Research On Some Key Techniques For Image Homing Guidance Based On Strap-down Platform

With the fast development of modern Infrared(IR) imaging technology and automatic control technology for guidance, infrared homing guidance information processing systemes(IRHGIPS) are used by interceptors and precise guided weapons more and more. With the inertial measurement acquired from the strap-down inertial sensor, the kinematic information of target in inertial space can be abstract from the images captured by camera on the same platform, and it is a novel idea that detecting and tracking by inertial information togother with imagery, which might be an important research area in the future.This thesis focuses on the techniques for IRHGIPS used by strap-down guided platform, with respect to the decoupling for staring imaging system, the detection of small target in clutter background imagery captured by moving camera, near range imaging tracking with compound technique and the design techniques for real-time information process system.The motion of camera platform make the target images deviate from theoretical location, which makes IRHGIPS can not analyze and use the motion information of targets. Analyzing the relationship between target imaging, motion and rotation of imaging platform, an exact decoupling algorithm is proposed. On the premise of small rotation, a simplified decoupling algorithm is proposed to solving the problem of high consumption on computation recourses of extact decoupling, and then the composition and propagation of error in decoupling are analyzed. By employing the inertia measurement, the inertia decoupling algorithm is more precise and reliable, and the simplified algorithm also has the advantages such as simple structure, easy to be accelerated by hardware.The suppression of clutter in background and the detection of dim target are the two important issues in homing guidance detecting system. A background prediction algorithm based on two dimensional least mean square filter (TDLMS) is proposed with the fusion of prediction images in different directions which is different from he traditional adaptive spatial filter which use the correlation in one direction only, and the proposed algorithm can suppress the clutter in background better. The motion of camera can make the detecting algorithm based on trajectory assocaition invalid. To overcome this problem, a novel detecting algorithm based on recursive decoupling and least square (LS) prediction trajectory association is proposed. This algorithm uses the attitude angles between two frames to decouple, which avoid using large angles to decouple and ensure the precision and reliability of detection.When target comes close to the camera, the trajectory of target in imagery becomes nonlinear. To solve the near range tracking problem, a bearings-only tracking system model is proposed in this thesis, and then the tracking algorithm based on two-variable extended kalman filter (EKF) and unscented kalman filter (UKF) are employed. When near rang tracking, the image features are going to be valid too. But usually the features in IR images are poor and unstable.To overcome these shortages of IR imagery, firstly a nonlinear re-quantizing method for IR image is proposed to extend the feature characterization of target, and then a novel multi-kernel cascading scheme for mean shift algorithm is given so as to improve the feature expression of target in mean shift procedure. Finally a novel tracking scheme for IR target based on bearings-only tracking and improved mean shift algorithm is achieved.At the end of this thesis, the structure of IRHGIPS is given, and then the schedule algorithm of tasks is discussed, the accelerating logic circuit of the key algorithm is designed too. The main works done in this paper have been used in the IRGIPS of some important project. Test results indicate that this system is fast and precise.