Model for Automatic Text Classification and Categorization for Image Indexing and Retrieval

Traditional Content-based Information Retrieval Sys- tems (CBIR) use low level characteristics, this is, primary characteristics such as the color, shape, texture and also in textual attributes related to images. Although, users make queries based on semantics, which are not representatives just by such low level characteristics. Recent works on content-based image retrieval have demonstrated that re- searchers have been trying to map visual low level charac- teristics and high level semantics. These work have moti- vated this paper which proposes a model for automatic text classification and categorization for image searching by us- ing an self-organizing neural network architecture. Experi- mental results confirm this text-based model is complemen- tary to image-driven techniques such as Retin.