Aerial Target Tracking System Based On DSP

Moving target recognized and tracked is an important research task in applied visual field. It has been widely used in military filed such as plane's navigation, weapon control and guiding system. In this thesis, the aerial target tracking system is under the complicated, contrast lower background to track the flight targets (such as aircraft, missiles, etc.). It is very significant on military affairs.Design of the tracking system is divided into two parts, which are hardware design and software algorithm design. Hardware design is based on DSP Development Board. The hardware design including:1. Under the structure of system, the whole tracking system can be divided into six tasks, which are image input, image processing, image output, tracking equipment control, network communication and parameters setting. In DSP integrated development environment the tasks are configured by DSP/BIOS, in order to complete the reasonable scheduling of tasks.2. The input and output interfaces of DSP Development Board are configured to realize communications between DSP Development Board and other equipments in the tracking system, including the controlling system and tracking equipment.3. PID controller is designed in order that system can control the tracking equipment better. Using the theory of PID control, the tracking equipment is effectively controlled in process of tracking moving targets.Software algorithm of tracking system is divided into two parts, which are target acquisition and target tracking. Target acquisition is the foundation and prerequisite of target tracking. In target acquisition phase, images of the target are recorded by CCD, and then the images will be identified from the background through image preprocessing, image segmentation, parameter extraction and target matching.The main content in target acquisition process includes:1. Median filtering algorithm in image preprocessing stage is improved and realized with DSP.2. Method of maximal between-class variance is used to determine the optimal threshold value of image segmentation. And method of region growth is also improved.3. Extract the target's parameters, and compare three edge detection algorithms with DSP. The most suitable algorithm is chosen and used in the system. Forecasting the location of target is the major work in target tracking stage. In this thesis, linear forecast algorithms is compared with Kalman forecast algorithms. The three-point linear forecast algorithms is chosen and used in the tracking system.Experimental results show that the aerial target tracking system can track moving target stably. It is fully demonstrated that the hardware configuration of tracking system is reasonable and the software is suitable. The whole system based on DSP is smaller than the tracking system designed by PC, and it is believed that this system will have a very good application.