Structural High-resolution Satellite Image Indexing

Gui-Song Xia (1) , Wen Yang (2) , Julie Delon (1) , Yann Gousseau (1) , Hong Sun (2) , Henri Maître (1, 3)

1 LTCI - Laboratoire Traitement et Communication de l'Information
2 DSP - Signal Processing Lab
3 TSI - Département Traitement du Signal et des Images

Satellite images with high spatial resolution raise many challenging issues in image understanding and pattern recognition. First, they allow measurement of small objects maybe up to 0.5 m, and both texture and geometrical structures emerge simultaneously. Second, objects in the same type of scenes might appear at different scales and orientations. Consequently, image indexing methods should combine the structure and texture information of images and comply with some invariant properties. This paper contributes to the indexing of high-resolution satellite images. We suggest a satellite image indexing method relying on topographic maps (Caselles et al., 1999) and a shape-based image indexing scheme (Xia et al., 2009). The proposed approach contains both the textural and structural information of satellite images and is also robust to changes in scale, orientation and contrast. Experimental analysis on a real satellite image database confirms the efficiency of the approach.

Mots clés: high-resolution satellite image; structural features; texture features; Image indexing

Domaines: Vision par ordinateur et reconnaissance de formes [cs.CV]; Traitement du signal et de l'image [eess.SP]; Traitement du signal et de l'image [eess.SP]

Gui-Song Xia, Wen Yang, Julie Delon, Yann Gousseau, Hong Sun, et al.. Structural High-resolution Satellite Image Indexing. ISPRS TC VII Symposium - 100 Years ISPRS, Jul 2010, Vienna, Austria. pp.298-303. ⟨hal-00458685v2⟩