A new search algorithm for feature selection in hyperspectral remote sensing images

A new suboptimal search strategy suitable for feature selection in very hig h-dimensional remote sensing images (e.g., those acquired by hyperspectral sensors) is proposed. Each solution of the feature selection problem is rep resented as a binary string that indicates which features are selected and which are disregarded. In turn, each binary string corresponds to a point o f a multidimensional binary space. Given a criterion function to evaluate t he effectiveness of a selected solution, the proposed strategy is based on the search for constrained local extremes of such a function in the above-d efined binary space. In particular, two different algorithms are presented that explore the space of solutions in different ways. These algorithms are compared with the classical sequential forward selection and sequential fo rward floating selection suboptimal techniques, using hyperspectral remote sensing images (acquired by the airborne visible/infrared imaging spectrome ter [AVIRIS] sensor) as a data set. Experimental results point out the effe ctiveness of both algorithms, which can be regarded as valid alternatives t o classical methods, as they allow interesting tradeoffs between the qualit ies of selected feature subsets and computational cost.