Study On Rocket Recognition And Tracking Control System Based On Video Image

A while after the rocket launching, the visibility of the rocket is a very important part for the command decision-making. Especially, when some launching errors occur, such as Eying error, the traditional outside detecting systems are hard to capture the target, so hardly the command controller can get the rocket Eying data, let alone making the correct decision. So, at the launching phase of the rocket, using the CCD camera for auto-tracking is more important.In the paper, based on the study of a moving aircraft, some advanced technologies of imaging tracking in recent years are studied. A high efficient and real-time method of imaging tracking is deduced. By this,realize the rocket servo tracking based on the continuous video moving image which is for the purpose of providing the rocket launching phase integrity the flight data and the picture record.In view of the rocket launch scene is hard to be distinguished in the daytime, This paper discussed the choice of tracking algorithm under different backgrounds. At the beginning, we used YUV color criterions to judge whether the rocket image is covered by smog: if the smog ratio surpasses the threshold, use the correlation algorithm unifies SSDA fast algorithm, multi-template matching as well as the template updating strategy, which improves the precision and overcomes the unsatisfied real-time performance; If the smog ratio has not achieved the threshold, sets the central point of rocket outline to be the tracking control point, we used the digital image processing technologies such as preprocessing, edge detection, improved two-dimensional Otsu binarization and 8-neighborhood image thinning. Moreover, considered the possiblilty of track rejection, the template mismatch and other suddenly situations, we take the Kalman filter to forecast to the location, thus maintain the tracking stability and the continuity. Take the different position of target and the camera center as fuzzy PID controller's input, which control camera auto-tracking the rocker image.The hardware and software of the system which are presented in this paper have passed test in lab, the experimental results show that the system has characteristics, including automatic recognition of the rocket under different backgrounds; permitting tracking and prediction when the rocket become shaded. The results prove that the system is effective and can be implemented in real time.