Target detection algorithm based on deep learning has become a hotspot with its practicability and adaptability. However, due to the small scale of small targets and lack of feature information, it is weak to effectively detect small targets in air by deep learning algorithm. To solve this problem, this paper proposes a novel detection scheme for small aerial targets based on salient region. This detection scheme combines a haze removal algorithm, a skyline detection algorithm and a Frequency-Tuned (FT) salient region detection algorithm to detect small targets in air. Through the usability of the haze removal algorithm, a relatively clear image can be obtained for subsequent detection processing. Next, by applying the skyline detection algorithm to the image obtained in the previous step, the sky background region with simple pixel level features can be obtained, which is free from the interference of complex ground features on the aerial target detection. Finally, the Frequency-Tuned salient region detection algorithm is used to detect the significant pixel area in the sky background to obtain the target position. Experimental results demonstrate that the detection scheme proposed in this paper can effectively detect small aerial targets.
Research on Small Aerial Target Detection Based on Salient Region
Lect. Notes Electrical Eng.
International Conference on Autonomous Unmanned Systems ; 2021 ; Changsha, China September 24, 2021 - September 26, 2021
Proceedings of 2021 International Conference on Autonomous Unmanned Systems (ICAUS 2021) ; Chapter : 69 ; 696-705
2022-03-18
10 pages
Article/Chapter (Book)
Electronic Resource
English
Research on Small Aerial Target Detection Based on Salient Region
British Library Conference Proceedings | 2022
|Multiscale salient region-based visual tracking
British Library Online Contents | 2017
|A dense subgraph based algorithm for compact salient image region detection
British Library Online Contents | 2016
|A dense subgraph based algorithm for compact salient image region detection
British Library Online Contents | 2016
|