Medical JPEG image steganography based on preserving inter-block dependencies

https://doi.org/10.1016/j.compeleceng.2017.08.020Get rights and content

Highlights

  • We investigate an adaptive data hiding strategy that preserves the differences among DCT coefficients at the same position in adjacent DCT blocks.

  • A novel medical JPEG image steganographic scheme based on the inter-block dependencies is proposed, so as to improve the security of the patients personal information.

  • Experimental results show that the proposed scheme can cluster the inter-block embedding changes and perform better than the state-of-the-art works.

Abstract

With the development of computer and biomedical technologies, medical JPEG images contain the patients’ personal information and the security of the private information attracts great attention. Steganography is utilized to conceal the private information, so as to provide privacy protection of medical images. Most of existing JPEG steganographic schemes embed messages by modifying discrete cosine transform (DCT) coefficients, but the dependencies among DCT coefficients would be disrupted. In this paper, we propose a new medical JPEG image steganographic scheme based on the dependencies of inter-block coefficients. The basic strategy is to preserve the differences among DCT coefficients at the same position in adjacent DCT blocks as much as possible. The cost values are allocated dynamically according to the modifications of inter-block neighbors in the embedding process. Experimental results show that the proposed scheme can cluster the inter-block embedding changes and perform better than the state-of-the-art steganographic method.

Introduction

With the rapid reform and development of the biomedical system, digital medical images have become increasingly important in recent years [1]. Medical images can be transmitted conveniently through the networks for the purposes of research, education, and consultations. Since medical images contain the patients’ personal information, information security and privacy protection have become greatly significant during transmitting medical images over the Internet [2], [3], [4]. Therefore, steganography is introduced to provide protection and confidentiality for medical images, and it could make the patients’ information undetectable [5], [6].

Bremnavas et al. [7] presented a new steganographic method to hide the patient’s information into medical images. The information is embedded by the least significant bit (LSB) method, and then the medical image is encrypted using chaos algorithms. Pandey et al. [8] combined image cryptography and steganography techniques for the secure transmission of medical images. The medical image is first encrypted and then embedded with the patients’ information. There are some steganographic schemes for the encrypted medical images. Qin et al. [9] proposed an inpainting-assisted reversible steganographic scheme using histogram shifting mechanism. They designed an effective reversible steganographic scheme for the privacy protection of medical image content [10]. Liao et al. investigated reversible data hiding in encrypted medical images based on the absolute mean difference of multiple neighboring pixels [11].

Recently, the JPEG format has been increasingly adopted for medical image storage and transmission, since it can achieve not only higher compression rate but also good visual quality. Hence, JPEG image steganography can be utilized to embed the patients’ personal information into medical JEPG images. Researchers have made much helpful progress on JPEG image steganography. A classic method called F5 was proposed by A. Westfeld [12]. It only embeds messages into the non-zero alternating current (AC) DCT coefficients, but introduced the shrinkage effect if a coefficient becomes zero after embedding. Non-shrinkage F5 [13], an improved version of F5, assigned infinite costs to some DCT coefficients, and thus alleviated the negative effect. Guo et al. [14] spread the embedding modification to each DCT coefficient evenly and designed a cost function for homogeneous embedding according to the principles of the spread spectrum communication. Huang et al. [15] presented a new channel selection rule for JPEG image steganography, aiming to find the DCT coefficients that may introduce minimal detectable distortion. Wang et al. [16] proposed an efficient JPEG steganography scheme based on the block entropy of DCT coefficients and syndrome trellis coding (STC) [17]. In 2013, Huang et al. [18] divided DCT coefficients into two portions and assigned different weights for them, and designed the cost function based on the quantization step, quantified coefficients and quantitative disturbance error. Filler et al. [19] constructed the cost function and JPEG image steganographic scheme by designing and optimizing a multi-parameter model with specific statistical features. Lately, they proposed JPEG universal wavelet relative distortion (J-UNIWARD) [20] which evaluates the embedding costs of DCT coefficients in the spatial domain by using inverse DCT, and implements the embedding operations in JPEG domain. Wang et. al [21] exploited block fluctuation and quantization steps to design a hybrid distortion function for JPEG image steganography.

In a JPEG image, DCT coefficients exhibit two kinds of complex dependencies, intra-block dependencies, and inter-block dependencies. Intra-block dependencies refer to the relationship among coefficients with similar frequency in the same block, while inter-block dependencies describe the relationship among coefficients at the corresponding positions in different DCT blocks. However, the existing JPEG image steganographic schemes might destroy the inter-block dependencies. As a modern JEPG image steganalysis approach, the union of JPEG and spatial rich model (JSRM) [22] could detect the data hiding traces according to the dependencies of DCT coefficients. Thus, the security performance of JPEG image steganographic schemes could be improved by preserving the inter-block dependencies.

In 2015, clustering modification directions (CMD) strategy [23] was presented, which mainly focused on preserving the correlation between neighboring pixels in the spatial domain. Consequently, it can synchronize the modification directions, and enhance the performance evaluated by the powerful spatial image steganalysis. In this paper, inspired by CMD, we propose an adaptive JPEG image steganographic scheme, and it preserves the correlation among inter-block adjacent coefficients by adjusting cost values in the embedding process. The initial cost values of all coefficients are firstly computed by one of the existing distortion functions. The original JPEG image is divided into several non-overlapping sub-images, ensuring that the neighboring DCT coefficient blocks belong to different sub-images. For a given DCT coefficient, it has four corresponding points at the same locations in the four adjacent DCT blocks (we name them inter-block neighbors). The cost value of each coefficient would be dynamically adjusted in accordance with the modifications of its neighbors. Experimental results show that the proposed scheme performs better than J-UNIWARD in resisting the modern JPEG image steganalysis.

The rest of the paper is organized as follows. A strategy for preserving inter-block dependencies is introduced in Section 2. In Section 3, we describe the proposed JPEG image steganographic scheme in details. The comparative experiments are presented in Section 4. Finally, the conclusion is given in Section 5.

Section snippets

The strategy for preserving inter-block dependencies

JPEG image steganographic schemes usually hide information into an image by adding or subtracting the values of DCT coefficients. In the ternary embedding framework, the coefficients might be modified by plus one or minus one. From the perspective of steganalysis, modern steganalytic methods always detect the data hiding traces by capturing the fluctuations. When inter-block dependencies remain unchanged, the fluctuations might be reduced.

In order to maintain the inter-block dependencies, the

The proposed image JPEG steganographic scheme

In this subsection, a novel adaptive JPEG image steganographic scheme based on preserving inter-block dependencies is proposed. The most important operation is the process of updating the cost values. The cost values assignment fully explores the mutual embedding impacts of inter-block coefficients. The detailed steps of embedding and extracting algorithms are as follows.

Experimental results

In this section, some experimental results and analyses are presented to demonstrate the effectiveness of the proposed scheme. In Section 4.1, some experiments are performed to determine the adjusting parameter α. The complexity analysis is presented in section 4.2. Section 4.3 illustrates the visualizing embedding changes of the proposed scheme, and verifies that the proposed scheme could effectively cluster embedding changes and preserve the inter-block dependencies. The experimental results

Conclusion

Nowadays, the need for sharing medical images is growing rapidly, and advanced medical information system is changing the way that medical images are stored, accessed and distributed. A large amount of patients’ personal information is included in medical JPEG images. Thus, the privacy protection of medical JEPG images has become an important issue. Steganography is a useful tool to conceal patients’ information in the medical images. Most of existing JPEG image steganographic schemes might

Acknowledgments

This work is supported by National Natural Science Foundation of China (Grant Nos. 61402162, 61572182, 61370225, 61472131, 61272546, 61300220), the Scientific Research Fund of Hunan Provincial Education Department (Grant No. 16B089), Hunan Provincial Natural Science Foundation of China (Grant No. 2017JJ3040), Opening Project of Shanghai Key Laboratory of Integrated Administration Technologies for Information Security(Grant No. AGK201605), CCF-Venustech Research Fund, Science and Technology Key

Xin Liao received the B.E. degree and Ph.D. degree in information security from Beijing University of Posts and Telecommunications, Beijing, China, in 2007 and 2012, respectively. He is currently an associate professor with Hunan University, China, where he joined in 2012. His research interests include steganography, watermarking, and multimedia forensic.

References (25)

  • C. Qin et al.

    An inpainting-assisted reversible steganographic scheme using a histogram shifting mechanism

    IEEE Trans Circuits Syst Video Technol

    (2013)
  • A. Westfeld

    F5-A steganographic algorithm: high capacity despite better steganalysis

    Springer international workshop on information hiding

    (2001)
  • Cited by (0)

    Xin Liao received the B.E. degree and Ph.D. degree in information security from Beijing University of Posts and Telecommunications, Beijing, China, in 2007 and 2012, respectively. He is currently an associate professor with Hunan University, China, where he joined in 2012. His research interests include steganography, watermarking, and multimedia forensic.

    Jiaojiao Yin received the Bachelors degree in engineering from Henan Normal University of China, Henan, China. Currently, she is pursuing the Masters degree in College of Computer Science and Electronic Engineering, Hunan University, China. Her research interests include steganography and image processing.

    Sujing Guo received the Bachelors degree in College of Computer Science and Electronic Engineering, Hunan University, China. Her research interests include steganography and image processing.

    Xiong Li received his Ph.D. degree from Beijing University of Posts and Telecommunications, China, in 2012. He is currently an associate professor at Hunan University of Science and Technology, China. He has published more than 60 referred journal papers in the area of cryptography and information security, and won JNCA 2015 Best Paper Award.

    Arun Kumar Sangaiah received his Ph.D. degree in Computer Science and Engineering from the VIT University, India. He is presently working as an associate professor in School of Computer Science and Engineering, VIT University, India. His research interests include software engineering, computational intelligence, wireless networks, bioinformatics, and embedded systems.

    Reviews processed and recommended for publication to the Editor-in-Chief by Guest Editor Dr. R. C. Poonia.

    View full text