The aim of the work is to increase the efficiency of selection of objects of different nature in digital monochrome images formed in remote sensing systems. For this purpose, algorithms for the formation of features of objects with respect to which boundary values are specified are introduced into the structure of multi-threshold processing. New schemes of multi-threshold processing and selection of objects of interest with threshold setting based on selection results are proposed. Algorithms of multi-threshold selection of objects by area and other scale-invariant geometric features, such as the elongation coefficient of the perimeter of the object and the elongation coefficient of the main axis of the describing ellipse, are obtained and tested. The binarization threshold is set for each of the selected objects based on the extremum of the applied geometric criterion. The new invariant geometric features used are different for round and elongated objects and provide independence of characteristics with changes in the image scale. Results of processing of typical models of images, and also results of selection of objects on the real television and infrared images showing efficiency of the proposed selection method are presented.


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Multi-threshold object selection in images of remote sensing systems


    Beteiligte:
    V. Yu. Volkov (Autor:in) / M. I. Bogachev (Autor:in) / O. A. Markelov (Autor:in)


    Erscheinungsdatum :

    2019




    Medientyp :

    Aufsatz (Zeitschrift)


    Format :

    Elektronische Ressource


    Sprache :

    Unbekannt





    Object and Topology Extraction from Remote Sensing Images

    Maire, C. / Datcu, M. | British Library Conference Proceedings | 2005


    Multi-Branch Parallel Networks for Object Detection in High-Resolution UAV Remote Sensing Images

    Qihong Wu / Bin Zhang / Chang Guo et al. | DOAJ | 2023

    Freier Zugriff

    ICA-based multi-temporal multi-spectral remote sensing images change detection

    Gu, Juan / Li, Xin / Huang, Chunlin et al. | SPIE | 2008