Title :
Biased isomap projections for interactive reranking
Author :
Bian, Wei ; Cheng, Jun ; Tao, Dacheng
Author_Institution :
Shenzhen Inst. of Adv. Technol., Chinese Acad. of Sci., Shenzhen, China
fDate :
June 28 2009-July 3 2009
Abstract :
Image search has recently gained more and more attention for various applications. To capture users´ intensions and to bridge the gap between the low level visual features and the high level semantics, a dozen of interactive reranking (IR) or relevance feedback (RF) algorithms have been developed and achieved significant performance improvements. In this paper, we develop a novel subspace learning based IR algorithm by using the patch alignment framework, termed the biased ISOMap projections or BIP for short. BIP models both the intraclass local geometry for query relevant images and the interclass discrimination between query relevant images and irrelevant images. In addition, BIP never meets the small samples size problem. We present experimental evidence suggesting that BIP is effective for targeting the intensions of users and reducing the semantic gaps for image search.
Keywords :
computational geometry; image retrieval; learning (artificial intelligence); relevance feedback; biased isomap projection; geometry; high level semantics; image search; interactive reranking algorithm; interclass discrimination; low level visual feature; patch alignment framework; query irrelevant image; relevance feedback algorithm; sample size problem; subspace learning; Application software; Bridges; Feedback; Image retrieval; Information retrieval; Kernel; Linear discriminant analysis; Principal component analysis; Radio frequency; Solid modeling; Interactive reranking; dimensionality reduction; manifold learning; web image search;
Conference_Titel :
Multimedia and Expo, 2009. ICME 2009. IEEE International Conference on
Conference_Location :
New York, NY
Print_ISBN :
978-1-4244-4290-4
Electronic_ISBN :
1945-7871
DOI :
10.1109/ICME.2009.5202832