Abstract
The chromatic diversity of natural scenes has been quantified from hyperspectral imaging data. Do artistic paintings mimic the same color diversity found in natural scenes? The purpose of this work was to analyze and compare the chromatic diversity of natural scenes and art paintings. Fifty images of natural scenes and 43 images of art paintings were digitized using a hyperspectral imaging system composed by a monochromatic CCD camera and a fast tunable filter capable of a spectral resolution from 400 to 720 nm in 10 nm steps. The images contained reflectance data, estimated from a grey reference with known reflectance presented at the scene at the time of acquisition. The radiance was then estimated for each pixel of the images assuming the CIE D65 illuminant. The chromatic diversity was assumed to be the number of discernible colors (NDC) in each image. The NDC was computed by representing all the image's pixels in the CIELAB color space and by segmenting it into unitary cubes. It was assumed that all the colors that were inside the same cube could not be discernible, and by counting the number of non empty cubes the NDC was obtained. It was found that the chromatic diversity in the form of NDC of the art paintings is different from that of natural scenes. The distributions of colors indicate that paintings are richer chromatically than natural scenes. This result suggest that artists are capable to generate richer chromatic information despite the gamut limitations imposed by the pigments used. Acknowledgement: This work was supported by FEDER through the COMPETE Program and by the Portuguese Foundation for Science and Technology (FCT) in the framework of the project PTDC/MHC-PCN/4731/2012.
Meeting abstract presented at VSS 2016