Color Image Classification Using Support Vector Machines

An efficient method using various histogram-based(high-dimension al)im age con tent descriptors for automatically classifying general color photos into relevant cate gories is pre-sent ed.Principal component analysis(PCA)is used to project the original high dimensional his tograms onto their eigenspaces.Lower dimensional eigenfeatures are then used to train sup-port vector machines(SVMs )to classify images into their cate gories.Experimen tal results show that even though different descriptors perform differently,they are all highly redundant.It is shown that the dimension ality of all these descriptors,re gard less of their performances,can be signifi cant ly reduced without affecting classification accura cy.Such scheme would be useful when it is used in an interactive setting for relevant feedback in con tent -based image re-trieval,where low dimen sional content descriptors will enable fast on line learn ing and re clas-sification of results.