Image segmentation using adaptively selected color space

Image segmentation is an important and difficult task in computer vision applications. Various methods have been introduced in the past to use gray-level histogram in deciding the segmentation threshold for monochrome images. With the reducing price of color cameras, different color spaces have also been considered in color image based segmentations. In this paper, a study of the effect of color spaces is presented and a segmentation strategy is introduced to select the most effective space in which the segmentation result could be improved. Experimental results show that the proposed method can provide robust segmentation outcomes subject to parts with different colors and under different illumination conditions.