Research On Aerial Object Tracking Algorithm Based On Multi-tracker Relay

Aerial object tracking technology is widely used in military missile guidance,air traffic control and drone counters.In practical applications,there are various interference factors in the aerial object tracking process,such as illumination variation,motion blurring,attitude variation and occlusion.How to achieve robust tracking of aerial objects under different interference conditions is the main problem studied in this thesis.The tracking algorithm using only a single feature is difficult to adapt to the complex tracking environment.To solve this problem,this thesis proposed a multi-feature fusion tracking algorithm based on confidence evaluation.The algorithm chooses various features with different advantages to track separately,and achieves multi-feature reliable fusion by evaluating the confidence of different features.At the same time,in order to solve the problem of large variation of aerial object scale,a parallel scale estimation strategy based on the Discriminative Scale Space Tracking algorithm is proposed in this thesis.This method estimates the optimal width and height of the object respectively,which can estimate the scale variation of aerial object more stably.For the occlusion problem,an occlusion-aware part-based model is proposed in this thesis.In this model,the object is divided into several local parts,and the confidence of each part is evaluated by its stability and similarity with the whole object.Then the parts having high confidence are used for realizing anti-occlusion tracking.The part-based local model can deal with the occlusion problem effectively,while the global model is more suitable for dealing with the fast motion and motion blur of the object.Therefore,a tracking method based on multi-tracker relay is proposed in this thesis.In this method,the tracking state is judged according to the confidence of the model.In the normal tracking state,the global model based on multi-feature fusion is used,and when occlusion interference exists,the tracking model is replaced by the parts-based local model.In this way,the algorithm can effectively deal with various tracking situations.In this thesis,the data set of aerial object tracking is used to verify the performance of the algorithm proposed,and the performance of different algorithms is compared and analyzed in various evaluation methods.Experimental results show that,compared with other algorithms tested,the proposed algorithm achieves overall performance improvement and is robust to interference factors such as attitude change,scale variations and occlusion of the aerial object.