Enhancing Supervised Hyperspectral Unmixing using Spatial Correlation under Nonlinear Mixing Model

In this paper, a supervised hyperspectral unmixing solution that integrate the spatial information in the abundance estimation procedure is presented. The proposed method is applied on a nonlinear model based on Polynomial Postnonlinear Mixing Model (PPNMM) where characterizes each pixel reflections composed of nonlinear function of pure spectral signatures corrupted by noise. The image is iteratively classified to classes where contains similar spectral reflectance so share the same abundance vector. Then the abundance vector is estimated for all pixels belong to each class. To make classification, the spatial correlation between pixels belonging to each class is modelled by Markov Random Field (MRF). A Bayesian framework is proposed to estimate the classes and corresponding abundance vectors alternatively. Due to complexity of derived likelihood function, a Markov Chain Monte-Carlo (MCMC) algorithm is used to estimate the abundance vector based on generated samples. The result of implementation on simulated data shows around 20% prominence of proposed approach in comparison to nonlinear unmixing and MRF-based linear unmixing algorithms in the sense of estimation and reconstruction error.