Adaptive grid-based confidence assessment for synthetic optoelectronic images by Physical Reasonable Infrared Scene Simulation Engine (PRISSE)

Abstract

Like visible spectrum, synthetic infrared scenes reflect the invisible world of infrared features. Propagation of a typical infrared radiation involves a variety of sources of different sizes, shapes, intensities, roughness, temperature, etc., all of which would impose impacts on the fidelity of the synthetic images. Assessing the confidence of a synthetic infrared scene is therefore not so intuitive as evaluating the quality of a visible image. An adaptive grid-based method is proposed in this paper for similarity assessments between synthetic infrared images and the corresponding real infrared images, which is on a grid-by-grid basis. Different from many traditional methods, each grid in our work is weighted by a value that is simulated by a 2D Gamma distribution. Introducing adaptive grids and exerting a weighting value on each grid are the aspects of our method. To investigate the effectiveness of our method, an experiment was conducted for taking real mid-wavelength infrared (MWIR) images, and the corresponding synthetic MWIR images were simulated by Physical Reasonable Infrared Scene Simulation Engine (PRISSE). The confidence of similarity assessments evaluated by our method is then compared to some popularly-used traditional assessment methods.