Sub-Pixel Mapping with Multiple Shifted Hyperspectral Images Based on Multiobjective Evolutionary Algorithm

摘要: Sub-pixel mapping (SPM) can interpret the sub-pixel spatial distribution of land-cover classes in hyperspectral image, which is an ill-posed problem due to inadequate information a single image. Auxiliary provided by multiple shifted (MS) images make SPM well-posed and improve accuracy. The maximum posteriori (MAP) technique incorporate auxiliary MS images, but it introduces fixed weight parameter fuse prior information, heavily influencing result. This paper proposed novel method model into two objective functions, be simultaneously optimized devised multiobjective evolutionary algorithm. Therefore, there no need parameter, functions intelligently integrated during evolution. Experimental results analysis have indicated superiority method.