In order to better balance the relationship between steganographic capacity and imperceptibility, this paper proposed an optimization scheme based on SteganoGAN. Firstly, the steganography network and extraction network of SteganoGAN were deepen to enhance the complexity and learning capability of the model; Secondly, in order to achieve more covert information embedding, discrete wavelet transform (DWT) and inverse discrete wavelet transform (IDWT) modules were introduced in the steganography network, which enabled secret information to be effectively embedded in the wavelet domain of the image; Finally, an improved channel and spatial attention module (ICAM-SAM) was incorporated into the steganography and extraction networks, enabling the model to focus on highly covert areas in the image and achieve more accurate information hiding. Experimental results show that the accuracy of the improved model is increased by 0.84 percentage point, indicating that its steganography and extraction processes are more accurate. In addition, the reed-solomon bits per pixel (RS-BPP) is increased by 1.71%, indicating that the improved model can hide more information in images of the same size. Meanwhile, the peak signal to noise ratio (PSNR) is increased by 12.53% and the structural similarity index (SSIM) is increased by 5.14%, indicating that the embedded information has a smaller impact on the original image and the improved model has higher image quality. In summary, these results indicate that the improved model has better imperceptibility and larger steganography capacity.
CHENJunfu, FUZhangjie, ZHANGWeiming, et al. Review of image steganalysis based on deep learning [J]. Journal of Software, 2021, 32(2): 551-578. (in Chinese)
[4]
GOODFELLOWI J, POUGET-ABADIEJ, MIRZAM, et al. Generative adversarial networks[J]. Advances in Neural Information Processing Systems, 2014, 27: 2672-2680.
BIXinliang, YANGXiaoyuan, LIUWenchao, et al. Research on cover selection image steganography algorithm[J]. Application Research of Computers, 2021, 38(8): 2465-2468. (in Chinese)
[7]
MARTÍNA, HERNÁNDEZA, ALAZABM, et al. Evolving generative adversarial networks to improve image steganography[J]. Expert Systems with Applications, 2023, 222: 119841-119855.
[8]
ZHANGR, DONGS, LIUJ. Invisible steganography via generative adversarial networks[J]. Multimedia Tools and Applications, 2019, 78(7): 8559-8575.
[9]
ZHUJ, KAPLANR, JOHNSONJ, et al. Hidden: hiding data with deep networks [C]//Proceedings of the European Conference on Computer Vision (ECCV), 2018: 657-672.
[10]
JINGJ, DENGX, XUM, et al. HiNet: deep image hiding by invertible network [C]//Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021: 4733-4742.
[11]
WUP C, CHENL G. An efficient architecture for two-dimensional discrete wavelet transform[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2001, 11(4): 536-545.
[12]
GUANZ, JINGJ, DENGX, et al. DeepMIH: Deep invertible network for multiple image hiding[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 45(1): 372-390.
[13]
HUJ, SHENL, SUNG. Squeeze-and-excitation networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 7132-7141.
[14]
WANGQ, WUB, ZHUP, et al. ECA-Net: Efficient channel attention for deep convolutional neural networks[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020: 11534-11542.
[15]
WOO S, PARKJ, LEEJ Y, et al. Cbam: Convolutional block attention module[C]//Proceedings of the European Conference on Computer Vision (ECCV), 2018: 3-19.
[16]
TANJ, LIAOX, LIUJ, et al. Channel attention image steganography with generative adversarial networks[J]. IEEE Transactions on Network Science and Engineering, 2021, 9(2): 888-903.
[17]
PENGW, LIS, QIANZ, et al. Text steganalysis based on hierarchical supervised learning and dual attention mechanism[J]. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2023, 31: 3513-3526.
[18]
ZHANGK A, CUESTA-INFANTEA, XUL, et al. SteganoGAN: high capacity image steganography with GANs[DB/OL].(2019-01-30)[2024-03-27].
[19]
LIS, WANGJ, LIUP, et al. Detection of multiple steganography methods in compressed speech based on code element embedding, Bi-LSTM and CNN with attention mechanisms[J]. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2021, 29: 1556-1569.
[20]
AGUSTSSONE, TIMOFTER. Ntire 2017 challenge on single image super-resolution: dataset and study [C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, 2017: 126-135.
[21]
LINT Y, MAIREM, BELONGIES, et al. Microsoft COCO: common objects in context[C]//Computer Vision-ECCV 2014. Cham: Springer International Publishing, 2014: 740-755.