Added dataset

.gitattributes

@@ -57,3 +57,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 # Video files - compressed
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
+*.exr filter=lfs diff=lfs merge=lfs -text
+*.mexw64 filter=lfs diff=lfs merge=lfs -text
+*.mexmaci64 filter=lfs diff=lfs merge=lfs -text

README.md

@@ -1,3 +1,44 @@
 ---
 license: unknown
 ---
+
+# Description
+
+The `exr` folder contains a high-resolution dataset, consisting of 30 hyperspectral images covering a wide spectral range.  
+The acquisitions have 31 channels in the range 420-720 nm, with a separation of 10 nm.
+
+# Capture setup
+- Camera: Pointgrey Grasshopper 9.1MP Monochromatic (GS3-U3-91S6M-C)
+- Lens: Jenoptik UV-VIS-IR 60mm f/4 apochromatic lens
+- Filters: Liquid Crystal Tunable Filters (VariSpec VIS 400-720)
+- Light Source: Xenon Illumination (Thorlab HPLS-30-4)
+
+
+# How to read the files
+
+An example MATLAB code for reading a hyperspectral EXR file is given in
+`example.m` under `code`.
+For `exrreadchannels`, you can refer to the GitHub repository:
+<https://github.com/KAIST-VCLAB/openexr-matlab>
+
+# Credits
+
+The dataset was originally collected from:
+<http://vclab.kaist.ac.kr/siggraphasia2017p1/kaistdataset.html>
+
+If you use this dataset, please cite:
+```bibtex
+@Article{DeepCASSI:SIGA:2017,
+  author  = {Inchang Choi and Daniel S. Jeon and Giljoo Nam
+             and Diego Gutierrez and Min H. Kim},
+  title   = {High-Quality Hyperspectral Reconstruction
+             Using a Spectral Prior},
+  journal = {ACM Transactions on Graphics (Proc. SIGGRAPH Asia 2017)},
+  year    = {2017},
+  volume  = {36},
+  number  = {6},
+  pages   = {218:1-13},
+  doi     = "10.1145/3130800.3130810",
+  url     = "http://dx.doi.org/10.1145/3130800.3130810",
+}
+```

code/example.m

@@ -0,0 +1,26 @@
+%% Example code for reading hyperspectral image from EXR file
+% input: .exr file
+% Output: RGB, HSI
+%  - RGB: linear sRGB image (nrows x ncols x 3)
+%  - HSI: hyperspectral image (nrows x ncols x 31)
+
+%% 
+% file to read
+fn = '../exr/scene01_reflectance.exr';
+
+% read exr file
+% please refer to 'openexr-matlab' github repository for 'exrreadchannels'
+% https://github.com/KAIST-VCLAB/openexr-matlab
+mapObj = exrreadchannels(fn);
+
+% get linear sRGB image (RGB)
+RGB(:,:,1) = mapObj('R');
+RGB(:,:,2) = mapObj('G');
+RGB(:,:,3) = mapObj('B');
+
+% get hyperspectral image (HSI)
+wvls2b = 420:10:720;
+for i=1:length(wvls2b)
+    HSI(:,:,i) = mapObj(sprintf('w%dnm', wvls2b(i)));    
+end
+

code/exrreadchannels.m

@@ -0,0 +1,34 @@
+function exrreadchannels( filename )
+%EXRREADCHANNELS    Read the raw channel data of an OpenEXR image.
+%   M = EXRREADCHANNELS(FILENAME) reads all the channels of the OpenEXR
+%   file and returns a containers.Map object on which the keys are
+%   strings with the channel names and the values are the actual
+%   channel data.
+%
+%   M = EXRREADCHANNELS(FILENAME,C1,C2,...) reads only the channels named
+%   by the strings C1,C2,...,Cn. Also returns a containers.Map object.
+%   Note that if there is only a single channel the result is just a
+%   matrix with the value of the channel.
+%
+%   M = EXRREADCHANNELS(FILENAME, CARRAY) behaves as above, but it
+%   receives a cell array with the names of the desired channels.
+%
+%   [M1,...] = EXRREADCHANNELS(FILENAME,C1,...) reads the channels
+%   specified by the strings C1,C2,...,Cn and stores the channel data in
+%   the corresponding matrices M1,M2,...,Mn.
+%
+%   [M1,...] = EXRREADCHANNELS(FILENAME, CARRAY) behaves as above, but it
+%   receives a cell array with the names of the desired channels.
+%
+%   For all these methods it is an error to request a channel which does
+%   not exist. Use EXRINFO to get a list of the channels available for
+%   a given file.
+%
+%   Note: this implementation uses the ILM IlmImf library version 1.7.
+%
+%   See also CONTAINERS.MAP,EXRINFO,EXRREAD,TONEMAP
+
+% Edgar Velazquez-Armendariz (eva5@cs.cornell.edu)
+%
+% (The help system uses this file, but actually doing something with it
+% will employ the mex file).

code/exrreadchannels.mexmaci64

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0a2fa87240c7714f1e3535238605381bb840880d7357c4a04fd18e3455f7fa0
+size 20752

code/exrreadchannels.mexw64

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0c65e48e3d9c36ef90202361374741149b0927c915a09f5abbaf09ac3b71865
+size 840704