%0 Conference Proceedings
%4 dpi.inpe.br/marte/2011/07.21.18.24
%2 dpi.inpe.br/marte/2011/07.21.18.24.49
%@isbn 978-85-17-00056-0 (Internet)
%@isbn 978-85-17-00057-7 (DVD)
%T Registro automático de imagens SAR interferométricas
%D 2011
%A Salles, Roberto Neves,
%A Mura, José Claudio,
%A Fonseca, Leila Maria Garcia,
%@affiliation Instituto de Estudos Avançados - IEAv/CTA
%@affiliation Instituto Nacional de Pesquisas Espaciais - INPE
%@affiliation Instituto Nacional de Pesquisas Espaciais - INPE
%@electronicmailaddress salles@ieav.cta.br
%@electronicmailaddress mura@dpi.inpe.br
%@electronicmailaddress leila@dpi.inpe.br
%E Epiphanio, José Carlos Neves,
%E Galvão, Lênio Soares,
%B Simpósio Brasileiro de Sensoriamento Remoto, 15 (SBSR).
%C Curitiba
%8 30 abr. - 5 maio 2011
%I Instituto Nacional de Pesquisas Espaciais (INPE)
%J São José dos Campos
%P 8288-8294
%S Anais
%1 Instituto Nacional de Pesquisas Espaciais (INPE)
%K remote sensing, interferometry, sensoriamento remoto, interferometria, registro subpixel, FFT.
%X This paper explores an automatic registration procedure for use on single-look complex InSAR image pairs. The procedure is divided on two distinct parts. First a coarse registration is done using the phase correlation technique. The coarse registration is a global registration method that calculates the vertical and horizontal shifts on a pixel scale. The calculated translation is applied on the second image without need of interpolation. The new pair is then processed with the fine registration. At this point a grid of small window pairs is equally distributed through the images. Each pair is zero-padded to avoid circular convolution issues and is transformed using Fourier transform. Subpixel resolution is achieved zero-padding the frequency by a given factor before applying the same phase correlation technique, now as a local registration method. Control points are defined as the center of each pair of windows processed and are saved on a list with calculated subpixel shifts. At the end a polynomial transformation is calculated based on least squares using the list of control points and subpixel shifts. Finally the second image is interpolated. This procedure was implemented on IDL/ENVI and tested on two data sets, X-Band R99 and L-Band SIR-C. Results show that it is possible to automatically produce better interferograms as the number of residues is reduced.
%9 SAR: Processamento e Aplicações
%@language pt
%3 p1104.pdf