Visual Querying By Color Perceptive Regions

Image databases are now currently employed in an eclectic range of different areas such as entertainment, art history, advertising, medicine and industry among others. In all these contexts, a key problem regards the modality of accessing visual image content. The visual content of an image should be intended as what humans remember after looking at it for a couple of minutes. It may include shapes of relevant objects, color distribution or color patches, textured surfaces, or the arrangement of visual elements on the image plane. In image databases, unlike textual and numerical databases, the visual content by itself has no ordering rules, unless textual labels are associated with images. Conventional attempts to cast visual features into textual keywords [Srihari] has been far recognized to be inadequate in indexing pictures. The minor expressiveness of text with respect to visual features doesn't allow to fully exploit capabilities of human memory. Items retrieved through a textual query could not be relevant at all for user's expectation. This is the reason why retrieval based on visual content has been identified as the means to overcome this modal clash. Pattern recognition and image analysis algorithms are used to extract parameters of the visual content and filter out unrelevant images. The relevance of visual elements for the final user depends on his subjectivity (which is not known in advance, when the database is created) and on the context of application (which is known, instead). On the one hand, user's subjectivity implies that modules are employed which are able to cope with imprecision and lack of knowledge about image content. On the other hand, the fact that the context of application is known in advance, can drive the choice of pattern recognition algorithms and functions that must be included in the system. For example, spatial arrangement is relevant to a doctor examining a heart section; color arrangement is important when looking at paintings, but the presence of particular color tones or the disposition of color patches on the canvas will be better remembered by an expert than by an occasional user.