Authors:
Hye-Yoon Woo
and
Hang-Bong Kang
Affiliation:
Catholic Univ. of Korea, Korea, Republic of
Keyword(s):
Color Transformation, Affection, Interactive Genetic Algorithm.
Related
Ontology
Subjects/Areas/Topics:
Computer Vision, Visualization and Computer Graphics
;
Image Enhancement and Restoration
;
Image Formation and Preprocessing
Abstract:
In this paper, we propose a new method to transform colors in the image according to the user’s emotion. First, we labeled the images into four basic emotions such as happy, sad, fear and anger from the users’ evaluation. After that, we compute color histograms using 36 quantized HSV scheme. After that, we intersect color histograms and finally construct color templates for basic emotions. To reflect the user’s preferences in color perception for specific emotion, we designed interactive genetic algorithm. First, we generate first stage individuals using color templates from the input image. From them, we extract chromosomes for HSV color space and then generate next stage individuals using crossover and mutation operations. After a few number of iterations, we stop the process and extract personalized color templates from the finalized image. Using personalized color templates, the user can generate an affection-based color transformed image according to his own preference. Survey r
esults show that desirable results are obtained from our proposed personalized color templates.
(More)