Comparative Overview of Color Models for Content-Based Image Retrieval

Today CBIR (content-based image retrieval) in contrast to conventional TBIR (text-based image retrieval) has become a focusing research area in image processing due to numerous application possibilities. These applications vary from medical and security to business and SNS applications, to name a few. Color is the most widely used image characteristic since it is independent of image resolution or orientation. There is no single best color representation. The target application has a big role in determining which color model is best to use. To speed up the retrieval time and get good accuracy, one must understand which color model to employ. In this paper, we provide a critical review of existing color models, explain their attributes, analyze them from various perspectives and provide a context-aware comparative evaluation. We study when and why certain color models outperform others in certain applications under certain circumstances. Color spaces described in this paper are not only well-known e.g., RGB, CMYK, HSV, and Munsell, but also novice systems, like fuzzy color models, that can process higher color semantic levels.