Multiple Targets Inequality Constrained Energy Minimization for Multispectral Imagery

Abstract Multiple Targets Constrained Energy Minimization is the extension of Constrained Energy Minimization in the case that the number of targets of interest is more than one. It designs a finite-impulse response filter in such a manner that the average output energy is minimized subject to unit responses to all desired targets. However, constraining the responses of all targets to a specified value is too rigid and, in many situations, not necessary. In this paper, we have found that by simply relaxing the equality constraints in Multiple Targets Constrained Energy Minimization to inequality constraints, not only can the solution space be theoretically enlarged, but also the detection accuracy of targets can be effectively improved in practical applications. Experiments with simulated and real data demonstrate the effectiveness of the presented method.