Colorido: Identificação da Cor Dominante de Fotografias

In this paper we propose a solution for the identification of the dominant colors (DCs) in pictures taking into account the way people perceive them. We accomplished this with six different algorithms. The developed algorithms took as their basis the non-uniform segmentation of the HSV color space with fuzzy membership functions and a palette of 12 colors. To know how many colors people usually identify in pictures we started by performing a survey with users. We also used this survey to test if the chosen palette was adequate for the identification of the DCs. After the development of the algorithms, experimental tests with users proved that our solution is suitable for the resolution of the identified problem, and that the two best algorithms are the 3x3 Segmentation and the Fuzzy Histogram. These algorithms can identify the same DCs as the users in 69,1% and 64,7% of the times, respectively.