Maritime surveillance systems have been commonly exploited in vessel traffic services. The maritime visual information can be obtained through shore‐borne, ship‐borne or air‐borne cameras. However, the obtained visual data often suffers from blur effects due to the shaky imaging devices in wild conditions (e.g. wind, waves and currents). It is necessary to develop image stabilisation methods to promote visual quality. In this work, we propose a two‐step image stabilisation method by introducing the blind deblurring technology. In particular, we first propose a hybrid regularised blur kernel estimation method to robustly and accurately capture the motion trajectory of imaging devices during shaking. Once the estimation of motion trajectory is obtained, we then present a robust non‐blind deblurring method in the second step to implement restoration of latent sharp image (i.e. image stabilisation). All non‐convex and non‐smooth minimisation problems related to blur kernel estimation and non‐blind deblurring are effectively solved using the alternating direction method of multipliers‐based numerical methods. Extensive experiments on both synthetic and realistic blurred images have been performed. All experimental results have demonstrated the satisfactory performance of our image stabilisation method under different imaging conditions. Moreover, the comparisons on other state‐of‐the‐art methods have also indicated that our method could provide a good representation of promoting object detection in vision‐enabled maritime surveillance systems.


    Access

    Download


    Export, share and cite



    Title :

    A two‐step image stabilisation method for promoting visual quality in vision‐enabled maritime surveillance systems


    Contributors:

    Published in:

    Publication date :

    2023-02-01


    Size :

    15 pages




    Type of media :

    Article (Journal)


    Type of material :

    Electronic Resource


    Language :

    English






    Enhancing Automatic Maritime Surveillance Systems With Visual Information

    Bloisi, Domenico D. / Previtali, Fabio / Pennisi, Andrea et al. | IEEE | 2017