High-Dynamic-Range Tone Mapping in Intelligent Automotive Systems

被引:6
|
作者
Shopovska, Ivana [1 ]
Stojkovic, Ana [1 ]
Aelterman, Jan [1 ]
Van Hamme, David [1 ]
Philips, Wilfried [1 ]
机构
[1] Univ Ghent, IMEC, IPI, B-9000 Ghent, Belgium
关键词
high dynamic range; tone mapping; deep learning; object detection; autonomous driving; OBJECT DETECTION; OPERATOR; NETWORK;
D O I
10.3390/s23125767
中图分类号
O65 [分析化学];
学科分类号
070302 ; 081704 ;
摘要
Intelligent driver assistance systems are becoming increasingly popular in modern passenger vehicles. A crucial component of intelligent vehicles is the ability to detect vulnerable road users (VRUs) for an early and safe response. However, standard imaging sensors perform poorly in conditions of strong illumination contrast, such as approaching a tunnel or at night, due to their dynamic range limitations. In this paper, we focus on the use of high-dynamic-range (HDR) imaging sensors in vehicle perception systems and the subsequent need for tone mapping of the acquired data into a standard 8-bit representation. To our knowledge, no previous studies have evaluated the impact of tone mapping on object detection performance. We investigate the potential for optimizing HDR tone mapping to achieve a natural image appearance while facilitating object detection of state-of-the-art detectors designed for standard dynamic range (SDR) images. Our proposed approach relies on a lightweight convolutional neural network (CNN) that tone maps HDR video frames into a standard 8-bit representation. We introduce a novel training approach called detection-informed tone mapping (DI-TM) and evaluate its performance with respect to its effectiveness and robustness in various scene conditions, as well as its performance relative to an existing state-of-the-art tone mapping method. The results show that the proposed DI-TM method achieves the best results in terms of detection performance metrics in challenging dynamic range conditions, while both methods perform well in typical, non-challenging conditions. In challenging conditions, our method improves the detection F2 score by 13%. Compared to SDR images, the increase in F2 score is 49%.
引用
收藏
页数:15
相关论文
共 50 条
  • [1] Tone mapping for high-dynamic-range images using localized gamma correction
    Qiao, Motong
    Ng, Michael K.
    JOURNAL OF ELECTRONIC IMAGING, 2015, 24 (01)
  • [2] Deep Learning Tone-Mapping and Demosaicing for Automotive Vision Systems
    Stojkovic, Ana
    Aelterman, Jan
    Van Hamme, David
    Shopovska, Ivana
    Philips, Wilfried
    SENSORS, 2023, 23 (20)
  • [3] Gamut Mapping in a High-Dynamic-Range Color Space
    Preiss, Jens
    Fairchild, Mark D.
    Ferwerda, James A.
    Urban, Philipp
    COLOR IMAGING XIX: DISPLAYING, PROCESSING, HARDCOPY, AND APPLICATIONS, 2014, 9015
  • [4] Perceptual Tone Mapping Model for High Dynamic Range Imaging
    Mehmood, Imran
    Shi, Xinye
    Khan, Muhammad Usman
    Luo, Ming Ronnier
    IEEE ACCESS, 2023, 11 : 110272 - 110288
  • [5] Inverse Tone Mapping High Dynamic Range Images
    Touil, Med Amine
    Ellouze, Noureddine
    2017 INTERNATIONAL CONFERENCE ON CONTROL, AUTOMATION AND DIAGNOSIS (ICCAD), 2017, : 338 - 342
  • [6] Tone mapping for high dynamic range displays
    Meylan, Laurence
    Daly, Scott
    Suesstrunk, Sabine
    HUMAN VISION AND ELECTRONIC IMAGING XII, 2007, 6492
  • [7] Tone mapping-based high-dynamic-range image compression: study of optimization criterion and perceptual quality
    Narwaria, Manish
    Da Silva, Matthieu Perreira
    Le Callet, Patrick
    Pepion, Romuald
    OPTICAL ENGINEERING, 2013, 52 (10)
  • [8] EMBEDDING HIGH DYNAMIC RANGE TONE MAPPING IN JPEG COMPRESSION
    Liu, Jian
    Hassan, Firas
    Carletta, Joan
    IMAGE PROCESSING: ALGORITHMS AND SYSTEMS XI, 2013, 8655
  • [9] An Adaptive Tone Mapping Algorithm for High Dynamic Range Images
    Zhang, Jian
    Kamata, Sei-ichro
    COMPUTATIONAL COLOR IMAGING, 2009, 5646 : 207 - 215
  • [10] A New Tone Mapping Workflow for High Dynamic Range Content
    Zhai, Jiefu
    Llach, Joan
    Wang, Zhe
    2009 CONFERENCE FOR VISUAL MEDIA PRODUCTION: CVMP 2009, 2009, : 91 - 99