In this paper, we propose a new re-coloring algorithm to enhance the accessibility for the color vision deficient (or colorblind). Compared to people with normal color vision, people with color vision deficiency (CVD) have difficulty in distinguishing between certain combinations of colors. This may hinder visual communication owing to the increasing use of colors in recent years. To address this problem, we re-color the image to preserve visual detail when perceived by people with CVD. We first extract the representing colors in an image. Then we find the optimal mapping to maintain the contrast between each pair of these representing colors. The proposed algorithm is image content dependent and completely automatic. Experimental results on natural images are illustrated to demonstrate the effectiveness of the proposed re-coloring algorithm.

1. Image Recolorization for the Colorblind
Jia-Bin Huang1, Chu-Song Chen1, Tzu-Cheng Jen2, Sheng-Jyh Wang2
1 Institute of Information Science, Academia Sinica, Taipei, Taiwan
2 Institute of Electronics Engineering, National Chiao Tung University, Hsinchu, Taiwan
Problem The Proposed Algorithm Experimental Results
• People with color vision deficiency (CVD) have difficulty in • Image Representation via Gaussian Mixture Modeling • Sample results
distinguishing between some colors. Use the CIEL*a*b* color space
• How the colorblind perceive colors? Approximate the distribution by K Gaussians: p (x |Θ) = K 1 ωi Gi (x |θi )
i=
Learn the parameters by the Expectation-Maximization (EM) algorithm
Select the optimal number of K by the Minimum Description Length (MDL) principle
• Target Distance
Generalize the concept of key colors from “point” to “cluster”.
Background Use symmetric Kullback-Leibler (KL) divergence as our dissimilarity measure The symmetric KL
divergence: DsKL(Gi , Gj ) = DKL(Gi ||Gj ) + DKL(Gj ||Gi )
For Gaussians, analytical solutions exists
• CVD results from partial or complete loss of function of one or DsKL(Gi , Gj ) = (µi − µj )T (Σ−1 + Σ−1)(µi − µj )+
i j
more types of cone cells. tr (Σi Σ−1 + Σ−1Σj − 2I )
j i
• Simulation of color perception of people with CVD [1]
• Optimization
• Related works addressing CVD accessibility:
Define the color mapping functions Mi (·), i = 1, . . . , K
Guidelines for designers to avoid ambiguous color combinations The error introduced by the ith and jth key colors:
(Semi-)automatic methods for recoloring images
Ei ,j = [DsKL(Gi , Gj ) − DsKL(Sim(Mi (Gi )), Sim(Mj (Gj )))]2
Introduce weights for each color values: αj = ||xj − Sim(xj )||
Obtain weight for each cluster:
(a) (b) (c)
Contributions λi = K
N
j =1 αj p (i |xj , Θ)
Figure: (a) Original images. (b) Simulated views of the original images for
N protanopia (first row), deuteranopia (second row), and tritanopia (third row).
i =1 j =1 αj p (i |xj , Θ) (c) Simulation results of the re-colored images.
• Generalize the concept of key colors in a image Rewrite the objective function
• Propose to measure the contrast between two key colors by the
i =K j =K • Comparison with [2]
E= (λi + λj )Ei ,j .
symmetric KL divergence i =1 j =i +1
• Interpolate colors to ensure local smoothness with a few key Minimize the objective function via direct search optimization method.
colors • Gaussian Mapping for Interpolation
Compute the transformed colors using
K
T (xj )H = xjH + p (i |xj , Θ)(Mi (µi )H − µH )
Reference i =1
i
(a) (c) (e)
1 H. Brettel, F. Vienot, and J.D. Mollon, “Computerized simulation
of color appearance for dichromats”, J. Optic. Soc. Amer. A, Future Directions
1997.
2 L. Jefferson and R. Harvey, “Accommodating color blind • Relax the assumption of Gaussian distribution (e.g., use non-parametric modeling)
computer users,” in ACM SIGACCESS, 2006. (b) (d) (f)
• More principled optimization procedure
Figure: Comparison with [2]. (a) The original image. (b) The simulated view
• Subjective evaluation
of (a) for tritanopia. (c)(d) The re-colored result by the proposed method and
its simulated view. (e)(f) The re-colored result by [2] and its simulated view.
Jia-Bin Huang et al. (Academia Sinica and National Chiao-Tung University, Taiwan) International Conference on Acoustics, Speech, and Signal Processing 2009