Hyperspectral images of King, Magnificent, and hybrid King of Holland's Bird-of-Paradise (Cicinnurus regius, C. magnificus, and C. magnificus x C. regius)
Data files
Oct 09, 2024 version files 128.45 GB
-
CM_M_01_Back.bil
4.44 GB
-
CM_M_01_Back.bil.hdr
2.72 KB
-
CM_M_01_Face.bil
3.18 GB
-
CM_M_01_Face.bil.hdr
2.72 KB
-
CM_M_01_Front.bil
3.76 GB
-
CM_M_01_Front.bil.hdr
2.72 KB
-
CM_M_01_Left.bil
3.13 GB
-
CM_M_01_Left.bil.hdr
2.72 KB
-
CM_M_01_Right.bil
3.61 GB
-
CM_M_01_Right.bil.hdr
2.72 KB
-
CM_M_01_TBL.bil
1.91 GB
-
CM_M_01_TBL.bil.hdr
2.72 KB
-
CM_M_01_TBR.bil
2.11 GB
-
CM_M_01_TBR.bil.hdr
2.72 KB
-
CM_M_02_Back.bil
3.43 GB
-
CM_M_02_Back.bil.hdr
2.72 KB
-
CM_M_02_Face.bil
2.90 GB
-
CM_M_02_Face.bil.hdr
2.72 KB
-
CM_M_02_Front.bil
3.70 GB
-
CM_M_02_Front.bil.hdr
2.72 KB
-
CM_M_02_Left.bil
3.09 GB
-
CM_M_02_Left.bil.hdr
2.72 KB
-
CM_M_02_Right.bil
3.04 GB
-
CM_M_02_Right.bil.hdr
2.72 KB
-
CM_M_02_TBL.bil
1.84 GB
-
CM_M_02_TBL.bil.hdr
2.72 KB
-
CM_M_02_TBR.bil
1.86 GB
-
CM_M_02_TBR.bil.hdr
2.72 KB
-
CR_M_01_Back.bil
3.46 GB
-
CR_M_01_Back.bil.hdr
2.72 KB
-
CR_M_01_Face.bil
2.71 GB
-
CR_M_01_Face.bil.hdr
2.72 KB
-
CR_M_01_Front.bil
3.45 GB
-
CR_M_01_Front.bil.hdr
2.72 KB
-
CR_M_01_Left.bil
3.37 GB
-
CR_M_01_Left.bil.hdr
2.72 KB
-
CR_M_01_Right.bil
3.34 GB
-
CR_M_01_Right.bil.hdr
2.72 KB
-
CR_M_01_TBLR.bil
2.35 GB
-
CR_M_01_TBLR.bil.hdr
2.72 KB
-
CR_M_02_Back.bil
3.53 GB
-
CR_M_02_Back.bil.hdr
2.72 KB
-
CR_M_02_Face.bil
2.52 GB
-
CR_M_02_Face.bil.hdr
2.72 KB
-
CR_M_02_Front.bil
3.28 GB
-
CR_M_02_Front.bil.hdr
2.72 KB
-
CR_M_02_Left.bil
2.67 GB
-
CR_M_02_Left.bil.hdr
2.72 KB
-
CR_M_02_Right.bil
3.11 GB
-
CR_M_02_Right.bil.hdr
2.72 KB
-
CR_M_02_TBL.bil
2.17 GB
-
CR_M_02_TBL.bil.hdr
2.72 KB
-
CR_M_02_TBR.bil
2.33 GB
-
CR_M_02_TBR.bil.hdr
2.72 KB
-
Figure_dataframes.zip
58.99 MB
-
Gouldian.zip
366.31 MB
-
H_M_01_Back.bil
3.28 GB
-
H_M_01_Back.bil.hdr
2.72 KB
-
H_M_01_Face.bil
3.27 GB
-
H_M_01_Face.bil.hdr
2.72 KB
-
H_M_01_Front.bil
3.62 GB
-
H_M_01_Front.bil.hdr
2.72 KB
-
H_M_01_Left.bil
2.65 GB
-
H_M_01_Left.bil.hdr
2.72 KB
-
H_M_01_Right.bil
2.77 GB
-
H_M_01_Right.bil.hdr
2.72 KB
-
H_M_01_TBL.bil
2.31 GB
-
H_M_01_TBL.bil.hdr
2.72 KB
-
H_M_01_TBR.bil
2.50 GB
-
H_M_01_TBR.bil.hdr
2.72 KB
-
H_M_02_Back.bil
3.21 GB
-
H_M_02_Back.bil.hdr
2.72 KB
-
H_M_02_Face.bil
2.82 GB
-
H_M_02_Face.bil.hdr
2.72 KB
-
H_M_02_Front.bil
3.58 GB
-
H_M_02_Front.bil.hdr
2.72 KB
-
H_M_02_Left.bil
3.17 GB
-
H_M_02_Left.bil.hdr
2.72 KB
-
H_M_02_Right.bil
2.94 GB
-
H_M_02_Right.bil.hdr
2.72 KB
-
H_M_02_TBLR.bil
2.61 GB
-
H_M_02_TBLR.bil.hdr
2.72 KB
-
matmask.zip
8.82 GB
-
MetaDepthNormal.zip
176.97 MB
-
README.md
4.55 KB
Abstract
Hyperspectral imaging—a technique that combines the high spectral resolution of spectrophotometry with the high spatial resolution of photography—holds great promise for the study of animal coloration. However, applications of hyperspectral imaging to questions about the ecology and evolution of animal color remain relatively rare. The approach can be expensive and unwieldy, and we lack user-friendly pipelines for capturing and analyzing hyperspectral data in the context of animal color. Fortunately, costs are decreasing and hyperspectral imagers are improving, particularly in their sensitivity to wavelengths (including ultraviolet) visible to diverse animal species. To highlight the potential of hyperspectral imaging for animal color studies, we developed a pipeline for capturing, sampling, and analyzing hyperspectral data (here, in the 325nm - 700nm range) using avian museum specimens. Specifically, we used the pipeline to characterize the plumage colors of the King Bird-of-Paradise (Cicinnurus regius), Magnificent Bird-of-Paradise (C. magnificus), and their putative hybrid, the King of Holland’s Bird-of-Paradise (C. magnificus x C. regius). We also combined hyperspectral data with 3D digital models to supplement hyperspectral images of each specimen with 3D shape information. Using visual system-independent methods, we found that many plumage patches on the hybrid King of Holland’s Bird-of-Paradise are—to varying degrees—intermediate relative to those of the parent species. This was true of both pigmentary and structurally colored plumage patches. Using visual system-dependent methods, we showed that only some of the differences in plumage patches among the hybrid and its parent species would be perceivable by birds. Hyperspectral imaging is poised to become the gold standard for many animal coloration applications: comprehensive reflectance data—across the entire surface of an animal specimen—can be obtained in a matter of minutes. Our pipeline provides a practical and flexible roadmap for incorporating hyperspectral imaging into future studies of animal color.
README: Hyperspectral images of King, Magnificent, and hybrid King of Holland's Bird-of-Paradise (Cicinnurus regius, C. magnificus, and C. magnificus x C. regius)
https://doi.org/10.5061/dryad.j0zpc86nf
The dataset consists of 43 hyperspectral images (.BIL, with associated .hdr), of Cicinnurus regius, Cicinnurus magnificus, and C. magnificus x C. regius, and associated metadata.
Description of the data and file structure
Naming for hyperspectral images has the following convention: Species_Sex_Individual_View. For example, CM_M_01_Back indicates a dorsal image of Cicinnurus magnificus male 1. "H" stands for hybrid (C. magnificus x C. regius). TBL and TBR stand for tailback (dorsal) left and right respectively, indicating dorsal images of the left and right tail wire. TBLR is used when the image contains a dorsal view of both the left and right tail wires. "Face" indicates a ventral image, with the specimen tilted beak toward the imager. Files are in standard ENVI format.
The subfolder "Gouldian" contains one hyperspectral image of a Gouldian Finch (Erythrura gouldiae, BIL, with associated .hdr).
The subfolder "matmask" contains metadata and regions of interest for the bird-of-paradise hyperspectral images in MATLAB (.mat) format. Each contains a series of values. The following fields are included in the metadata files, inside superstructure ImStruct:
- J: a false-color RGB image of the hyperspectral image.
- J_enhanced: see J, except with histogram equalization to brighten the false-color image.
- WholeROI: polygonal ROI enclosing the entire specimen in the hyperspectral image.
- BW: binary image mask where values within WholeROI bounds are 1, and other values 0.
- bwlist: array of mask names for additional ROIs, one for each measured patch on the hyperspectral image.
- masklist: array of binary image masks where values within the ROI corresponding to this cell in bwlist are 1, and other values 0. NaN values indicate that the patch was not visible in the hyperspectral image.
- StandardSize: a record of the diameter of the 99% reflectance standard placed in the image (mm).
- standardLocation: x, y location of center and diameter of the 99% reflectance standard in the image (pixels). In CM_M_01_Face, CR_M_01_Face, and CR_M_02_Face the second entry in the standard location corresponds to the location and size of the 50% reflectance standard.
- imScaleStandard: StandardSize/diameter of the standard to yield mm/pixel image scale. In CM_M_01_Face, CR_M_01_Face, and CR_M_02_Face the second entry in the standard location corresponds to the location and size of the 50% reflectance standard.
- ref99: vector of the measured radiance (0 - 1 scale) for each band in the hyperspectral image of the median of a rectangular selection of the pixels corresponding to the 99% reflectance standard in the image.
The subfolder "MetaDepthNormal" contains for each hyperspectral image a set of metadata images. These are generated by making renders of registered 3D models (see paper for analysis and processes) for each hyperspectral image. We include the following images:
- _depth.tif: grayscale image indicating the distance of the visible polygons of the 3D model from the camera colored from nearest the camera (value: 255) to furthest from the camera (value:1), with zero values indicating background.
- _mask.png: black and white image segregating the 3D model (value: 255) from the background (value: 0).
- _normal_G.tif: grayscale image indicating the vertical component of the angle of the surface normal for the visible polygons of the 3D model in relation to the camera. Pixels with low values point up relative to the camera, and those with high values point down relative to the camera.
- _normal_R.tif: grayscale image indicating the horizontal component of the angle of the surface normal for the visible polygons of the 3D model in relation to the camera. Pixels with low values point left relative to the camera, and those with high values point right relative to the camera.
- _render.png: RGB color image showing a render generated from a mosaic of the images used to generate the 3D model.
Subfolder "Figure_dataframes" (download and unzip folder) contains .csv files containing processed data underlying all data figures in the article, see also a .txt README explaining the column titles and other conventions in the datasets.
Code/Software
See the associated article for links to the code used in our analyses.
Methods
This dataset was collected using a Resonon Pika NUV hyperspectral camera. Each image (BIL format, with .hdr files) was spatially calibrated to a PTFE standard, and each image contains a 99% reflectance standard for spectral calibration.
Images were collected for 6 male specimens (see paper for details), and included dorsal, ventral, and two lateral images (left, right), as well as a ventral image with the specimen tilted beak-toward the imager, and images of the dorsal side of the tail wires of the specimens.
We include some manually generated meta-data, including the location and size of the 99% reflectance standard in each image, and manually generated ROIs for various patches of each bird. In addition, we include angular data metadata in the form of renders of a 3D model for each hyperspectral image.
The Gouldian Finch sub-folder contains a hyperspectral image of a Gouldian Finch also collected with a Resonon Pika NUV hyperspectral camera (BIL format, with .hdr file), and was spatially calibrated to a PTFE and spectrally calibrated to a 99% reflectance standard (not present in the image).