Design of video aerial target detection and tracking system based on AM5728

A detection and tracking system for aerial target in the dynamic background is designed based on AM5728 in this paper. Firstly, a moving target detection algorithm based on LK optical flow and inter-frame difference is proposed. According to the number of FAST feature points, the foreground target is extracted by inter-frame difference, or the matching points are calculated by the LK optical flow algorithm to obtain the registration image, and then the foreground target is extracted by the image difference. The false target is removed by 8-connected components labeling and multi­frame association in order to obtain the target in the current frame. Then, the KCF tracker is initialized by the result of the detection algorithm and tracks the target stably. Finally, target detection and tracking algorithms are transplanted to AM5728. Experiments demonstrate that the video aerial target detection and tracking system can obtain dynamic information of the aerial target in real time and track aerial target stably for a long time under the conditions of scale changes, light changes, etc.

[1]  Cai Qing-Yao,et al.  Application of Robust Regression Estimation in Moving Background Compensation , 2016, 2016 Sixth International Conference on Instrumentation & Measurement, Computer, Communication and Control (IMCCC).

[2]  Tom Drummond,et al.  Machine Learning for High-Speed Corner Detection , 2006, ECCV.

[3]  David G. Lowe,et al.  Distinctive Image Features from Scale-Invariant Keypoints , 2004, International Journal of Computer Vision.

[4]  Rui Caseiro,et al.  Exploiting the Circulant Structure of Tracking-by-Detection with Kernels , 2012, ECCV.

[5]  Rui Caseiro,et al.  High-Speed Tracking with Kernelized Correlation Filters , 2014, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[6]  Michael Felsberg,et al.  Discriminative Scale Space Tracking , 2016, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[7]  Luc Van Gool,et al.  SURF: Speeded Up Robust Features , 2006, ECCV.