With the promotion of economic globalization, the demand for air travel and maritime travel in all countries has increased significantly. However, the accompanying rate of accidents at sea is increasing rapidly year by year. After the accident, the survival rate of people is closely related to the speed of rescue. Artificial eyesight and radar electronic search and rescue (SAR) are limited due to inaccuracy and limited resources. The rapid development of unmanned aerial vehicle (UAV) technology in recent years has made it possible to obtain stable, high-resolution images from airborne cameras. In this paper, a method of SAR using images captured by UAVs is proposed. The detection of victim is divided into two steps. First, in order to improve the system’s real-time, the possible position of a victim may be located from the image through a simple feature to avoid exhaustive search. This step is called hypothesis generation (HG). Then, the pre-trained convolutional neural network (CNN) is used to test the HG region to ensure the accuracy of the system. This step is called hypothesis verification (HV). The result shows that taking advantage of pre-processing + CNN has better comprehensive performance in SAR at sea.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    A Deep-Learning-Based Sea Search and Rescue Algorithm by UAV Remote Sensing


    Beteiligte:
    Wang, Shubo (Autor:in) / Han, Yu (Autor:in) / Chen, Jian (Autor:in) / Zhang, Zichao (Autor:in) / Wang, Guangqi (Autor:in) / Du, Nannan (Autor:in)


    Erscheinungsdatum :

    01.08.2018


    Format / Umfang :

    631747 byte




    Medientyp :

    Aufsatz (Konferenz)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch