A Discrete Model for Color Naming

Abstract

The ability to associate labels to colors is very natural for human beings. Though, this apparently simple task hides very complex and still unsolved problems, spreading over many different disciplines ranging from neurophysiology to psychology and imaging. In this paper, we propose a discrete model for computational color categorization and naming. Starting from the 424 color specimens of the OSA-UCS set, we propose a fuzzy partitioning of the color space. Each of the 11 basic color categories identified by Berlin and Kay is modeled as a fuzzy set whose membership function is implicitly defined by fitting the model to the results of an ad hoc psychophysical experiment (Experiment 1). Each OSA-UCS sample is represented by a feature vector whose components are the memberships to the different categories. The discrete model consists of a three-dimensional Delaunay triangulation of the CIELAB color space which associates each OSA-UCS sample to a vertex of a 3D tetrahedron. Linear interpolation is used to estimate the membership values of any other point in the color space. Model validation is performed both directly, through the comparison of the predicted membership values to the subjective counterparts, as evaluated via another psychophysical test (Experiment 2), and indirectly, through the investigation of its exploitability for image segmentation. The model has proved to be successful in both cases, providing an estimation of the membership values in good agreement with the subjective measures as well as a semantically meaningful color-based segmentation map.

Learning Color Names for Real-World Applications

“…• CC-III: dataset of 1014 color chips assembled by Menegaz et al [9][10]. The sampling is based on the OSA-UCS color space [20] which is a perceptually uniform space.…”

Human-centered content-based image retrieval

“…In 2005, Mojsilović [15] introduced her "computational model for color", founded on "the National Bureau of Standards' recommendation for color names". Recently, Menegaz, Le Troter, Sequeira, and Boi [14] introduced their "discrete model for color naming", "starting from the 424 color specimens of the OSA-UCS set", they "propose a fuzzy partitioning of the colorspace".…”

Customising a qualitative colour description for adaptability and usability

“…Stanchev et al [10] defined 12 fundamental colours based on the CIELUV colour space and used Johannes Itten's theory of colour to define the contrasts light-dark, warm-cold, etc. Menegaz et al [11] presented a model for computational colour categorisation and naming based on the CIE lab colour space and fuzzy partitioning using the 11 basic colour categories proposed by Berlin and Kay [12] and the OSA-UCS colour appearance system developed by the Optical Society of America (OSA) [13]. All the approaches mentioned provide evidence for the effectiveness of using different colour models and spaces for colour quantisation and naming.…”

“…In the literature, multiple tests have been carried out to participants who are asked to label uniform colour images within a controlled experimental setup where colours are chosen from a preselected set of colour names, usually the 11 English basic colour terms by Berlin and Kay [12] ( [4,11,19]). From this labelled set of uniform colour images, the correspondence between the colour coordinates and the colour name is obtained.…”

“…From this labelled set of uniform colour images, the correspondence between the colour coordinates and the colour name is obtained. Several of these works have been applied to name colours in digital real-world images [4,6,11,19,20]. Another approach to colour naming was taken by van de Weijer [21] when using images from Google image search to learn the correspondence of the 11 English basic colour terms by Berlin and Kay [12] to their RGB coordinates.…”