Design Of Infrared Real-time Simulation System For Photoelectric Interference Coupling Scene

The wide application of infrared technology makes infrared simulation technology have important research value.With the rise of deep learning,there is an increasing demand for rapid and large-scale generation of high-reliability infrared image samples under a variety of imaging conditions.For the problem that the infrared image obtained by actual shooting is costly and the external conditions are difficult to control,the thesis has developed a system that can generate a large number of high-reliability infrared images in real time.The research work is listed below:Firstly,to improve the modeling accuracy,the thesis proposes a multi-granularity material classification method for the problem of large workload and low efficiency of largescale remote sensing image material classification.It effectively improves the efficiency of remote sensing image high-precision material classification by at least one time.Secondly,to get more accurate surface temperature data of surface materials,the physical model is established based on the principle of heat transfer for each type of material according to the surface material characteristics.The accuracy of the thesis calculation model was verified by the measured temperature data.Thirdly,to improve the realism of infrared simulation,the atmospheric transmission and detector imaging process are studied and simulated.The thesis proposes a convenient method for calculating atmospheric transmittance and path radiation.The geometric distortion,vignetting,blurring,and noise effects which will interfere with image quality during the imaging process are simulated according to the imaging principle of the detector.The thesis proposes a new radiance quantization method to improve the existing quantitative methods cannot show the time and weather changes.And the infrared texture is modulated based on the visible light remote sensing image texture to generate a realistic infrared image.Finally,the infrared smoke interference effect is simulated for the actual battlefield environment.Aiming at the actual application rendering rate requirement,the thesis proposes a parallel simulation rendering method for imaging simulation.The combination of CPU and GPU is used to realize infrared radiation energy calculation,quantization and imaging interference effect simulation on the GPU,which greatly improves the rendering rate.On this basis,a complete infrared real-time simulation system is designed and implemented.By comparing with the results of the existing simulation methods,it is proved that our system can generate high-reliability infrared images and reach the simulation level of foreign infrared mature commercial software.The system has been officially put into use by the department of defense.The actual use indicates that the system can simultaneously generate realistic infrared simulation images of arbitrary observation angles and multiple conditions in real time with 12 cameras.