期刊名称:International Journal of Advanced Computer Research
印刷版ISSN:2249-7277
电子版ISSN:2277-7970
出版年度:2013
卷号:3
期号:2
出版社:Association of Computer Communication Education for National Triumph (ACCENT)
摘要:Unclas sified region deceases t he efficiency and performance o f PLSA an d FLDA . The pro per selection o f feature sub set reduced th e un class ified region of PLSA and FLDA . No w a day .s bin ary classificatio n are widely used in image clas sification . Th e mapping of d ata o ne space to anot her space creates div ersit y of outlier and n oise and gen erate un clas sified reg ion for image classification . Fo r th e reduction o f un class ified region we us ed radial basis fu nctio n fo r samplin g of featu re an d reduce the noise an d ou tlier for featu re sp ace of data an d in creas e th e performance an d efficien cy of image clas sification . Our propos ed metho d op timized th e feature selection proces s and finally s ends dat a to FLDA clas sifier for classificatio n of data. Here we us ed fish er classifier. A s a classifier FLDA su ffering two p roblems (1) ho w to choo se o ptimal feature sub set in put and (2) h ow to set bes t kernel parameters. Th ese prob lems influ ence th e performance an d accuracy o f FLDA. Now the pre-samplin g of feat ure reduced th e feature s election proces s of FLDA fo r image class ificat ion