Research On Methods Of Visual Camouflage Technology For Moving Vehicles

From the stealth amulet of ancient myths to invisibility cloaks in science fiction movies,the human pursuit of camouflage and stealth abilities has never stopped.Nowadays,with the advancement of information and material technology,camouflage and stealth technology have been more applied in modern society.Focusing on the reconnaissance of human vision,camera and other visible light means,this paper does research on visual camouflage method used for moving vehicles and aims to show realistic scenery on the outer surface of the moving vehicles.First,the paper introduces the camouflage principle of moving vehicles based on monocular vision and the monocular camera is calibrated.Considering air reconnaissance,a camouflage method of upper direction is proposed and considering ground reconnaissance,a lateral camouflage method is proposed.The imaging model of camera and the meaning of various parameters and coordinates are introduced.The carbody coordinate and the world coordinate are established,and the camera is calibrated using a two-dimensional checkerboard target.Then,the paper realizes the acquisition of single frame camouflage images in different directions of moving vehicles.Two methods of calculating the perspective transformation matrix by using the camera's internal and external parameters and four corresponding points are introduced.For the upper direction,the robot's view captured by the camera is converted into airscape using inverse perspective transformation,and the perspective distortion is eliminated.Then,the camouflage image of upper direction is obtained by cutting the appropriate size area on the airscape.For the lateral direction,the observation model of human vision to vehicles based on central projection model is given,and the camouflage image of lateral direction is obtained through focus adjustment and image cutting.In the following,based on the acquired single frame camouflage image,the real-time camouflage of the moving vehicle is realized.The images collected by the camera on the upper are ahead of the vehicle position,so the real-time display method of advance acquisition and lag display is adopted.The paper gives the formula for calculating the lag time under the condition of uniform speed and non-uniform speed.While the images collected in the lateral direction are synchronous with the vehicle,the method of real-time acquisition and display is adopted.A one-dimensional motion system is used to simulate the movement of the vehicle to verify the effectiveness of the real-time camouflage method.Finally,the research of image restoration is studied considering the fact that the high speed or the vibration of the vehicle itself causes the motion blur.Based on the cepstrum of blurred image,edge detection is used to identify blur extent and the randon transform is used to identify blur direction.After identifying the parameters of motion blur,the Wiener filtering is used to restore the blurred image.In the experiment,the contrast of the camouflage images generated from the blurred image before and after the restoration is used to verify the necessity of the restoration of the blurred image and the validity of the method used in this paper.