Study On Recognition And Tracking Of Military Motion Targets

The recognition and tracking of motion targets have important research value to national economy, military field, scientific technical research and project application etc. Motion object recognition and tracking in image is the key technology of the precision guided munitions, which need to be solved urgently in the world. Target recognition and tracking technology has become a hot topic of research in the automatic control field, computer vision field and pattern recognition field. The thesis "Mobile Military target recognition" is proposed on the practical background of Second Artillery scientific research department project.In this thesis, significance for the Second Artillery's development and the domestic external study and development for motion targets are first introduced. First of all, the weapon information on the other's arms is obtained by using reconnaissance technology; and stored in database. Second, because the past recognition and tracking systems often use a single identification and tracking method with unreasonable length of time, a new idea is proposed based on the needs of actual combat, in other words, data in database is used in two steps. First, after moving targets are detected, the targets are preprocessed by the techniques of image enhancement, noise filtering, edge detection, then compare them with the information in database provided by inleligent department in accordance with different types of images. If the types of weapons are known, only pattern matching is needed for locating the target quickly, then track and monitor it in accordance with the data in database and its spatial location. Thus a lot of time is saved. Second, if the matching data with the target detected is not found in database, we have to identify it based on motion analysis method firstly, then track it. Eventually the results will be transmitted to the commanding officers to make the right decision.According to above ideas, a simulation software for moving targets recognition and tracing system based on Matlab6.5 is implemented. The system is efficient and feasiable for image's filter, enhancing, edge detecting, recognizing several kinds of fly objects and tracking a serial fly objects. Nowadays, each of methods and algorithms mentioned above is written in a subroutine by Matlab 6.5. In the future, they can be directly used in further research fields such as video surveillance, navigation and so on.