DESIGN And Research On Stabilized Platform Pointing And Tracking System

With the development of modern weapons, aircrafts are operating on a higher and higher altitude. Then how to exactly identify on-ground targets and make precise blow have become a problem needed to solve urgently. On-board pointing and tracking system help the pilot by conducting capturing, locking then tracking on the target on ground. Based on design task of the pointing and tracking system of stabilized platform, a deep study on visual tracking and visual based pose estimation algorithms was done in this thesis. Later a pointing and tracking system is designed based upon the algorithms proposed.Firstly, Mean Shift based tracking scheme was adopted. Advanced Mean Shift tracking methods was proposed to enhance the description ability as well as the adaptive scaling ability of the Mean Shift tracking algorithm. Description ability was enhanced through adaptive division of the color space, and LBP texture description added to the tracking model. Scaling ability was enhanced by adaptively choosing the feature scale based on Linderberg's scale selection theory. Through these enhancements, tracking ability under complex environment and of objects with large scale changes was greatly improved.Secondly, visual based pose estimation algorithm was proposed for the system. The proposed algorithm gives estimation of the inter-frame Euler angels of the platform. Detection and matching of feature points was done in the SIFT framework. Feature points detected were optimized through a RANSAC process. Simulation results showed that the average absolute estimation error of the proposed algorithm from 1000 tests can reach(±1 .5 ,±3 .5 ,±0.5 ) deg with Euler angles lie within (±15 ,±1 5 ,±15 ) and Signal to Noise Ratio(SNR) equals to 20dB.Thirdly, compensation ability of the Kalman predictor and adaptive LMS predictor was studied relating to the latency introduced by the tracking and pose estimation module. Based on their performance, an adaptive LMS compensation module was developed. Tracking error of the position loop showed that the adaptive LMS predictor effectively compensate the latency, thus enhanced the stability and tracking accuracy of the system. Finally, a tracking and pointing system was designed through a module-based schematic work. Video capturing module, image processing module and serial communicating module was designed based on their functions. Then system testing was given to the developed system, which shows that the system performance well through the whole capturing, tracking, pose estimating process.