A High-Capacity Steganography Algorithm Based on Adaptive Frequency Channel Attention Networks
Deep learning has become an essential technique in image steganography. Most of the current deep-learning-based steganographic methods process digital images in the spatial domain. There are problems such as limited embedding capacity and unsatisfactory visual quality. To improve capacity-distortion...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-10-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/22/20/7844 |
| _version_ | 1797470040689213440 |
|---|---|
| author | Shanqing Zhang Hui Li Li Li Jianfeng Lu Ziqian Zuo |
| author_facet | Shanqing Zhang Hui Li Li Li Jianfeng Lu Ziqian Zuo |
| author_sort | Shanqing Zhang |
| collection | DOAJ |
| description | Deep learning has become an essential technique in image steganography. Most of the current deep-learning-based steganographic methods process digital images in the spatial domain. There are problems such as limited embedding capacity and unsatisfactory visual quality. To improve capacity-distortion performance, we develop a steganographic method from the frequency-domain perspective. We propose a module called the adaptive frequency-domain channel attention network (AFcaNet), which makes full use of the frequency features in each channel by a fine-grained manner of assigning weights. We apply this module to the state-of-the-art SteganoGAN, forming an Adaptive Frequency High-capacity Steganography Generative Adversarial Network (AFHS-GAN). The proposed neural network enhances the ability of high-dimensional feature extraction through overlaying densely connected convolutional blocks. In addition to this, a low-frequency loss function is introduced as an evaluation metric to guide the training of the network and thus reduces the modification of low-frequency regions of the image. Experimental results on the Div2K dataset show that our method has a better generalization capability compared to the SteganoGAN, with substantial improvement in both embedding capacity and stego-image quality. Furthermore, the embedding distribution of our method in the DCT domain is more similar to that of the traditional method, which is consistent with the prior knowledge of image steganography. |
| first_indexed | 2024-03-09T19:31:09Z |
| format | Article |
| id | doaj.art-dc51d5eef2ba4f87954d8f8fbd4cab5d |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-09T19:31:09Z |
| publishDate | 2022-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-dc51d5eef2ba4f87954d8f8fbd4cab5d2023-11-24T02:27:11ZengMDPI AGSensors1424-82202022-10-012220784410.3390/s22207844A High-Capacity Steganography Algorithm Based on Adaptive Frequency Channel Attention NetworksShanqing Zhang0Hui Li1Li Li2Jianfeng Lu3Ziqian Zuo4School of Computer Science and Technology, Hangzhou Dianzi University, Hangzhou 310018, ChinaSchool of Computer Science and Technology, Hangzhou Dianzi University, Hangzhou 310018, ChinaSchool of Computer Science and Technology, Hangzhou Dianzi University, Hangzhou 310018, ChinaSchool of Computer Science and Technology, Hangzhou Dianzi University, Hangzhou 310018, ChinaSchool of Computer Science and Technology, Hangzhou Dianzi University, Hangzhou 310018, ChinaDeep learning has become an essential technique in image steganography. Most of the current deep-learning-based steganographic methods process digital images in the spatial domain. There are problems such as limited embedding capacity and unsatisfactory visual quality. To improve capacity-distortion performance, we develop a steganographic method from the frequency-domain perspective. We propose a module called the adaptive frequency-domain channel attention network (AFcaNet), which makes full use of the frequency features in each channel by a fine-grained manner of assigning weights. We apply this module to the state-of-the-art SteganoGAN, forming an Adaptive Frequency High-capacity Steganography Generative Adversarial Network (AFHS-GAN). The proposed neural network enhances the ability of high-dimensional feature extraction through overlaying densely connected convolutional blocks. In addition to this, a low-frequency loss function is introduced as an evaluation metric to guide the training of the network and thus reduces the modification of low-frequency regions of the image. Experimental results on the Div2K dataset show that our method has a better generalization capability compared to the SteganoGAN, with substantial improvement in both embedding capacity and stego-image quality. Furthermore, the embedding distribution of our method in the DCT domain is more similar to that of the traditional method, which is consistent with the prior knowledge of image steganography.https://www.mdpi.com/1424-8220/22/20/7844image steganographygenerative adversarial networksdiscrete cosine transformattention mechanism |
| spellingShingle | Shanqing Zhang Hui Li Li Li Jianfeng Lu Ziqian Zuo A High-Capacity Steganography Algorithm Based on Adaptive Frequency Channel Attention Networks Sensors image steganography generative adversarial networks discrete cosine transform attention mechanism |
| title | A High-Capacity Steganography Algorithm Based on Adaptive Frequency Channel Attention Networks |
| title_full | A High-Capacity Steganography Algorithm Based on Adaptive Frequency Channel Attention Networks |
| title_fullStr | A High-Capacity Steganography Algorithm Based on Adaptive Frequency Channel Attention Networks |
| title_full_unstemmed | A High-Capacity Steganography Algorithm Based on Adaptive Frequency Channel Attention Networks |
| title_short | A High-Capacity Steganography Algorithm Based on Adaptive Frequency Channel Attention Networks |
| title_sort | high capacity steganography algorithm based on adaptive frequency channel attention networks |
| topic | image steganography generative adversarial networks discrete cosine transform attention mechanism |
| url | https://www.mdpi.com/1424-8220/22/20/7844 |
| work_keys_str_mv | AT shanqingzhang ahighcapacitysteganographyalgorithmbasedonadaptivefrequencychannelattentionnetworks AT huili ahighcapacitysteganographyalgorithmbasedonadaptivefrequencychannelattentionnetworks AT lili ahighcapacitysteganographyalgorithmbasedonadaptivefrequencychannelattentionnetworks AT jianfenglu ahighcapacitysteganographyalgorithmbasedonadaptivefrequencychannelattentionnetworks AT ziqianzuo ahighcapacitysteganographyalgorithmbasedonadaptivefrequencychannelattentionnetworks AT shanqingzhang highcapacitysteganographyalgorithmbasedonadaptivefrequencychannelattentionnetworks AT huili highcapacitysteganographyalgorithmbasedonadaptivefrequencychannelattentionnetworks AT lili highcapacitysteganographyalgorithmbasedonadaptivefrequencychannelattentionnetworks AT jianfenglu highcapacitysteganographyalgorithmbasedonadaptivefrequencychannelattentionnetworks AT ziqianzuo highcapacitysteganographyalgorithmbasedonadaptivefrequencychannelattentionnetworks |