Multitask Identity-Aware Image Steganography via Minimax Optimization

Cui, Jiabao; Zhang, Pengyi; Li, Songyuan; Zheng, Liangli; Bao, Cuizhu; Xia, Jupeng; Li, Xi

doi:10.1109/tip.2021.3107999

Cited by 13 publications

(13 citation statements)

References 53 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…ArcFace is a loss function that trains a feature to have a higher cosine similarity with features of the same class and a lower cosine similarity with features of different classes by adding a margin only to the positive class. Multitask Identity-Aware Image Steganography (MIAIS), which is a combination of steganography and face recognition, has been proposed [11]. MIAIS consists of a steg-generator to hide the secret images into the cover images, two steg-classifiers to recognize the container images, i.e., the stego images in this paper, and a steg-restorer to restore the secret images from the container images.…”

Section: Face Recognitionmentioning

confidence: 99%

“…To address this problem, Deep Steganography (DS) using Convolutional Neural Network (CNN) has been proposed, which can embed the information of a single image into another image of the same size using CNN [10]. Using this characteristic, Multitask Identity-Aware Image Steganography (MIAIS) has been proposed to embed a face image into an arbitrary image and extract features corresponding to the face image [11]. MIAIS is not secure as a face recognition method since the quality of the embedded image is low and the face image is visible in the embedded image.…”

Section: Introductionmentioning

confidence: 99%

“…The proposed method can generate higher-quality stego images than MIAIS without sacrificing the face recognition accuracy with the introduction of the three loss functions. We demonstrate the effectiveness of the proposed method compared to DS [10] and MIAIS [11] through performance and security evaluation experiments using the Labeled Faces in the Wild (LFW) dataset 1 , the CASIA-WebFace (CASIA) dataset [13], and the Asian Face Dataset 2 (AFD) [14], which are large-scale public datasets of face images. In addition, one of the potential applications of the proposed method is presented.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Cancelable Face Recognition Using Deep Steganography

Ito,

Kozu,

Kawai

et al. 2024

IEEE Trans. Biom. Behav. Identity Sci.

View full text Add to dashboard Cite

In biometrics, the secure transfer and storage of biometric samples are important for protecting the privacy and security of the data subject. One of the methods for authentication while protecting biometric samples is cancelable biometrics, which performs transformation of features and uses the transformed features for authentication. Among the methods of cancelable biometrics, steganography-based approaches have been proposed, in which secret information is embedded in another to hide its existence. In this paper, we propose cancelable biometrics based on deep steganography for face recognition. We embed a face image or its face features into a cover image to generate a stego image with the same appearance as the cover image. By using a dedicated face feature extractor, we can perform face recognition without restoring the embedded face image or face features from the stego image. We demonstrate the effectiveness of the proposed method compared to conventional steganography-based methods through performance and security evaluation experiments using public face image datasets. In addition, we present one of the potential applications of the proposed method to improve the security of face recognition by using a QR code with a one-time password for the cover image.

show abstract

Section: Face Recognitionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Cancelable Face Recognition Using Deep Steganography

Ito,

Kozu,

Kawai

et al. 2024

IEEE Trans. Biom. Behav. Identity Sci.

View full text Add to dashboard Cite

show abstract

“…Multitask Identity-Aware Image Steganography (MIAIS) for restoring secret images initialized by [25]. The MIAIS system is to improving system robustness and increasing the security in steganography.…”

Section: Literature Review and Problem Statementmentioning

confidence: 99%

Identifying optimal message embedding location in audio steganography using generative adversarial networks

Hajer¹,

Anbar

2022

EEJET

View full text Add to dashboard Cite

Audio steganography (AS) uses the auditory redundancy of the human ear to conceal the hidden message inside the audio track. In recent studies, deep learning-based steganalysis has swiftly revealed AS by extracting high-dimensional stego acoustic features for categorization. There is still an opportunity for improvement in the current audio steganography required for managing communication confidentiality, access control and data protection. The main objective of this research is to improving the data protection by identifying the data embedding location in the audio. Generative Adversarial Network-based Audio Steganography Framework (GAN-ASF) is presented in this study, and it can automatically learn to provide better cover audio for message embedding. The suggested framework's training architecture comprises a generator, a discriminator, and a steganalyzer learned using deep learning. The Least Significant Bit Matching (LSBM) message embedding technique encrypts the secret message into the steganographic cover audio, which is then forwarded to a trained steganalyzer for misinterpretation as cover audio. After performing the training, stenographic cover audio has been generated for encoding the secret message. Here, Markov model of co-frequency sub images to generate the best cover frequency sub-image to locate an image's hidden payload. Steganographic cover audio created by GAN-ASF has been tested and found to be of excellent quality for embedding messages. The suggested method's detection accuracy is lower than that of the most current state-of-the-art deep learning-based steganalysis. This payload placement approach has considerably increased stego locations' accuracy in low frequencies. The test results GAN-ASF achieves a performance ratio of 94.5 %, accuracy ratio of 96.2 %, an error rate of 15.7 %, SNR 24.3 %, and an efficiency ratio of 94.8 % compared to other methods.

show abstract

“…[7][8][9] The key technology of image steganography has been well-developed in the past decade, and its fundamental purpose is to securely transfer encrypted images on a public channel without raising suspicions. [10][11][12] However, existing methods of image steganography are generally in a digital route, and can only achieve a single channel, which means one sender transfers encrypted information to one receiver. [13] Therefore, it will be urgent to develop strategies for multiplexing imaging steganography, which is essential for multifunctional integration and miniaturization.…”

Section: Introductionmentioning

confidence: 99%

Multiplexing Optical Images for Steganography by Single Metasurfaces

Cao

Tang

et al. 2023

Small

View full text Add to dashboard Cite

Image steganography based on intelligent devices is one of the effective routes for safely and quickly transferring secret information. However, optical image steganography has attracted far less attention than digital one due to the state‐of‐the‐art technology limitations of high‐resolution optical imaging in integrated devices. Optical metasurfaces, composed of ultrathin subwavelength meta‐atoms, are extensively considered for flat optical‐imaging nano‐components with high‐resolutions as competitive candidates for next‐generation miniaturized devices. Here, multiplex imaging metasurfaces composed of single nanorods are proposed under a detailed strategy to realize optical image steganography. The simulation and experimental results demonstrate that an optical steganographic metasurface can simultaneously transfer independent secret image information to two receivers with special keys, without raising suspicions for the general public under the cloak of a cover image. The proposed optical steganographic strategy by metasurfaces can arbitrarily distribute a continuous grayscale image together with a black‐and‐white image in separate channels, implying the distinguishing feature of high‐density information capacity for integration and miniaturization in optical meta‐devices.

show abstract

Multitask Identity-Aware Image Steganography via Minimax Optimization

Cited by 13 publications

References 53 publications

Cancelable Face Recognition Using Deep Steganography

Cancelable Face Recognition Using Deep Steganography

Identifying optimal message embedding location in audio steganography using generative adversarial networks

Multiplexing Optical Images for Steganography by Single Metasurfaces

Contact Info

Product

Resources

About