Visual Taxonomy for Professional Image Retrieval and Automated Annotation of Images

There is an ever-growing need for new methods related to the management and retrieval of image material. The area of usage, such as the publishing industry, defines specific demands for the tools being developed. This paper presents a visual taxonomy for professional image retrieval. The purpose of the taxonomy is to provide modifiers for specifying queries from visual perspective. The development is conducted using existing literature and image descriptions provided by professionals. Procedures generally regarded as good practice as well as well-defined criteria are used in the formulation of the vocabulary. The taxonomy consists of two major levels: visual impressions and visual characteristics. The taxonomy can be converted into an ontology and it can be used as a basis for designing computational methods serving automated annotation of images.