Some Key Techniques Of Target Tracking In Defense

The development of Active Protect System is becoming an important task of countries all around the world. Based on the project requirement, this paper goes deep into some key techniques of the project. Several algorithms are proposed to solve them, and a set of infrared dim target detection and tracking system is designed, which are composed of infrared imaging sensor, high-speed image information processing system and a precise servo. This paper could be divided into six chapters, which are organized as follows:Chapter one is the introduction of the paper. The background and significance of the research are presented. In addition, the status quo of at home and abroad and the content of this paper are included in this chapter.Chapter two concentrates on the study of detection method. The outfield data collection scheme is given, then, some detection methods are validated based on the practical data. Two methods, M-TR method and improved maximum background prediction method, for detecting dim target are proposed. The analysis and comparison between the traditional methods and proposed ones are given in the end of the chapter.Based on the work of chapter two, chapter three develops target tracking algorithms. The first proposed method named FC-MEP(Fuzzy Clustering-Maximum Entropy Principle)is dead against the requirement of real-time tracking for dim infrared target. Another problem is the difficulty of tracking in nonlinear system. In this chapter, method based on Unscented Kalman filter is presented. Experiments show the effectiveness of the proposed methods.Chapter four discusses the track initiation method. A new track initiation method based on fuzzy Hough Transform is presented, in which the measurements are mapped into a fuzzy set in parameter space for the presence of system error and random error. Then, the fuzzy accumulator array in parameter space is computed. Target track is achieved by using fuzzy reasoning.Chapter five mainly discusses data association in multi-target tracking. A new method derived from ACO(Ant Colony Optimization) is developed. According to the characteristics of data association, some conception of parameters in ACO is redefined. Then, the data association model is constructed. In ACDA, distance information, direction information and gray information are integrated to work as the measurement of association. Since the convergence of ACO has not been proved in strict math manner to this day, the system are not ensure to be reliable and stability. We introduce the FCM data association method to solve the problem, which is combined with ACDA to perform the data association.In chapter six, the practical realization of infrared detection and tracking system is designed and analyzed. The software based on the algorithm presented in this paper is accomplished, and the scheme of hardware realization is presented, which is based on DSP and embedded system.Chapter seven summarizes the dissertation systematically, including the content of the work and main achievement in the study. In addition, the work needed to be studied in the future is presented.