Raining and snowing have negative effects on the visual quality and degrade the performance of various vision algorithms in outdoor environment. In this chapter, we propose three rain and snow removal algorithms including two for videos and one for single images. We first introduce an algorithm to remove snowflakes in videos by using the global and local low-rank property; i.e., assume the snowflake is sparse in snowflake-removed scenes. The proposed framework contains two steps of low-rank decomposition. The first step is to extract stationary background with the whole frames, and the second step focuses on local blocks obtained by patch-matching for removing snowflakes in front of moving foreground. Then, we introduce a rain and snow removal algorithm based on matrix decomposition no longer with the assumption that all the snowflakes/rain streaks are sparse in snow/rain scenes. We divide snowflakes and rain streaks into two categories: sparse ones and dense ones. With background fluctuations and optical flow information, the detection of moving objects and sparse snowflakes/rain streaks is formulated as a multi-label Markov Random Fields (MRFs), making the rain and snow removal model adaptable to different rain and snow scenes. Finally, we propose an encoder–decoder deraining net for single-image rain removal in heavy scenes, which is only supervised by MSE loss and adopts dually skip connections for rain removal in one stage. The dual connections, skip sum connection and skip concatenation connection, enable the deraining net to promote information flow between layers, and thus can help to localize rain streaks. In addition, the learnable pixel-wise attention can adaptively recalibrate the decoded rain features to preserve more image details.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Rain and Snow Removal


    Weitere Titelangaben:

    Research on Intelligent Manufacturing


    Beteiligte:
    Tian, Jiandong (Autor:in)


    Erscheinungsdatum :

    2021-12-02


    Format / Umfang :

    39 pages





    Medientyp :

    Aufsatz/Kapitel (Buch)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch




    Vehicle rain and snow removing method, vehicle rain and snow removing system and vehicle with vehicle rain and snow removing system

    HUANG HAITAO / HUANG ZHONGSHUI / WANG YAN et al. | Europäisches Patentamt | 2023

    Freier Zugriff

    RAIN AND SNOW INTRUSION ARRESTER

    KITAMURA TOSHIRO | Europäisches Patentamt | 2022

    Freier Zugriff

    RAIN-SNOW DISCRIMINATION WITH POLARIMETRIC RADAR

    Brandes, E. A. / Vivekanandan, J. / Rasmussen, R. M. et al. | British Library Conference Proceedings | 2002


    Snow and rain deflector assembly for vehicles

    BLOMBERG THOMAS ALBERT | Europäisches Patentamt | 2019

    Freier Zugriff

    SNOW AND RAIN DEFLECTOR ASSEMBLY FOR VEHICLES

    BLOMBERG THOMAS ALBERT | Europäisches Patentamt | 2019

    Freier Zugriff