Burns are some of the most severe forms of accidental trauma across the world. Burn injuries require specialized care and an early and accurate distinction between superficial dermal burns and deep dermal burns which require further surgical procedures, as they do not heal spontaneously. This paper proposes a multispectral imaging based diagnosis support system for the identification of severe burns. The acquisition is performed in both visual and infrared domains, simultaneously, with a thermal camera recording in parallel color and thermal images of the unconstrained patient and patient environment. The classification of burns is developed under a supervised scenario, according to a ground truth defined by specialist surgeons from a large pediatric case database, by an ensemble of methods that gather votes from both convolutional neural network and standard pattern recognition systems.