In steganography, the cover medium is widely treated as a mere container for the embedded information, even though it affects the stego-image quality, security, and robustness. In addition, there is no consensus on the characteristics of a suitable cover image. In this work, we introduce and practically prove the most suitable cover image (MSCI) framework to automatically select a cover image for a given secret image. This paper proposes choosing the most suitable cover from a set of images based on two steps. First, a set of cover images is filtered based on relative entropy and a histogram in order to identify the most suitable candidates. Second, the local block pixel intensity features of the candidates are analyzed to select the most suitable cover image. Furthermore, cover image local blocks were optimized, using rotation and flipping, during the embedding process to further improve stego-image representation. The proposed framework demonstrated high visual image quality when compared with existing solutions. Steganalysis tests indicated that the proposed solution for cover selection provided an increased resistance to modern steganalyzers with up to 30% lowered detection rate, which improved security.

中文翻译：

---

基于空间块分析和DCT嵌入的有效封面图像选择

在隐写术中，尽管覆盖介质会影响隐写图像的质量，安全性和鲁棒性，但它被广泛视为嵌入信息的简单容器。另外，对于合适的封面图像的特征还没有达成共识。在这项工作中，我们介绍并实践证明了最合适的封面图像（MSCI）框架，可以自动为给定的秘密图像选择封面图像。本文建议基于两个步骤，从一组图像中选择最合适的封面。首先，根据相对熵和直方图过滤一组封面图像，以识别最合适的候选图像。其次，分析候选者的局部块像素强度特征以选择最合适的覆盖图像。此外，使用旋转和翻转功能对封面图像局部图块进行了优化，在嵌入过程中，以进一步改善隐秘图像的表示。与现有解决方案相比，所提出的框架展示了高视觉图像质量。隐写分析测试表明，建议的封面选择解决方案提高了对现代隐写分析仪的抵抗力，降低了30％的检测率，从而提高了安全性。