Hierarchical image browsing based on visual information

The aim of this paper was to develop an image browsing system, where only visual information is available for users and computers as well. Large image set is structured in albums and the idea was to select the most representative image from each album in the level of hierarchy, and then the next upper level in the hierarchy consists of these representative images. Selection of the most representative image is based on semantic features, which helps users in browsing process. Quantitative and qualitative analysis have been completed, where the goodness of the representative image selection is tested and comparison with other systems is accomplished respectively. Finally it can be concluded that the system developed possesses semantic and structured features.