000148268 001__ 148268
000148268 005__ 20190316234759.0
000148268 0247_ $$2doi$$a10.1109/TIP.2011.2126584
000148268 022__ $$a1057-7149
000148268 02470 $$2ISI$$a000294132800020
000148268 037__ $$aARTICLE
000148268 245__ $$aScalable Feature Extraction for Coarse-to-Fine JPEG2000 Image Classification
000148268 269__ $$a2011
000148268 260__ $$bInstitute of Electrical and Electronics Engineers$$c2011
000148268 336__ $$aJournal Articles
000148268 520__ $$aIn this paper, we address the issues of analyzing and classifying JPEG 2000 code-streams. An original representation, called integral volume, is first proposed to compute local image features progressively from the compressed code-stream, on any spatial image area, regardless of the code- block borders. Then, a JPEG2000 classifier is presented, that uses integral volumes to learn an ensemble of randomized trees. Several classification tasks are performed on various JPEG2000 image databases and results are close or even better than the ones obtained in the literature with non-compressed version of these databases. Finally, a cascade of such classifiers is considered, in order to specifically address the image retrieval issue, i.e. bi-class problems characterized by a highly skewed distribution and by a large amount of test samples compared to learn samples. An efficient way to learn and optimize such cascade is proposed. We show that staying in a JPEG 2000 framework, initially seen as a constraint to avoid heavy decoding operations, is actually an advantage as it can benefit from the multi-resolution and multi-layer paradigms inherently present in this compression standard. In particular, unlike other existing cascaded retrieval systems, the features used along our cascade are increasingly discriminant and lead therefore to a better complexity vs performance trade-off.
000148268 6531_ $$awavelets
000148268 6531_ $$aclassification
000148268 6531_ $$aimage retrieval
000148268 6531_ $$ajpeg2000
000148268 6531_ $$alts2
000148268 6531_ $$aLTS2
000148268 700__ $$aDescampe, Antonin
000148268 700__ $$aDe Vleeschouwer, Christophe
000148268 700__ $$0240428$$g120906$$aVandergheynst, Pierre
000148268 700__ $$aMacq, Benoît
000148268 773__ $$j20$$tIEEE Transactions on Image Processing$$k9$$q2636-2649
000148268 8564_ $$uhttps://infoscience.epfl.ch/record/148268/files/Descampe-manuscript.pdf$$zn/a$$s1155883
000148268 8564_ $$uhttps://infoscience.epfl.ch/record/148268/files/Descampes.pdf$$zn/a$$s1277747
000148268 909C0 $$xU10380$$0252392$$pLTS2
000148268 909CO $$qGLOBAL_SET$$pSTI$$ooai:infoscience.tind.io:148268$$particle
000148268 917Z8 $$x120906
000148268 917Z8 $$x120906
000148268 917Z8 $$x120906
000148268 917Z8 $$x120906
000148268 937__ $$aEPFL-ARTICLE-148268
000148268 973__ $$rREVIEWED$$sPUBLISHED$$aEPFL
000148268 980__ $$aARTICLE