Expert assisted exploration of photographs

While digital technology extinguished the obligatory use of the photographic film as well as the time consuming chemical photograph development techniques, the number of taken photographs has rapidly increased. As the computer harddisks as well as the photography websites substituted the photo albums, finding photographs from digital archives became difficult. Conventional image retrieval systems that seek to bridge the semantic gap, optimize photograph discovery based on associated tags or their low-level features, which can define the information regarding the content of a photograph, however not the photographs’ aesthetic value. This dissertation investigates the possible benefits of augmenting traditional query techniques with subjective expert knowledge based on the manipulation of raw low-level data in order to empower users in exploring visual media. A novel tool, XPhoto is presented which aims to enable users in retrieving photographs from large collections using not only objective tags but also subjective expertise based on photographs’ color space. Evaluation results showed that conventional tag-based systems ignore the appreciative expressions hence limit the users to search via tagged simplifications of a photograph rather than the aesthetics of the photograph itself. Injecting expert knowledge into a conventional system that only offers tag-based searching, allows users to freely express both the aesthetics of the photograph they want as well as the picture it conveys. XPhoto provided users an alternative pathway to access their large photograph collections. Initial user tests showed promise and indicate that this approach can be used to grant users the freedom they seek in relation to photograph discovery.