• A CUDA-based Fast Robust Tracking Method of Space Target   [CET 2015]
  • Author(s)
  • Rufei Kang
  • Influenced by spatial complex background, lighting change and relative motion, traditional space target recognition algorithms have poor robustness, large computation and low real-time capacity. Based on CUDA parallel computing architecture, a fast robust tracking method was proposed in this paper. First, on the basis of gray gradient method, it introduced shape matching, area screening and adaptive gray threshold adjustment to improve the robustness of contour extraction. An improved CUDA-based GVF Snake contour tracking algorithm was then designed based on the contour extraction result to improve the real-time capacity of tracking algorithm in terms of the sequential features and computational efficiency of initial contour. The experimental results show that this method ensures the accuracy of contour detection and can significantly improve the computational efficiency and achieve real-time robust tracking.
  • Fast Robust, Space Target
  • References

Engineering Information Institute is the member of/source content provider to