Title :
Cascading classifiers for consumer image indexing
Author :
Lim, Joo-Hwee ; Jin, Jesse S.
Author_Institution :
Inst. for Infocomm Res., Singapore
Abstract :
We propose a cascading framework of binary classifiers to extract and combine intra-image and inter-class semantics for image indexing and retrieval. Support vector detectors are first trained on semantic support regions without image segmentation. The reconciled and aggregated detection-based indexes then serve as input for support vector learning of image classifiers to generate class-relative image indexes. During retrieval, similarities based on both indexes are combined to rank images. Query by-example experiments on 2400 heterogeneous consumer photos with 16 semantic queries show that the combined matching approach is better than matching with single index. It also outperformed combined matching of color and texture features by 55% in average precision.
Keywords :
content-based retrieval; image classification; image colour analysis; image matching; image retrieval; image texture; learning (artificial intelligence); binary classifiers; consumer image indexing; heterogeneous consumer photos; image classifiers; interclass semantics; intraimage semantics; query by-example experiments; semantic queries; support vector learning; Cities and towns; Detectors; Image segmentation; Indexing; Labeling; Layout; Object detection; Object recognition; Support vector machines; Videos;
Conference_Titel :
Pattern Recognition, 2004. ICPR 2004. Proceedings of the 17th International Conference on
Print_ISBN :
0-7695-2128-2
DOI :
10.1109/ICPR.2004.1333917