| As one of the key techniques of guided weapon, infrared guidance based on image, by using real-time image processing, can trace the target throughout the day and obtain the information which can be used for guiding. As the battle field becomes more complicated and the types and quantities of time sensitive targets increase, there is growing significance of researching on infrared multi-targets tracking techniques in precision strike and targets selection. In this thesis, following researches have been conducted:Based on the aerial time-critical guided bomb, Firstly, the working principle of yaw-pitch half strapdown seeker and particle filter have been proposed and analyzed. As observation model based only on gray-scale feature can't describe infrared targets precisely, the gray-scale feature combined with gradient feature described by histograms of oriented gradient (HOG) has been used as a new model. Adaptive template to the algorithm was applied in order to solve targets'missing problem caused by targets rotating or occlusion, by improving its accuracy and stability. In order to improve the particle filter's real-time capability, we also used multithread implementation to increase computing efficiency of parallel program. The experiment results indicate that the improved method has excellent tracking precision, stronger robustness in tracking, and it also reduces computing time significantly. Our method provides a new theory for the engineering of infrared multi-target guidance and can be used as an efficient tracking method for precision strike.In addition, in order to simulate the terminal guidance of the aerial time-critical guided bomb, a flight experiment is conducted with an unmanned aerial vehicle installed with a yaw-pitch infrared seeker. Using ARM as embedded image parallel processing system of the infrared seeker, the image data are processed and the seeker is guided to aim at the targets. Using this experiment, the merits and disadvantages of this method are discussed. |
PDF Full Text Request