Optimizing image segmentation using color model mixtures

Several mathematical color models have been proposed to segment images based on their color information content. The most frequently used color models of such sort include RGB, HSV, YCbCr, etc. These models were designed to represent color and in some cases emulate how the reflection of light on a given entity is perceived by the human eye. They were, however, not designed specifically for the purpose of image segmentation. In this study, the efficiency of several color models for the application of image segmentation is assessed and more efficient color models, consisting of color model mixtures, are explored. It was observed that two of the studied models, YCbCr and linear, were more efficient for the purpose of image segmentation. Additionally, by employing multivariate analysis, it was observed that the model mixtures were more efficient than the most commonly used models studied, and thus optimized the segmentation