack data architecture. An example script is also provided to show the usage syntax. 

% -------------------------------------------------------------------------

2–SpecTex sample spectra
The SpecTex_spectra.mat contains two sets of spectral signatures from the database. Both sets are sampled by a random grid with uniform distribution in the CIEL*a*b* color space. The first set includes a total of 178,684 spectra that come from sampling each class separately. The second set contains 10,000 spectra which are achieved by applying a similar sampling method in above to the first set. Therefor, the second set represents the whole gamut of the SpecTex dataset. For further information on the data collection method and the usage syntax, refer to the comments in (example_spectex.m).

% -------------------------------------------------------------------------

3–List of the provided Matlab functions:
example_spectex.m = an example code to show the usage syntax of the rest of the functions
t_tif_read.m = reading the TIFF-stack data structure of the SpecTex image cubes into Matlab workspace
t_tif_write.m = writing an image from Matlab workspace to a TIFF file with the proposed TIFF-stack data structure
t_spd2sth.m = a utility function to calculate the tristimulus values of a given spectrum under any light source 
t_mom_ftr.m = a function to calculate the moment features of a spectral image as described in [1]

% -------------------------------------------------------------------------

SpecTex images were captured with a spectral line-scanning camera ImSpector V8 (Specie Spectral Imaging Ltd., Finland) with a GretagMacbeth SpectraLight III unit (X-Rite, Inc., USA) light source with D65 standard illuminant simulator. A Navitar Zoom 7000 objective lens (Navitar, Inc., USA) with manual focus collected and guided the light to the camera. A moving table with a stepper motor unit performed the push-broom scanning. The light source lit the sample from above with 45 degrees angle and the camera points perpendicularly to the sample. 

Samples were collected in two batches. The first batch is from sample T01 to T23 and the second is from sample T24 to T60. The size of the spectral images is 640x640x39. The spectral dimension is gathered in wavelength range [400nm, 780nm], with 5nm intervals. For more details about the database refer to https://www.uef.fi/web/spectral/spectex and the related article [1].

% -------------------------------------------------------------------------

References:
[1] Mirhashemi, A. “Introducing spectral moment features in analyzing the SpecTex hyperspectral texture database” Machine Vision and Applications (2017). https://doi.org/10.1007/s00138-017-0892-9