Infrared thermal imaging technology can not only detect ships through obstacles such as smoke, fog, and haze, and realize continuous passive observation day and night, but also observe the external details of ships and subsequently identify and track targets. Therefore, infrared ship target recognition plays a vital role in maritime traffic management and ship navigation safety. Different deep-learning algorithms and infrared images of different bands can easily affect the effect of ship target recognition. This paper selects three representative deep learning algorithms, RetinaNet, Cascade R-CNN and CenterNet, to carry out ship recognition experiments on multi-band infrared ship images. In terms of experimental accuracy, the Cascade R-CNN algorithm demonstrates a high level of recognition accuracy, particularly for small ships. It outperforms other algorithms in terms of recognition accuracy. On the other hand, the RetinaNet algorithm excels in achieving both high precision and fast processing speed. When it comes to experimental speed, the CenterNet algorithm exhibits a faster running speed compared to others. However, CenterNet’s recognition accuracy is slightly lower, especially for small ships. From the perspective of multi-band infrared images, it is observed that long-wavelength Infrared (LWIR) images have a poor recognition effect on small ships. Conversely, short-wavelength infrared (SWIR) images show better recognition performance compared to the LWIR images. By comprehensively considering the accuracy and speed, evaluating the performance of the algorithms in multi-band infrared images, and selecting the best combination to achieve the optimal effect of ship target recognition in practical application, to meet the needs of maritime surveillance.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Performance Analysis of Ship Target Recognition in Multi-Band Infrared Images Based on Deep Learning


    Contributors:
    Li, Yuli (author) / Zhang, Di (author) / Fan, Liang (author) / Ma, Haowei (author) / Xu, Zidong (author)


    Publication date :

    2023-08-04


    Size :

    749947 byte





    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English




    Multiscale Edge Detection for Infrared Ship Target Recognition

    Lu, J. / Yu, W. / Jin, Y. et al. | British Library Conference Proceedings | 1995


    Multiscale edge detection for infrared ship target recognition

    Lu Jun / Yu Wenxian / Jin Yue et al. | IEEE | 1995


    Research on Ship Target Recognition Based on Maritime Radar Images

    Li, Yi / Xie, Lei / Guo, Tao et al. | IEEE | 2023


    Deep Learning Methods for Ship Classification: From Visible to Infrared Images

    Liu, Tianci / Qin, Hengjia / Zhan, Zhuo et al. | IEEE | 2023