Using A Crop-Pest Ontology To Facilitate Image Retrieval
Soonho Kim, Yun-Chul Jung, Howard Beck · Computers in Agriculture and Natural Resources, 23-25 July 2006, Orlando Florida · 2013
Professionals in the agricultural field need a facility to retrieve photographic images related totheir work, especially as the volume of such images continues to increase. However, current keyword-basedimage retrieval suffers from limitation of finding relevant images and helping users to find proper keywords.A new approach to image retrieval using an ontology addresses the limitations posed by the currentkeyword-based image retrieval, which is achieved by browsing images associated with an ontology. Anontology is a collection of concepts and relationships between them in a particular domain such as crops andrelated pests. The presented work used two hundred and ninety-one images for developing the ontologybasedapproach in the domain of crops and related pests. To enable browsing of images associated with thecrop-pest ontology, images were indexed based on the ontology. The indexing process included analyses ofeach image caption based on grammatical structure and word meaning. A graphical interface wasimplemented for browsing images associated with concepts in the crop-pest ontology and was evaluated forfinding relevant images and helping users to find proper keywords. The results of evaluation indicated thatparticipants achieved high relevancy in the retrieval of images using the crop-pest ontology, comparing tothe keyword-based image retrieval (p = .02). The results also indicated the strong possibility that the imageretrieval using the crop-pest ontology can help users by transferring domain knowledge to them. Thepresented work showed the new approach of indexing and browsing images using the crop-pest ontology,which can help professionals retrieve images more easily and accurately in the agricultural field.