A secure visual secret sharing (VSS) scheme with CNN-based image enhancement for underwater images
- PDF / 4,876,063 Bytes
- 15 Pages / 595.276 x 790.866 pts Page_size
- 23 Downloads / 207 Views
ORIGINAL ARTICLE
A secure visual secret sharing (VSS) scheme with CNN-based image enhancement for underwater images Nikhil C. Mhala1 · Alwyn R. Pais1 Accepted: 31 August 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract Nowadays, underwater images are being used to identify various important resources like objects, minerals, and valuable metals. Due to the wide availability of the Internet, we can transmit underwater images over a network. As underwater images contain important information, there is a need to transmit them securely over a network. Visual secret sharing (VSS) scheme is a cryptographic technique, which is used to transmit visual information over insecure networks. Recently proposed randomized VSS (RVSS) scheme recovers secret image (SI) with a self-similarity index (SSIM) of 60–80%. But, RVSS is suitable for general images, whereas underwater images are more complex than general images. In this paper, we propose a VSS scheme using super-resolution for sharing underwater images. Additionally, we have removed blocking artifacts from the reconstructed SI using convolution neural network (CNN)-based architecture. The proposed CNN-based architecture uses a residue image as a cue to improve the visual quality of the SI. The experimental results show that the proposed VSS scheme can reconstruct SI with almost 86–99% SSIM. Keywords Visual secret sharing (VSS) · Visual cryptography · Super-resolution · Underwater images · Discrete cosine transform (DCT)
1 Introduction An underwater image contains important information about the different types of objects like minerals, metals, etc. [15,18]. As these images contain valuable information, it needs to be transmitted securely over a network. Nowadays, researchers proposed various schemes to detect salient objects present in an image. Recently, Fu. Keren et al. [14] proposed a generalized convolution neural network (CNN) architecture namely deepside to detect salient objects present in an image. They have fused the hierarchical CNN features using the segmentation-based pooling. Most of the neural network-based technique suffers from the problem of coarse object boundaries. To solve this problem, Zhao et al. [37] proposed an edge guidance network (EGNet) technique. They have three main stages in EGNet as: (1) The first stage extracts the salient object features. (2) The second stage integrates the local and global edge information, and (3)
B 1
Nikhil C. Mhala [email protected] Information Security Research Laboratory, Department of Computer Science and Engineering, National Institute of Technology Karnataka, Surathkal, Mangalore, India
the third stage couples the same salient edge features with salient object features at various resolutions. A novel joint learning and densely-cooperative fusion (JL-DCF) architecture for RGB-D salient object detection are proposed in [13]. They have used RGB and depth channels jointly to learn about the salient objects using a Siamese network. Further, in [11] authors provided a new dataset and ben
Data Loading...
