Abstract: Image steganography algorithms based on deep learning are often trained using either spatial- or frequency-domain features. It is difficult for features from a single domain to comprehensively express the content of an entire image, which usually leads to poor performance because steganography is commonly multi-task. To solve this problem, this paper proposes a robust image steganography algorithm based on feature score maps, called the secure and robust image steganography network (SRIS-Net). First, instead of spatial-domain steganography, our proposed algorithm utilizes a convolutional neural network to obtain shallow spatial-domain features. These features are decomposed by Laplacian pyramid frequency-domain decomposition (LPFDD) to hide secret information in the different frequency sub-bands with a progressive assisted hiding strategy that significantly reduces the influence of the secret information on the cover image, achieving significant invisibility and robust performance. In addition, we propose a global–local embedding module (GLEM) to achieve embedding by considering the overall structure of the image and the local details, and a dual multi-scale aggregation sub-network (DMSubNet) to perform multi-scale reconstruction to improve the quality of the carrier image. For security, we propose a dual-task discriminator structure, while giving a real/fake judgment of the image, which can generate a feature score map of the cover image’s region of interest (ROI) to guide the embedding module to generate a carrier image with higher imperceptibility and undetectability. Experimental results on BOSSBase show that our SRIS-Net outperforms mainstream methods in terms of undetectability and robustness, with more than 9.2 and 3.4 dB improvement in visual quality, respectively, and the capacity can be increased up to approximately 72–96 bits per pixel.

SRIS-Net：基于特征评分图的鲁棒图像隐写算法

[1]Baluja S, 2017. Hiding images in plain sight: deep steganography. Proc 31^st Int Conf on Neural Information Processing Systems, p.2066-2076.

[2]Baluja S, 2020. Hiding images within images. IEEE Trans Patt Anal Mach Intell, 42(7):1685-1697.

[3]Barni M, Bartolini F, Piva A, 2001. Improved wavelet-based watermarking through pixel-wise masking. IEEE Trans Image Process, 10(5):783-791.

[4]Bas P, Filler T, Pevný T, 2011. “Break our steganographic system”: the ins and outs of organizing boss. Proc 13^th Int Conf on Information Hiding, p.59-70.

[5]Cheddad A, Condell J, Curran K, et al., 2010. Digital image steganography: survey and analysis of current methods. Signal Process, 90(3):727-752.

[6]Chen BJ, Wang JX, Chen YY, et al., 2020. High-capacity robust image steganography via adversarial network. KSII Trans Int Inform Syst, 14(1):366-381.

[7]Denemark T, Fridrich J, Holub V, 2014. Further study on the security of S-UNIWARD. Proc IS&T/SPIE Electronic Imaging, Article 902805.

[8]Duan XT, Jia K, Li BX, et al., 2019. Reversible image steganography scheme based on a U-Net structure. IEEE Access, 7:9314-9323.

[9]Duan XT, Gou MX, Liu N, et al., 2020a. High-capacity image steganography based on improved Xception. Sensors, 20(24):7253.

[10]Duan XT, Liu N, Gou MX, et al., 2020b. SteganoCNN: image steganography with generalization ability based on convolutional neural network. Entropy, 22(10):1140.

[11]Fridrich J, Kodovsky J, 2012. Rich models for steganalysis of digital images. IEEE Trans Inform Forens Secur, 7(3):868-882.

[12]Fu ZJ, Wang F, Cheng X, 2020. The secure steganography for hiding images via GAN. EURASIP J Image Video Process, 2020:46.

[13]Holub V, Fridrich J, Denemark T, 2014. Universal distortion function for steganography in an arbitrary domain. EURASIP J Inform Secur, 2014:1.

[14]Hu DH, Wang L, Jiang WJ, et al., 2018. A novel image steganography method via deep convolutional generative adversarial networks. IEEE Access, 6:38303-38314.

[15]Huang JJ, Cheng SY, Lou SH, et al., 2019. Image steganography using texture features and GANs. Proc Int Joint Conf on Neural Networks, p.1-8.

[16]Isola P, Zhu JY, Zhou TH, et al., 2017. Image-to-image translation with conditional adversarial networks. Proc IEEE Conf on Computer Vision and Pattern Recognition, p.1125-1134.

[17]Jing JP, Deng X, Xu M, et al., 2021. HiNet: deep image hiding by invertible network. Proc IEEE/CVF Int Conf on Computer Vision, p.4733-4742.

[18]Johnson J, Alahi A, Li FF, 2016. Perceptual losses for real-time style transfer and super-resolution. Proc 14^th European Conf on Computer Vision, p.694-711.

[19]Kodovsky J, Fridrich J, Holub V, 2012. Ensemble classifiers for steganalysis of digital media. IEEE Trans Inform Forens Secur, 7(2):432-444.

[20]Lai WS, Huang JB, Ahuja N, et al., 2019. Fast and accurate image super-resolution with deep Laplacian pyramid networks. IEEE Trans Patt Anal Mach Intell, 41(11):2599-2613.

[21]Li XL, Yang B, Cheng DF, et al., 2009. A generalization of LSB matching. IEEE Signal Process Lett, 16(2):69-72.

[22]Li ZZ, Yang XY, Shen KQ, et al., 2022. Dual branch parallel steganographic framework based on multi-scale distillation in framelet domain. Neurocomputing, 514:182-194.

[23]Liu LS, Meng LZ, Wang XL, et al., 2022. An image steganography scheme based on ResNet. Multim Tools Appl, 81(27):39803-39820.

[24]Lu SP, Wang R, Zhong T, et al., 2021. Large-capacity image steganography based on invertible neural networks. Proc IEEE/CVF Conf on Computer Vision and Pattern Recognition, p.10816-10825.

[25]Mao XD, Li Q, Xie HR, et al., 2017. Least squares generative adversarial networks. Proc IEEE Int Conf on Computer Vision, p.2794-2802.

[26]Otazu X, Gonzalez-Audicana M, Fors O, et al., 2005. Introduction of sensor spectral response into image fusion methods. IEEE Trans Geosci Remote Sens, 43(10):2376-2385.

[27]Ren S, Gong H, Zheng SY, 2025. Algorithm for 3D point cloud steganalysis based on composite operator feature enhancement. Front Inform Technol Electron Eng, 26(1):62-78.

[28]Ruanaidh JJKO, Dowling WJ, Boland FM, 1996. Phase watermarking of digital images. Proc 3^rd IEEE Int Conf on Image Processing, p.239-242.

[29]Shi HC, Dong J, Wang W, et al., 2018. SSGAN: secure steganography based on generative adversarial networks. Proc 18^th Pacific Rim Conf on Multimedia, p.534-544.

[30]Shi WZ, Caballero J, Huszár F, et al., 2016. Real-time single image and video super-resolution using an efficient sub-pixel convolutional neural network. Proc IEEE Conf on Computer Vision and Pattern Recognition, p.1874-1883.

[31]Singh B, Sharma PK, Huddedar SA, et al., 2022. StegGAN: hiding image within image using conditional generative adversarial networks. Multim Tools Appl, 81(28):40511-40533.

[32]Tamimi AA, Abdalla AM, Al-Allaf O, 2013. Hiding an image inside another image using variable-rate steganography. Int J Adv Comput Sci Appl, 4(10):18-21.

[33]Tancik M, Mildenhall B, Ng R, 2020. StegaStamp: invisible hyperlinks in physical photographs. Proc IEEE/CVF Conf on Computer Vision and Pattern Recognition, p.2117-2126.

[34]ur Rehman A, Rahim R, Nadeem S, et al., 2018. End-to-end trained CNN encoder-decoder networks for image steganography. Proc Computer Vision-ECCV Workshops, p.723-729.

[35]Van TP, Dinh TH, Thanh TM, 2019. Simultaneous convolutional neural network for highly efficient image steganography. Proc 19^th Int Symp on Communications and Information Technologies, p.410-415.

[36]Wengrowski E, Dana K, 2019. Light field messaging with deep photographic steganography. Proc IEEE/CVF Conf on Computer Vision and Pattern Recognition, p.1515-1524.

[37]Wu P, Yang Y, Li XQ, 2018. StegNet: mega image steganography capacity with deep convolutional network. Fut Int, 10(6):54.

[38]Xu GS, Wu HZ, Shi YQ, 2016. Structural design of convolutional neural networks for steganalysis. IEEE Signal Process Lett, 23(5):708-712.

[39]Xu YM, Mou C, Hu YJ, et al., 2022. Robust invertible image steganography. Proc IEEE/CVF Conf on Computer Vision and Pattern Recognition, p.7875-7884.

[40]Yang H, Xu YT, Liu XH, et al., 2024. PRIS: practical robust invertible network for image steganography. Eng Appl Artif Intell, 133: 108419.

[41]Ye J, Ni JQ, Yi Y, 2017. Deep learning hierarchical representations for image steganalysis. IEEE Trans Inform Forens Secur, 12(11):2545-2557.

[42]Ying QC, Zhou H, Zeng XH, et al., 2022. Hiding images into images with real-world robustness. Proc IEEE Int Conf on Image Processing, p.111-115.

[43]Yu C, 2020. Attention based data hiding with generative adversarial networks. Proc AAAI Conf on Artificial Intelligence, p.1120-1128.

[44]Zamir SW, Arora A, Khan S, et al., 2022. Restormer: efficient transformer for high-resolution image restoration. Proc IEEE/CVF Conf on Computer Vision and Pattern Recognition, p.5728-5739.

[45]Zhang L, Lu Y, Li J, et al., 2023. Deep adaptive hiding network for image hiding using attentive frequency extraction and gradual depth extraction. Neur Comput Appl, 35(15):10909-10927.

[46]Zhang R, Dong SQ, Liu JY, 2019. Invisible steganography via generative adversarial networks. Multim Tools Appl, 78(7):8559-8575.

[47]Zhang R, Zhu F, Liu JY, et al., 2020. Depth-wise separable convolutions and multi-level pooling for an efficient spatial CNN-based steganalysis. IEEE Trans Inform Forens Secur, 15:1138-1150.

[48]Zhou DQ, Hou QB, Chen YP, et al., 2020. Rethinking bottleneck structure for efficient mobile network design. Proc 16^th European Conf on Computer Vision, p.680-697.