Adaptive Multi-feature Fusion Improved ECO-HC Image Tracking Algorithm Based on Confidence Judgement for UAV Reconnaissance

Abstract

Owing to the advantages of target tracking algorithm based on correlation filtering on the efficiency in speed and the tracking accuracy, it has been widely applied in the image real-time tracking, especially in the field of the video surveillance, the human-computer interaction, and the intelligent transportation. However, considering the tracking timeliness and the ability to deal with deformation, position changes and occlusion in complex background, especially in the application for the unmanned aerial vehicle (UAV) reconnaissance, an adaptive multi-feature fusion improved tracking algorithm with confidence judgement strategy is proposed in this paper. On the basis of the efficient convolution operators handle-crafted (ECO-HC) method, which is the best algorithm with excellent performance based on correlation filtering, an adaptive multi-feature strategy and a strategy of confidence state recognition with confidence judgement and relocation are described in detail. After quantitative and qualitative comparison tests with other advanced algorithms, the results fully verify the superiority in tracking accuracy and robustness against the interference of the complex background.