Technical Program

Paper Detail

Paper:	IMDSP-P4.8
Session:	Image/Video Indexing and Retrieval
Time:	Tuesday, May 16, 16:30 - 18:30
Presentation:	Poster
Topic:	Image and Multidimensional Signal Processing: Image Indexing and Retrieval
Title:	Optimizing Metrics Combining Low-Level Visual Descriptors for Image Annotation and Retrieval
Authors:	Qianni Zhang, Ebroul Izquierdo, Queen Mary, University of London, United Kingdom
Abstract:	Recognizing objects and semantic concepts in images has become a major research topic in image processing. It is the pillar for automatic annotation and retrieval of visual in-formation and in a more generic context it is the base for high-level image understanding. However the transition between the automatic classification of low-level primitives and higher concepts remains an open problem and is referred as the ‘semantic gap’. To bridge this gap two profound challenges are evident: how to deal with the subjective interpretation of images by different users under different conditions; and how to link a semantic-based concept with low-level metadata. This paper presents an object oriented approach for keyword annotation and considers com-binations of low-level descriptors and suitable metrics to represent and measure similarity between single objects present in images.