Aero-image Processing And Multi-target Tracking On Moving Carrier Base On Wavelet Transform

Multi-target tracking is widely used in military field. Such as trajectory missile recovery, aero-guard, multi-target attacking in air, surveillance on sea and so on. In the close years, surveillance equipments of infrared and visible light are widely used in military field, so the technology of multi-target tracking is developed rapidly. Firstly, the methods of multi-target tracking are summarized. The method of detecting and tracking targets based on wavelet transform is emphasized. In this paper, a method of magnifying image based on two dimension discrete wavelet is introduced. Comparing with traditional interpolation methods, it not only enhanced resolution of the aero-image, but maintained the detail character of the image. Secondly, two methods of targets detecting based on wavelet transform are suggested in this paper. One is targets detecting method based on wavelet energy aiming at nature texture background. Compared with traditional segmentation method, it can remove background, which gray is close to targets, at the same time, targets is reserved. The other is arithmetic based on wavelet transform multi-layer block matching aiming at characteristic background. Compared with traditional block matching tracking arithmetic, because combining of coarse orientation and accurate orientation, it overcome some disadvantages, such as a mass of operation, inaccurate matching in the low SNR image. Otherwise, fuzzy reasoning method is used for data association. The arithmetic is mainly used while some targets fall in one tracking door. Compared with traditional arithmetic of data association, it's operation speed is faster, and it is better fit for engineering application. Combining Kalman filter, the experiment of multi-target tracking and forecasting is done by this arithmetic. The experiment result is satisfactory. Finally, the tracking characteristic of three mobile target model is researched by emulational experiment under several mobile condition. The advantages and disadvantages of these mobile target model are compared. At last, the develop direction of multi-target tracking is forecasted.