Scholarlink

<p>Deep learning-based steganography methods often achieve state-of-the-art performance on various datasets. More and more methods explore how to enhance steganography capacity without compromising security. In this paper, we propose a novel end-to-end deep learning-based HDR image steganography method. To the best of our knowledge, this is the first steganography method to hide one HDR image within another HDR image. The proposed network hides the HDR secret image of the same size in the HDR cover image by the <i>hiding network</i> and recovers the HDR secret image by the <i>extract network</i>. To overcome the numerical overflow problem of HDR images during training, we designed a reversible normalization method for HDR images. The experimental results show that the proposed method has significant advantages in steganography capacity and can generate images with a satisfactory visual effect.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Large-capacity HDR image steganography based on deep learning

  • 期刊论文

  • 发表日期：

    2025年12月2日

查看全文
  • Longzhi Wang,
  • Yongqing Huo,
  • Wenke Jiang

摘要

Deep learning-based steganography methods often achieve state-of-the-art performance on various datasets. More and more methods explore how to enhance steganography capacity without compromising security. In this paper, we propose a novel end-to-end deep learning-based HDR image steganography method. To the best of our knowledge, this is the first steganography method to hide one HDR image within another HDR image. The proposed network hides the HDR secret image of the same size in the HDR cover image by the hiding network and recovers the HDR secret image by the extract network. To overcome the numerical overflow problem of HDR images during training, we designed a reversible normalization method for HDR images. The experimental results show that the proposed method has significant advantages in steganography capacity and can generate images with a satisfactory visual effect.