ABSTRACT
Color in general plays a crucial role in today’s world. Color alone can influence the way humans think, act, and even react. It can irritate or soothe eyes, create ill-tempers, or result in loss of appetite. Whenever used in a right form, color can save on energy; otherwise, it can cause detrimental effects. Particularly, the color composition of an image can turn out to be a powerful cue for the purpose of content-based image retrieval (CBIR), if extracted in a perceptually oriented way and kept semantically intact. Furthermore, the color structure in a visual scenery is robust to noise, image degradations, changes in size, resolution, and orientation. Eventually, most of the existing CBIR systems use various color descriptors in order to retrieve relevant images (or visual multimedia material); however,
their retrieval performance is usually limited especially on large databases due to lack of discrimination power of such color descriptors. One of the main reasons for this is because most of such systems are designed based on some heuristics or naive rules that are not formed with respect to what humans or more specifically the human visual system (HVS) finds relevant in terms of color similarity. The word relevance is described as “the ability (as of an information retrieval system) to retrieve material that satisfies the needs of the user.” Therefore, it is of decisive importance that human color perception is respected while modeling and describing any color composition of an image. In other words, if and only when a particular color descriptor is designed based entirely on HVS and human color perception rules, further discrimination power and hence certain improvements in the retrieval performance can be achieved.
