A Locally-Adaptive Metric for Contrast-Fusion of Noisy Multimodal Imagery
EQUINOX CORP NEW YORK NY
Pagination or Media Count:
We propose a new technique for reducing the effect of noise on contrast fusion of multimodal imagery, yielding higher quality results than can be obtained with previous methods. We rely on local non-parametric estimation of band gradient entropies as a relative measure of geometric structure. This work builds upon previous work by the author, and is exemplified with remote sensing applications. Quantitative measures of performance are given on ground-truth data, which indicate the advantages of the new technique over existing approaches.