Research On Low-light Level Target Visual Simulation And Detection And Tracking Technology Based On VieWTerra

With the development of the low-light-level(LLL)night vision imaging technology,various kinds of military weapon systems and civilian equipment based on the principle of LLL imaging have been developed.Moreover,the performance and design time of night vision system can be improved with the LLL night vision imaging technology.Besides,the simulation system can provide the video for the LLL target detection tracking technology,which can further promote the development of detection and tracking technology.The vieWTerra platform was developed with Visual C++ program language,OpenGL graphics application interface and OpenCV computer vision library to realize the simulation of LLL night vision system.Firstly,the theory of LLL night vision system is described.Moreover,the mathematical models of the corresponding physical effect are established based on the transfer process of LLL radiation and the imaging mechanism of ICCD;Secondly,a visual scene simulation system is developed with vieWTerra and 3ds Max,which can produce the visual scene that can be regarded as the video source of ICCD simulation.To ensure the real time of the simulation of LLL imaging,the two preprocess methods are proposed before the simulation of LLL night vision system:the establishment of the LLL atmospheric transmittance database and the target texture material setting.Furthermore,based on the theory of LLL night vision,,the signal transfer characteristic,the spatial transfer characteristic and the noise characteristic are handled with the signal transfer characteristic mathematical caculation equation,the optoelectronic imaging system model based on modulation transfer function,the Gaussion noise,respectively.The system achieves the purpose of generating LLL video image based on the software simulation.Finally,the target in the LLL simulation video can be detected and tracked through an improved Gaussian Mixture model and an algorithm that combins Kalman filter and multi-feature Mean Shift,which verifies the practicability of the LLL simualtion video.