Capturing And Tracking Of Simulative Targets In The Missile Trainer

The moving target real-time detection and tracking is an important problem, which almost includes all aspects of image processing and analyzing. It is used widely in computer vision, video early-warning, machine navigation, target recognition, secure monitoring, TV guidance, and so on.At present, the missile aiming simulative trainer is based on infrared guidance theory. But the aiming training is so special that the trainer has many shortages, such as high price, few function, low stability, difficult tending, etc. Aiming to such shortages, it is discussed that a new missile aiming simulative trainer, based on the technology of target detection and tracking, substitute for the available trainer in the paper. In the new trainer, a television camera will replace the seeker of a missile, a computer will replace the guidance and tracking circuit in a missile. In the computer, the real-time video images are processed by the computer where many technologies are synthetically used, such as image filter, image segmentation, target detection and target tracking. The new trainer will have low price, high stability, complete functions, easy upgrade.The moving target real-time capturing and tracking is emphasized in the trainer.Firstly, the technology of moving target detection is discussed: the three common ways are introduced. On the conditions of dynamic background and real-time processing, we select the frame difference method through comparing the three ways. The two ways of corner detection are introduced. According to the practice of system, affine transformation is selected. In many algorithms of solution of coefficients of affine transformation, the robust LAD method replace the normal LMS method that is easy to be influent by outer point. Secondly, the paper introduces correlation techniques of post-processing of detection and target tracking way based Kalman Filter.Lastly, the paper put forward the project of the new training simulator: Firstly, we make use of SUSAN method to detect corners of image. Considering the velocity and direction of actual targets, we form the relation between corers and chose the best affine coefficients through the LAD method. The target moving area has been gotten via making affine transform for images and frame difference method. Secondly, the difference images are processed using many methods, for example binaryzation, morphology filter, and so on. At last, the way of target tracking is used for abstracting targets ,tracking targets and getting target parameters. The key arithmetic of the paper is come true through using OpenCV in the VC++6.0 programme environment. The tests indicate that the algorithm can get enough precision and fast velocity which are requested by the training simulator.