• Acta Photonica Sinica
  • Vol. 50, Issue 12, 1228001 (2021)
Shuo WEI1、2, Nanxiang ZHAO1、2、*, Yihua HU1、2, Wanshun SUN1、2, and Biao LIU1、2
Author Affiliations
  • 1State Key Laboratory of Pulsed Power Laser Technology, Electronic Warfare Academy State Key Laboratory of Pulsed Power Laser Technology , National University of Defense Technology, Hefei 230037,China
  • 2Advanced Laser Technology Anhui Laboratory, Hefei 230037,China
  • show less
    DOI: 10.3788/gzxb20215012.1228001 Cite this Article
    Shuo WEI, Nanxiang ZHAO, Yihua HU, Wanshun SUN, Biao LIU. Long-distance Ship Type Recognition Based on Airborne Photon Radar[J]. Acta Photonica Sinica, 2021, 50(12): 1228001 Copy Citation Text show less
    Sea surface point cloud in MABEL data
    Fig. 1. Sea surface point cloud in MABEL data
    Sea scene detected from the right
    Fig. 2. Sea scene detected from the right
    The scene after removing the sea surface
    Fig. 3. The scene after removing the sea surface
    Scenes containing different types of ships
    Fig. 4. Scenes containing different types of ships
    Clustering scene
    Fig. 5. Clustering scene
    Concentric circular division method
    Fig. 6. Concentric circular division method
    The angle between the two ends of the ship
    Fig. 7. The angle between the two ends of the ship
    Models of different types of ships
    Fig. 8. Models of different types of ships

    Rings

    Number

    12345
    383%91%91%75%83%
    4100%91%91%100%91%
    591%91%91%91%91%
    683%83%91%83%83%
    791%83%91%83%91%
    891%91%91%91%100%
    991%91%91%91%91%
    10100%91%100%100%100%
    Table 1. Circular space division method experiment

    Interval

    Number

    12345
    2025%17%33%25%17%
    3017%17%17%17%17%
    4517%17%17%17%17%
    6017%17%17%33%17%
    Table 2. Classification results of routine method vector histogram

    Interval

    Number

    12345
    2067%58%67%75%75%
    3050%58%58%50%83%
    4567%75%75%67%100%
    6067%50%67%58%58%
    Table 3. Classification results of optimized post-method vector histogram
    Ship typeMLong/mWidth/mHigh/m
    Cruisers2.531832056(25)
    Aircraft carriers2.773157560
    Frigates2.861481632(24)
    Destroyers2.611602030
    Fishing ships2.841123082.5(30)
    Medical ships2.852703448
    Landing ships2.631191626(18)
    Crane ships2.681703873(28)
    Cargo ships2.821202440
    Scientific research ships2.731672632
    Container ships2.971202440
    Cruise ships2.64258100172
    Medium-sized aircraft carriers2.551983849
    Table 4. Partial data of different types of ship models
    Training numberNumber of decision treesFeature numberNumber of misclassificationsNumber of correct classificationsCorrect rate
    14012026100%
    24012026100%

    3

    4

    40

    40

    12

    12

    1

    0

    25

    26

    96.15%

    100%

    54012026100%
    6401212596.15%
    74012026100%
    Table 5. Random forest algorithm 7 classification results
    Training numberNumber of decision treesFeature numberNumber of misclassificationsNumber of correct classificationsCorrect rate
    140912596.15%
    240912596.15%
    340922492.31%
    440912596.15%
    540912596.15%
    6409026100%
    740922492.31%
    Table 6. Second classification result
    Shuo WEI, Nanxiang ZHAO, Yihua HU, Wanshun SUN, Biao LIU. Long-distance Ship Type Recognition Based on Airborne Photon Radar[J]. Acta Photonica Sinica, 2021, 50(12): 1228001
    Download Citation