Image Steganography is the process of hiding information which can be text, image or video inside a cover image. The secret information is hidden in a way that it not visible to the human eyes. Deep learning technology, which has emerged as a powerful tool in various applications including image steganography, has received increased attention recently. The main goal of this paper is to explore and discuss various deep learning methods available in image steganography field. Deep learning techniques used for image steganography can be broadly divided into three categories - traditional methods, Convolutional Neural Network-based and General Adversarial Network-based methods. Along with the methodology, an elaborate summary on the datasets used, experimental set-ups considered and the evaluation metrics commonly used are described in this paper. A table summarizing all the details are also provided for easy reference. This paper aims to help the fellow researchers by compiling the current trends, challenges and some future direction in this field.

https://northumbria-test.eprints-hosting.org/id/document/273545

Image Steganography: A Review of the Recent Advances

Internet of Things (IoT) is a domain where the transfer of big data is taking place every single second. The security of these data is a challenging task; however, security challenges can be mitigated with cryptography and steganography techniques. These techniques are crucial when dealing with user authentication and data privacy. In the proposed work, a highly secured technique is proposed using IoT protocol and steganography. This work proposes an image steganography procedure by utilizing the combination of various algorithms that build the security of the secret data by utilizing Binary bit-plane decomposition (BBPD) based image encryption technique. Thereafter a Salp Swarm Optimization Algorithm (SSOA) based adaptive embedding process is proposed to increase the payload capacity by setting different parameters in the steganographic embedding function for edge and smooth blocks. Here the SSOA algorithm is used to localize the edge and smooth blocks efficiently. Then, the hybrid Fuzzy Neural Network with a backpropagation learning algorithm is used to enhance the quality of the stego images. Then these stego images are transferred to the destination in the highly secured protocol of IoT. The proposed steganography technique shows better results in terms of security, image quality, and payload capacity in comparison with the existing state of art methods.

/pdf/ssii-secured-and-high-quality-steganography-using-32r07i5bjk.pdf

SSII: Secured and High-Quality Steganography Using Intelligent Hybrid Optimization Algorithms for IoT

Image steganography is the operation of hiding a message into a cover image. the message can be text, codes, or image. Hiding an image into another is the proposed approach in this paper. Based on LSB coding, a k-LSB-based method is proposed using k least bits to hide the image. For decoding the hidden image, a region detection operation is used to know the blocks contains the hidden image. The resolution of stego image can be affected, for that, an image quality enhancement method is used to enhance the image resolution. To demonstrate the effectiveness of the proposed approach, we compare it with some of the state-of-the-art methods.

An image steganography approach based on k-least significant bits (k-LSB)

Steganography is a technique that transmits secret data or message in an appropriate multimedia carrier, e.g., image, audio, and video files. It comes under the assumption that if the feature is visible, the point of attack is evident. However, such technology is always used by criminals who do not want to be easily discovered to hide harmful information in various media, especially in images. Massive spreading
of those harmful information will increase the difficulty of social security management. In this case, excellent image steganalysis should be developed and applied. Specially, real-time image steganalysis is necessary when information timelines need to be protected. If detection scene has large amounts of users, deep learning can be applied to improve performance of image steganalysis benefiting from its powerful processing capability. Using deep learning, real-time image steganalysis system gets higher accuracy and efficiency. In this paper, we give an account of preliminary knowledge first. A brief overview of the deep neural networks (DNN) is also presented. The combination of DNN and real-time image steganalysis is introduced. Then, we import the concept of CNN in DNN, and expound theory as well as advantages of combining CNN and image steganalysis. For multi-user scenarios, we analyze a practical real-time image steganalysis application based on outlier detection methods. At last, we prospect the future issues of real-time image steganalysis.

Deep learning for real-time image steganalysis: a survey

Transform domain steganography methods are always preferable than spatial domain methods to the research community due to following advantages: better feature identification, improved security and higher robustness. In this paper, an integer sequence named Catalan Transform (CT) has been exploited in the image steganography domain. At the outset, the cover image is decomposed into 2 × 2 non-overlapping blocks in row major order. Then, each such block i.e., 4-pixel group is converted into transform domain using CT. Secret bits are embedded into the transformed components in varying proportions which facilitates us to achieve a payload in the range of 1 to 4 bpp. Inverse Catalan Transform (ICT) is applied over transformed cum embedded quadruples to generate the stego-pixels in spatial domain. Successive embedding operation over an entire image ensures the formation of stego-image. Experimental results confirm that the Peak Signal to Noise Ratio (PSNR) values obtained by the proposed method are always above the acceptable level (i.e., 30 dB) and at the same time, it outperforms many state-of-the-art methods. Other metrics such as Mean Squared Error (MSE), Structural Similarity Index Measure (SSIM), Normalized Cross Correlation (NCC) and Global histograms with similarity measure have also been computed which prove the effectiveness of the proposed scheme. The robustness of the stego-images has further been tested using the StegExpose tool and it is found that the tool is unable to detect the presence of secret data. The code of this work is available at: https://github.com/Souradeep150/Catalan-based-Steganography
 .

Secured image steganography based on Catalan transform

A high security data hiding approach using image cropping and Least Significant Bit (LSB) steganography is proposed. The predefined certain secret coordinate crops are extracted from the cover image. The secret text message is divided into parts with the same image crops. Each part of the secret text message is embedded into an image crop with secret sequence using LSB approach. The embedding is done using the cover image of three color channels. The stego image is given by reassembling the image and the stego crops. A detailed comparative study is performed between the proposed approach and the other state-of-the-art approaches. This comparison is based on visualization to detect any degradation in stego image, difficulty of extracting the embedded data by any unauthorized viewer, Peak Signal-to-Noise Ratio (PSNR) of stego image, and the embedding algorithm CPU time. Experimental results shows that the proposed approach is more secure compared with the other traditional approaches.

High security data hiding using image cropping and LSB least significant bit steganography

This paper proposes a hybrid robust image watermarking scheme based on three levels of Discrete Stationary Wavelet Transform (DSWT) in the Discrete Cosine Transform (DCT) domain. The host image colors (red, green and blue) are separated, and then the DCT is applied on each color after separation. The DSWT is utilized to divide the DCT output into four sub-bands (3 levels). These sub-bands are (A, H, V, D) matrices with the same image size. The watermark is embedded on matrix A. The results of the proposed watermarking scheme are compared with other state-of-the-art schemes. The comparison is based on visualization to detect any degradation of the watermarked image, Peak Signal-to-Noise Ratio (PSNR) of the watermarked image, Normal Correlation (NC) of the extracted watermark after detection, applying attacks, and then calculating the PSNR and NC.

A hybrid scheme for robust color image watermarking using DSWT in DCT domain

This paper proposes a robust image watermarking approach using homomorphic based SVD in the DWT domain. The DWT is utilized to divide the host image into non-overlapping bands. The reflectance component of sub-band LL is extracted using homomorphic transform for each color (red, green and blue). The embedding watermark is done by applying SVD on the reflectance component of the sub-band LL. The results of the proposed watermarking approach are compared with those of the other traditional approaches. Evaluation is based on visualization, Peak Signal-to-Noise Ratio of watermarked image (PSNR), Normalized Correlation of watermark after detection (NC). Experimental results ensure that the proposed watermarking approach is both robust and immune to attacks studied in this paper compared with the other traditional approaches.

Ahmed Elmhalawy

Papers

High security data hiding using image cropping and LSB least significant bit steganography

A hybrid scheme for robust color image watermarking using DSWT in DCT domain

Efficient color image watermarking using homomorphic based SVD in DWT domain