Image preparation and calibration

Digital cameras are an accessible, affordable, and non-invasive method of data collection. The resulting images, however, are optimized for human vision and for display on commercial RGB monitors. The actual spectral reflectance of the photographed object is therefore distorted in a digital image. Accurate image calibration, including white balance, radiance normalization, and converting to the color sensitivities of non-human animals, is an essential step before image analysis. A comprehensive discussion of image calibration is beyond the scope of this paper, but see Troscianko & Stevens (2015), Byers (2006), Endler & Mielke (2005) and Schindelin et al. (2012).

Because colordistance does not include image calibration tools, images should be calibrated before being analyzed in R. There are a variety of tools available for image calibration, including simple white-balance correction in most image editing applications. The image calibration and analysis ImageJ toolbox by Troscianko & Stevens (2015) allows users to not only calibrate images, but also to correct for the non-linearity of RGB images and to incorporate ultraviolet (UV) channels to simulate animal color vision; the plug-in is free and comes with a comprehensive guide for users with camera RAW images.

Background masking is the last step of image preparation. Any part of an image that the user wants to ignore should be masked out with a uniform background color that is not similar to any of the colors in the object itself; the examples below use bright green (RGB triplet of (0, 1, 0) on a 0–1 scale) and white (RGB triplet of (1, 1, 1)). This can be accomplished with Photoshop, ImageJ, or other image editing software.