We present an energy matrix representation for multiband imageswhich captures spatial and spectral properties. Using a physical modelfor spectral reflectance, we derive a pseudoinverse method for thecomparison of energy matrices which is invariant to the spectralproperties of the scene illumination. At the same time, this methoddetermines the illumination change matrix allowing direct comparison ofimages obtained under different illumination conditions. We demonstratethe performance of the method for both illumination-invariantrecognition and illumination correction on a large set of multibandimages. The energy matrices are generated using a small set of orientedsteerable filters. We also demonstrate that a related set ofrotationally symmetric filters can be used for recognition invariant toboth illumination and rotation and that subsequent processing can beused to recover the rotation angle of a recognized object
展开▼
