Journal of Applied Optics, Volume. 45, Issue 2, 346(2024)

Infrared target detection algorithm based on improved Faster R-CNN

Xichen WANG1... Fulun PENG2, Yexun LI3 and Junju ZHANG1,* |Show fewer author(s)
Author Affiliations
  • 1School of Electronic and Optical Engineering, Nanjing University of Science and Technology, Nanjing 210094, China
  • 2Xi'an Institute of Applied Optics, Xi'an 710065, China
  • 3Jiangsu North Huguang Photoelectric Co.,Ltd., Wuxi 214100, China
  • show less

    In order to improve the detection accuracy of infrared targets, a Faster R-CNN infrared target detection algorithm introducing a frequency domain attention mechanism was proposed. Firstly, a parallel image enhancement preprocessing structure was designed to address the issues of edge blur and noise in infrared images. Secondly, a frequency domain attention mechanism was introduced into Faster R-CNN, and a new infrared target detection backbone network was designed. Finally, a path enhanced pyramid structure was introduced to fuse multi-scale features for prediction, and the rich location information of the underlying network was utilized to improve detection accuracy. The experiment was conducted on a dataset of infrared aircraft. The results show that the AP of improved Faster R-CNN target detection framework is 7.6% higher than that of the algorithm with ResNet50 as the main stem. In addition, compared with current mainstream algorithms, the proposed algorithm improves the detection accuracy of infrared targets and verifies the effectiveness of the algorithm improvement.

    Keywords
    Tools

    Get Citation

    Copy Citation Text

    Xichen WANG, Fulun PENG, Yexun LI, Junju ZHANG. Infrared target detection algorithm based on improved Faster R-CNN[J]. Journal of Applied Optics, 2024, 45(2): 346

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Research Articles

    Received: Apr. 11, 2023

    Accepted: --

    Published Online: May. 28, 2024

    The Author Email: ZHANG Junju (张俊举(1979—))

    DOI:10.5768/JAO202445.0202001

    Topics