Research On Location Solver Of Motion Target Based On Image Technology

In the modern military competition, the aerial refueling technology reflects a country’s Aircombat ability and its defense capacity. At present, the flexible aerial refueling technology that ourcountry uses needs the cooperation between the aerial refueling commander and the pilot toaccomplish the complex and dangerous docking process. Thus, it is necessary to develop a set ofassistant aerial refueling guidance system. The thesis researches on the detection、tracking andpositioning of the moving taper sleeve with iamge technology.Firstly, the methods of preprocessing refueling images are studied. The images are turned toHSV model from RGB model. Besides, threshold segmentation、closing operation and Gaussian filterare researched in V channel. For convenience of providing image features for detection, the thesisputs forward a sector-by-sector image-preprocessing solution which applies distance information.Secondly, the methods of motion detection are studied. According to the different deformationdegrees of the taper sleeve in different distances, a method of detecting sector by sector is put forward.In the long distance, the secondary detection of the target is researched which is based on advancedRandomized Hough Transform and Least Squares Fit. In the short distance, two methods are providedin order to solve the great deformation of the taper sleeve and lead to good result. One is gettingexternal contour by GVF Snake model and the other is getting internal contour by geometricconstraint.Furthermore, Kalman-based motion tracking method is studied, which begins from the analysisof the sequence of the video images, builds the Kalman model of the taper sleeve, discusses the initialsetting of the model parameters, brings in temporal occlusion coefficient to accomplish the automaticadjustment of the parameters and puts forward two detecting and tracking methods in the longdistance and short distance models.Finally, the spatial position calculating of the motion target is researched. It uses the pinholecamera model, deduces the mathematical formulas of the moving target’s position calculating, buildsthe system of real-time collecting of images and provides the experimental platform to test themethods. The aerial refueling docking process is simulated under different outdoor backgrounds andillumination. The experiment results show, the methods the thesis put forward meet the requirementsof the tracking effect and measuring accuracy basically and have the capacity of real-time processing.