Near Infrared Scene Simulation Based On Visible Image

Near infrared band is between infrared and visible band, which has the advantages of good penetrating, insensitive to temperature, etc. It is widely used in the research of camouflage and anti-camouflage detection technology in military field. And near infrared image can obtain the useful information which cannot be obtained by the visible and infrared band, which can adapt to the influence of the bad weather conditions and has important theoretical research significance and the strong scientific research application value.Target reflectivity characteristic feature is the basic parameters of the near-infrared scene simulation, this paper introduces a fast and accurate method for image reflectance inversion. First, in order to obtain the linear correspondence between gray and real scene irradiance, the relationship between gray image and the illumination of the camera pupil is analyzed, then correct the gray level of the image; Secondly, acquire the reflectance by using CMOS camera; Finally the reflectance of the corresponding scene is then performed in an image processing step. We demonstrate that the proposed method can produce reasonable results with an inversion uncertainty of less than 10% compared to the results derived from exact measurements.The simulation precision is affected because of the sun radiation energy of the visible and near infrared bands in the real scene are different. In order to establish an accurate model of visible and near infrared spectroscopy model, the energy distribution of the real scene in the visible and near-infrared bands is calculated. The instantaneous solar radiation is calculated by establishing the solar angle model at different time. The atmospheric transmittance of the near infrared band is calculated by MODTRAN based on the analysis of radiation in the atmosphere. Reindl model is used to calculate the atmospheric scattering rate. Finally, the visible and near infrared surface radiation energy are obtained. Using this method, we can calculate the surface radiation energy in the surface of earth at different time of any wave band.Finally, based on the similarity of visible and near infrared imaging principle, a near infrared simulation model is established,which use the reflectance data of visible and near infrared band, the surface radiation energy data and gray level of visible image as the parameters. In the process of the complex scenes simulation, the image segmentation algorithm based on graph theory minimum spanning tree is used to classify different materials in complex scenes, the near infrared scene is simulated based on visible image combining with the simulation model. In order to accurately describe the variation of the simulation image of different materials, the near infrared image is processed by the method of edge scan and smooth. Compare with real near infrared image, the simulation results highly similar to the real image and can better reflect the global characteristics of near infrared image. In this paper, this method create the conditions for the real time near infrared scene simulation.