Image Retrieval Using Natural Language and Content-Based Techniques
Kate G. Byrne, Ewan Klein · 2003
This MSc thesis deals with the application of Natural Language Processing and ContentBased Image Retrieval to the practical problem of finding database images to answer user queries. The data collection used contains about 50,000 digital images linked to a large text database. In contrast to many otherwise similar image retrieval projects, the text used comes from a range of data fields, not just captions individually linked to images. The data relates to archaeology and architectural history, and contains a lot of domain-specific language.