Adaptive Dynamic Target Tracking System Research And Design

Involving numerous subjects, and having been applied widely both in military and civic fields, locomotive target tracking is a hot research in the field of computer vision and signal processing. As the technology of DSP develops rapidly, the research of the algorithms on locomotive target detection and recognition and locomotion forecasting has been a key issue to track locomotive target. This thesis mainly brings forth a new algorithm in locomotion forecasting and then designs a practical system of adaptive dynamic target tracking with it.First, the dissertation studies the principles and some common algorithms of target tracking including contrast tracking, correlation tracking and cepstral tracking, ect. Then it analyzes their advantages and disadvantages. Secondly, it constructs a high speed hard and soft ware disposition platform utilizing SEED-VPM642/IV TMS320DM642 visual disposition system produced by SEED company. Based on these, it presents an adaptive α-β-γ filter design to solve the contradiction between tracking precision and convergence rate that traditional α-β-γ filter has and the limitation of unable to track high locomotive targets due to stable coefficient. This adaptive α-β-γ filter can use tracking error vector to adjust the output of the Fuzzy Logic System according to the variety of locomotive target to get the adaptive coefficients so that the tracking results converges quickly on the premise of assuring the tracking precision. This system can help to track targets with high locomotive performance. At last, With the adaptive α-β-γ filter worked out in the thesis as the forecasting algorithm, combing improved Centroid Track Algorithm, and making use of the SEED-VPM642/IV TMS320DM642, the thesis implements a Centroid Tracking System based on the adaptive α-β-γ filter. According to method which evolves correspondingly with the Centroid tracking, the system can track the targets precisely.