Art Image Retrieval Utilizing Colour Semantics

The field of art image retrieval has to overcome a major challenge: it needs to accommodate the obvious difference between the digital technologies that are limited within pixels capture, and human expectation for perceiving different semantic, aesthetic and cultural messages which the artwork sends to the viewer.In the process of perception, colour responses are more connected to human emotions than to a rational mind. The presence of one or more colours in different proportions conveys different messages which increase or suppress the perception of the observed objects. This work provides a detailed analysis of the successful colour combinations and one approach for their formalization in order to implement automated extraction from digitized artworks. Proposed and developed types of harmonies and contrasts features strive to narrow the semantic and abstraction gap between low-level automatic visual extraction and high-level human expression. The vividness of proposed features opens the door for indexing and searching in paintings repositories according to such characteristics and can be used as a step in the transition from Web 2.0 to Web 3.0.