The Precise Location Method for Initial Target
Based on Fused Information
Yan Junhua
(&)
, Xu Zhenyu, Zhang Yin, Yang Yong, and Su Kai
College of Astronautics, Nanjing University of Aeronautics and Astronautics,
Nanjing 210016, China
yjh9758@126.com
Abstract. If there is a large error in the initial position of the tracking method,
the tracking performance will decrease. This paper proposes an accurate locating
algorithm for initial tracking target with fusion of object-like information, saliency information, and priori information to provide accurate initial tracking
target for further tracking method. Firstly, the probability distribution of objectlike information is calculated by Edge Boxes which is used to detect the objectlike regions. Then, the probability distribution of saliency information is calculated by spectral residual method. And the priori probability distribution is
calculated according to the target initial position determined by the target
detection. Finally, these three probabilities are adaptively fused to build multiple
clues fused probability distribution. The precisely located initial tracking target
area is the Edge Boxes object-like box with the highest average probability of
fusion probability distribution. Experiments show that the accurate location
method in this paper can achieve accurate relocation of the target in the case of
selection box drift and scale changes and has good real-time performance.
Keywords: Unmanned aerial vehicle Á Target tracking Á Target relocation Á
Object-like detection Á Pixel saliency
1 Introduction
In UAV target tracking, the initial target position of the tracking module is determined
by the target detection module. When the UAV quickly flies approaching the target, the
movement state changes violently, and it is difficult to accurately locate the initial
tracking target. At this time, there are a large number of errors and background noises
in the initial tracking target.
Studies have shown that if background noise and position drift are introduced in the
initial tracking target, the performance of most tracking methods will be severely
reduced. In [1], the marked target box is scaled, and the range is [0.8, 1.2]. Taking the
image in the box area as the initial tracking target, it was found that the performance of
the tracking method was significantly reduced. In [2], the initial target tracking frame is
shifted, and the offset is 10% of the length and width of the marked target tracking. The
performance of the tracking method will also be significantly reduced. Although the
tracking method [3] has certain tolerance to the initial tracking target shifting, the
tolerance performance is limited, and it can only respond to small error and noise.
© Springer Nature Singapore Pte Ltd. 2020
Q. Liang et al. (Eds.): Artificial Intelligence in China, LNEE 572, pp. 563–575, 2020.
https://doi.org/10.1007/978-981-15-0187-6_67
Based on Fused Information
Yan Junhua
(&)
, Xu Zhenyu, Zhang Yin, Yang Yong, and Su Kai
College of Astronautics, Nanjing University of Aeronautics and Astronautics,
Nanjing 210016, China
yjh9758@126.com
Abstract. If there is a large error in the initial position of the tracking method,
the tracking performance will decrease. This paper proposes an accurate locating
algorithm for initial tracking target with fusion of object-like information, saliency information, and priori information to provide accurate initial tracking
target for further tracking method. Firstly, the probability distribution of objectlike information is calculated by Edge Boxes which is used to detect the objectlike regions. Then, the probability distribution of saliency information is calculated by spectral residual method. And the priori probability distribution is
calculated according to the target initial position determined by the target
detection. Finally, these three probabilities are adaptively fused to build multiple
clues fused probability distribution. The precisely located initial tracking target
area is the Edge Boxes object-like box with the highest average probability of
fusion probability distribution. Experiments show that the accurate location
method in this paper can achieve accurate relocation of the target in the case of
selection box drift and scale changes and has good real-time performance.
Keywords: Unmanned aerial vehicle Á Target tracking Á Target relocation Á
Object-like detection Á Pixel saliency
1 Introduction
In UAV target tracking, the initial target position of the tracking module is determined
by the target detection module. When the UAV quickly flies approaching the target, the
movement state changes violently, and it is difficult to accurately locate the initial
tracking target. At this time, there are a large number of errors and background noises
in the initial tracking target.
Studies have shown that if background noise and position drift are introduced in the
initial tracking target, the performance of most tracking methods will be severely
reduced. In [1], the marked target box is scaled, and the range is [0.8, 1.2]. Taking the
image in the box area as the initial tracking target, it was found that the performance of
the tracking method was significantly reduced. In [2], the initial target tracking frame is
shifted, and the offset is 10% of the length and width of the marked target tracking. The
performance of the tracking method will also be significantly reduced. Although the
tracking method [3] has certain tolerance to the initial tracking target shifting, the
tolerance performance is limited, and it can only respond to small error and noise.
© Springer Nature Singapore Pte Ltd. 2020
Q. Liang et al. (Eds.): Artificial Intelligence in China, LNEE 572, pp. 563–575, 2020.
https://doi.org/10.1007/978-981-15-0187-6_67