Uncooled infrared (IR) microbolometer cameras are gaining popularity in a variety of military and commercial applications due to their simplicity, compactness and reduced cost when compared to photon detectors. Three commercially available IR microbolometer cameras have been investigated for use in a system. The cameras have been characterized in terms of camera response and noise as function of camera temperature with the aim of modelling the cameras for use in simulation. Ideally, the camera systems, consisting of a detector, electronics, and optics, should be modelled from a low-level physical point of view and measurements should be performed for verification. However, the detector and electronic design parameters are not available for the commercially acquired cameras, and a black-box approach of the systems was adopted for modelling and characterization. The black-box approach entails empirical mathematical modelling of the camera response and noise through measurements and subsequent data analysis. A 3D noise model was employed to characterize camera noise in terms of orthogonal noise components, and an empirical temperature-dependent model was deduced for each component. The method of modelling through measurement is discussed, and the accuracy of specifically the empirical noise models is shown. The cameras are also compared in terms of measured noise performance.