Estimating color-concept associations from image statistics:
Code written by Ragini Rathore

Code for reproducing model estimates from Rathore et al. (in press) and for applying the code for new images and colors. In this study, we developed a new approach for estimating color-concept associations. Building on prior studies that used images downloaded from Google Images, we provide new insights into effectively estimating distributions of human color-concept associations across CIELAB color space. Specifically, we evaluated several methods for filtering the raw pixel content of the images in order to best predict color-concept associations obtained from human ratings. The most effective method extracted colors using a combination of cylindrical sectors in color space and color categories. This repo consists of the image dataset, jupyter notebooks, matlab scripts and data files required to predict the color-concept associations using the cylindrical sectors and color categories as features. This repo also contains additional notebooks used for analysis.


Code written by Laurent Lessard

A Matlab function for converting between different color coordinate systems. The purpose of the colorconvert function is to provide a simple way to convert between different color coordinate systems, using either a standard whitepoint such as D65 or a user-specified whitepoint. Although Matlab’s image-processing toolbox provides much of this functionality already, the colorconvert function is simpler to use.


Stimuli and Data

Mapping color to meaning in colormap data visualizations
Karen B. Schloss, Connor C. Gramazio, A. Taylor Silverman, Madeline Parker, Audrey S. Wang
IEEE Transactions on Visualization and Computer Graphics.      

Main article (pdf)
Supplementary material (pdf)

Colormap images (zip)
Exp 1 Data (csv)
Exp 2 Data (csv)