We present an energy matrix representation for multiband images that captures spatial and spectralproperties. Using a physical model for spectral reflectance, we derive a pseudoinverse method for thecomparison of energy matrices that is invariant to the spectral properties of the scene illumination. Atthe same time, this method determines the illumination change matrix, allowing direct comparison ofimages obtained under different illumination conditions. We demonstrate the performance of themethod for both illumination-invariant recognition and illumination correction on a large set ofmultiband images. The energy matrices are generated using a small set of oriented steerable filters.We also demonstrate that a related set of rotationally symmetric filters can be used for recognitioninvariant to both illumination and rotation and that subsequent processing can be used to recover therotation angle of a recognized object.
展开▼