ITB J. ………….. Vol. XX …, No. X, 20XX, XX-XX 1 Received ________, Revised _________, Accepted for publication __________ A NOVEL SWARM-BASED APPROACH TO COLOR QUANTIZATION Amir Poorsadeg-Zadeh-Yeganeh 1 , Mustafa Oral 2 , Serkan Kartal 3 Çukurova University, Faculty of Architecture and Engineering, Department of Computer Engineering, 01330, Adana, Turkey 1 Email: apoorsadeg@student.cu.edu.tr 2 Email: moral@cu.edu.tr 3 Email: skartal@cu.edu.tr Abstract. Color Quantization (CQ) which is considered as an NP-Hard optimization problem is a process that aims to reduce the number of distinct colors in a given image. The goal of CQ is to represent the image with fewer colors and less data while trying to keep the distortion of the final image as low as possible. In this paper, a novel approach based on Intelligent Water Drops (IWD) Algorithm is proposed for CQ problem. The proposed algorithm IWD- CQ is compared with five conventional methods of CQ; Uniform, Popularity, Median-Cut, Octree and K-Means, in terms of visual quality of the quantized image. The comparison results show the superiority of IWD-CQ to the all tested methods in this study. Keywords: color quantization; image processing; intelligent water drops algorithm; swarm intelligence; artificial intelligence. 1 Introduction Display devices such as monitors and LCD screens represent colors by using the three primary colors; Red, Green and Blue. Various intensities of primary colors are added together to make different colors (additive colors). In true- color digital images, the intensity range of each primary color is usually digitized to 256 levels using 8 bits. Therefore, 24 bits are needed to save color information of each pixel. This provides approximately 16.7 million (2 24 ) discrete colors which is not necessarily distinguishable to the human visual system. In some display devices, due to hardware limitations, it is impossible to display this range of colors simultaneously. The well-known solution to this problem is to decrease the number of distinct colors in an image by Color Quantization (CQ) process [1]. Although today’s display devices have improved considerably, the importance of CQ is not diminished. Processing, storage and transmission of 24 bits per pixel is problematic for many applications and causes inefficiency. Additionally, in some applications such as medical imaging, it is not necessarily needed to represent the images in true-color. CQ techniques are extensively