Application of the polarimetric matched image filter to the enhancement of characteristic features of POL-SAR imagery

Abstract

Y usung-Gu Tae-Jon.Shi 300-3 1 Korea KAIST-KIT/EML Spckle r.duction has lay ben rec0gnIr.d as the u i n problem of coh.rent it. nn vast u j o r i t y of then todmipus. -r, are of a scalar nature s i q l y b c a u n n c t o r / u t r i x I u g h q data are so s p r u ud have hcor available ml w y -r m n a y . w. do haw ~c k data, m s taken with the raJh/JpL ~v -9 9 8 ~I M p l a r i s a t i m &bud (1.225 O a l SWSynthatic Warture Radar) system, have bem M& available to U. W. 1nwstig.t. the potential of M exclwiwly p 1 a r i r t r i c Ilvye filtering ;/ holch, which takes full advantage Of the m t r i x data provided on a pixel-ky-pfxel basis. ud cclp1-t~ the existing scalar cmtrast optbization and s p c k l e reduction t . c h d q u s . W. wish t o streaa f r a the outset that our so.1 Is contrast o p t b i m t i m wiAtkart the help of 1ncOh.rent ~r a q i n g w a r pixels/look, because of tha correspading loss of spatial or t a p o r a l resolution. 1. P " m I c rmMRm6 ilvqing ud mny pronssing techniqus hm b..n advmced t o w . r m i o l l d n g [ I ], w define the receivinq a n t e m tanina1s with the &+Is directed C a r d tha target (pixel1 as Shoun i n ( 2 rig 11 Note that i n SAR a p p l i u t i a u eh. r m i v e r and t r a n d t t e r are co-l&ated'tat' lvy te different anterms so that eh. situation is slightly bistatic (31. Ru received -$ ud tha tr.nrritted -5 uaws, together With tha antmm height h (polarization of the r m i v i n g a n t e m vhen used as tr-tt e r [ ) ] I , CM J 1 b polarIration ration r t~-~lE/E~l, .nd + ard I I are the relative and absolute phases.respectively [31, ud related by the 2x2 ccqlex Sinclair scatterinq mdtrlx IS1 origin of the mOrdiMte system a t the written as pluu mves in terms of the caplex $ -lSl+ . (1) % voltage a t the receiving antenna terminals as a function of transmitter and receiver p l a r i u t i a u is given by v . $5 -$IS& , 121 where suparrcript f denotes the transpose (as oppcsed to hernutian conjuqate . +. , . n 12, pp. 1471-14731, (31 for details). Since a Coqlete scattering mtrix ( S I W ) ] is available for w e r y pixel, one can s i d a t e tha resof U. img. to m y t r a n b t t e d polarization -5 by u l a t d for an a r b i t r a p receinr p l s r l z a t i m h via (21. Bath +timu m i t b. 1 q l . m t . d for each pixel of UW entire imp.: RI. brightruss is than aisiqned to each pixel sCeOCdlng to P m V*V -I h ' s ) * ( _ k s ) . CalCUhtiV -5 Via (11. ? U C t h e m C * , the W S p U Of the lug. C M alS0 h Si-2. -8EKiL -2-W a V 5 Rl(n lRpux IG] Ih. TSP addresus t h f o l l d n g p r o b l a ( M [2,3] for -re detailsl: For a giwn pixel ( i . e . , know scattering matrix), find such t r W l t t t i n g .nd receiving polariratiau, for which tha receiwd p a r is r u i m a l ( r l n i r s l l . ruthmatically it translates into: find 5 _h such that v -$[SI% is optilvll for a qiwn [SI, subject t o the c a u t r a i n t s ll_hll -ll-$ll -1. nN w uUqli.hn this in t h r a up.rata st.9.s [1,21: st.9. 1 1 Ru m r g y &tufty in tha reflected uave (before it reachas tha I...

Radar target recognition by probabilistic filtering

“…Their use allows deriving the information about electrophysical and geometrical properties of radar targets [4,5].…”