Abstract
ABSTRACT Aircraft is a crucial mode of transport with several strategic objectives. It is important to detect the position and number of aircraft in remote sensing images accurately. Most of the current research on aircraft detection focuses on improving algorithms or networks based on public datasets or a single remote sensing image, with limited application to large-scale remotely sensed scenes. Moreover, the detection of small and weak targets is challenging for current detection algorithms. This paper proposes a top-down method for aircraft detection in large-scale scenes. First, multi-source data and spatial constraint rules are used to search for potential airport areas in a wide range, and U-Net is then used to extract the airport area. Finally, this paper proposes a feature enhancement faster R-CNN (FEF-R-CNN) by adding a feature enhancement module to the original Faster R-CNN to improve target detection. This can enhance the aircraft feature while suppressing the background information to a certain extent, for effectively detecting small and weak aircraft targets in the airport area. Top-down aircraft detection experiments were conducted on Hainan Island and Shanghai, and 10 of the 12 airports were accurately detected. Meanwhile, the average precision (AP) of aircraft detection by FEF-R-CNN in the public dataset and four large airports in the study area reached 97.71% and 95.63%, respectively, which are 2.14% and 11.31% higher than that of the traditional Faster R-CNN, respectively. Experimental results verify the effectiveness of the proposed method for aircraft detection in large-scale remotely sensed scenes.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.