Fusion of band selection methods for hyperspectral imaging

Each band selection (BS) method has its own merit by finding a different band subset for a given number of bands to be selected, n_BS. In order to take advantage of these different band subsets, a fusion of these band subsets is highly desirable. This chapter presents an approach to band selection fusion (BSF), which fuses bands produced by a set of various BS methods for the same n_BS. Its idea is to first find the union of all band subsets produced by a set of various BS methods as a joint band subset (JBS). Due to the fact that a band selected by one BS method in JBS may be also selected by other BS methods, in this case each band in JBS is prioritized by the frequency of this particular band appearing in the individual band subsets to be fused. Such frequency is then used to calculate its priority probability to be fused in JBS. Two versions of BSF, called progressive BSF (PBSF) and simultaneous BSF (SBSF), are developed for this purpose. Of particular interest is that BSF can prioritize bands without band de-correlation, which has been a major issue in many single band-selected BS methods.