1 / 1

Feature Selection for Satellite Image Indexing

Feature Selection for Satellite Image Indexing. Marine Campedel*, Eric Moulines*, Henri Maître* and Mihai Datcu** Competence Center on Information Extraction and Image Understanding for Earth Observation *GET-Telecom Paris 46 rue Barrault 75013 Paris – France - e-mail: Marine.Campedel@enst.fr

sunila
Download Presentation

Feature Selection for Satellite Image Indexing

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Feature Selection for Satellite Image Indexing • Marine Campedel*, Eric Moulines*, Henri Maître* and Mihai Datcu** • Competence Center on Information Extraction and Image Understanding for Earth Observation • *GET-Telecom Paris 46 rue Barrault 75013 Paris – France - e-mail: Marine.Campedel@enst.fr • **German Aerospace Center (DLR) – D-82234 Oberfaffenhaffen – Germany – e-mail : Mihai.Datcu@dlr.de Feature Selection Features (+labels) Cross validation loop Learning/Testing Feature Selection Feature Selection Classification Heuristics Classifier learning Classification error rates (means and variances) Redundancy Evaluation Experiments Features Extraction Features Selection Small Images 64x64 Spot5 (5m/pixel) @CNES 6 semantic classes : city, forest, field, sea, desert, cloud 100 images for each class Classification Classification error rates (means and variances) H representation entropy: high values indicate low redundancy Classification error rates (mean and variances) • Features sets efficiency is compared using automatic feature selection and classification algorithms. We observe that : • Automatic feature selection enhances the classification performances; • Selection and classification algorithms have to be chosen jointly; • Textural and geometrical features are complementary for Spot 5 images • (4/14 Fisher-FS selected features are geometrical).

More Related