Ensemble classification of PolSAR data using multi-objective heuristic combination rule

Polarimetric synthetic aperture radar (PolSAR) system provides a day-or-night, all-weather means of remote sensing and produces high-resolution images. The use of these images for terrain classification is of interest to researchers. On the other hand according to recent research results, ensemble of classifiers as an effective approach has more capabilities to single-classifiers. So an optimum ensemble of classifier using multiple objective particle swarm optimization (MOPSO) and considering accuracy and reliability as objective functions is proposed. A sparse representation-based classifier and other diverse single-classifiers are used as base classifiers. The experiments over a benchmark PolSAR image demonstrate the effectiveness of the proposed algorithms over the existing techniques.