Hyperspectral image unmixing via quadratic programming

Estimating abundance fraction of materials in hyperspectral images is an important area of study in the field of remote sensing. The need for fraction estimation in remotely sensed images arises from the fact that the sampling distance is generally larger than the size of the targets of interest. The abundance sum-to-one constraint and the abundance nonnegativity constraint are the two physical constraints often considered when developing estimation methods. In this paper, we relax the abundance sum-to-one constraint as this condition is rarely satisfied in reality and use the relaxed sum-to-one constraint instead to develop a constrained method. Experimental results demonstrate that this method provides more precise estimation of the fractions when the sum-to-one constraint is violated, while producing similar results to other existing methods when the constraint holds.