a survey on near-reversible data hiding techniques · 2018-09-29 · a survey on near-reversible...

16
A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES VELPURU MUNI SEKHAR 1 , CH SRAVAN 2 , P BUDDHA REDDY 3 , K ANVESH 4 1,2,3,4 Department of Information Technology, Vardhaman College of Engineering and India [email protected] June 12, 2018 Abstract Data hiding is process of hiding secret information into digital content (cover-content) such as image, video and audio. Steganography and watermarking are two data hiding application, i.e., are widely used in multimedia security applications such crime investigation, copy right protection, copy control, medical imaging, remote sensing and etc. Now researchers showing lot of interest towards data hiding research, due to rapid digitization. With rapid digitization, security of digital content became major challenge. This paper presents a survey on Near-reversible Data Hiding (NDH) techniques, its major applications, focus of the work and compares different techniques performance analysis. The near-reversibility gaining popularity over Reversible Data Hiding (RDH) techniques due to robustness of NDH technique and NDH techniques are resistive against signal processing attacks and compression. Moreover, applications like crime investigation, remote sensing, medical imaging and bio-metric support slight modification after extraction of cover-content. Keywords :Near-reversible Data Hiding, Reversible Data Hiding, Steganography, watermarking, cover-content. 1 International Journal of Pure and Applied Mathematics Volume 120 No. 6 2018, 3309-3324 ISSN: 1314-3395 (on-line version) url: http://www.acadpubl.eu/hub/ Special Issue http://www.acadpubl.eu/hub/ 3309

Upload: others

Post on 23-Jun-2020

4 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES · 2018-09-29 · A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES VELPURU MUNI SEKHAR1, CH SRAVAN2, P BUDDHA REDDY3, K ANVESH4

A SURVEY ON NEAR-REVERSIBLEDATA HIDING TECHNIQUES

VELPURU MUNI SEKHAR1, CH SRAVAN2,P BUDDHA REDDY3, K ANVESH4

1,2,3,4 Department of Information Technology,Vardhaman College of Engineering and India

[email protected]

June 12, 2018

Abstract

Data hiding is process of hiding secret information intodigital content (cover-content) such as image, video andaudio. Steganography and watermarking are two datahiding application, i.e., are widely used in multimediasecurity applications such crime investigation, copy rightprotection, copy control, medical imaging, remote sensingand etc. Now researchers showing lot of interest towardsdata hiding research, due to rapid digitization. With rapiddigitization, security of digital content became majorchallenge. This paper presents a survey on Near-reversibleData Hiding (NDH) techniques, its major applications,focus of the work and compares different techniquesperformance analysis. The near-reversibility gainingpopularity over Reversible Data Hiding (RDH) techniquesdue to robustness of NDH technique and NDH techniquesare resistive against signal processing attacks andcompression. Moreover, applications like crimeinvestigation, remote sensing, medical imaging andbio-metric support slight modification after extraction ofcover-content.

Keywords:Near-reversible Data Hiding, ReversibleData Hiding, Steganography, watermarking, cover-content.

1

International Journal of Pure and Applied MathematicsVolume 120 No. 6 2018, 3309-3324ISSN: 1314-3395 (on-line version)url: http://www.acadpubl.eu/hub/Special Issue http://www.acadpubl.eu/hub/

3309

Page 2: A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES · 2018-09-29 · A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES VELPURU MUNI SEKHAR1, CH SRAVAN2, P BUDDHA REDDY3, K ANVESH4

1 INTRODUCTION

Digital revolution make human to slaves of the computer. Todaymost of the things are getting done by computers, because of thatpresent era called as digital era [4]. Due to the wide digitizationnot only giving convenience and also trust of the digital contentbecome a major challenge [4, 5]. To generate trust on netizensvarious security models are in practice among that watermarkingand steganography are the two [6], both are instance of datahiding techniques. The data hiding is a process of hiding secretinformation into cover-content. The location of secret informationin cover-content is depend upon application [4,5,6,9,17]. Datahiding have many applications such as covert communication,copy control, broad caste monitoring, digital contentauthentication and etc. In data hiding existence of secretinformation is not visible to Human Visual System (HVS)[4,5,6,8,9,17]. The quality of data hiding is measures withfollowing parameters such as capacity (maximum size of secretinformation), security (strength of embedding and extractionalgorithm), imperceptibility (invisible to HVS and robustness(resistive against signal processing attacks and compression). Indata hiding technique exist two procedures called embedding andextraction.

Embedding ProcessEmbedding is a process of concealing secret information S intocover-content I in imperceptibly, to generate stego-object Stego byusing embedding algorithm embed as shown equation (1).

stegoembed(I, S) (1)

Figure 1: Components of embedding process

2

International Journal of Pure and Applied Mathematics Special Issue

3310

Page 3: A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES · 2018-09-29 · A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES VELPURU MUNI SEKHAR1, CH SRAVAN2, P BUDDHA REDDY3, K ANVESH4

Extraction ProcessIt is an art of separating secret information ’S|’ and cover content ’I|’from a stego- object ’Stego’ by using extraction algorithm extract.

[I|, S|]extract(Stego) (2)

Figure 2: extraction process

Based on embedding algorithm input and extraction algorithmoutput data hiding techniques are classified into three categoriesi.e..,

• Near-reversible data hiding (NDH) [3]

• Reversible data hiding (RDH) [1, 2,17]

• Irreversible data hiding [8]

If embedded secret message ’S’ and extracted secret message’S|’ are equal and cover content of embedding process ’I’ andextracting process ’I|’ are equal then that type of data hidingtechniques called as reversible data hiding [1, 2]; otherwiseirreversible data hiding [8]. The near-reversible paradigm is firstintroduced by barni et al.., [3]. The near-reversible techniquesproved that some applications like remote sensing, crimeinvestigation and bio-metric accept approximate recovery ofcover-content in extraction process. It means that absolutedifference of cover-content ’I’ in embedding and cover-content’I|’ after extraction are I I |.

Ahmed et al., [10] proposed a digital watermarking techniquebased on semi-reversible frequency domain data hiding. Thesemi-reversible [10] is used in medical image authentication fordistributed diagnosi and home health care systems. Tang et al.,

3

International Journal of Pure and Applied Mathematics Special Issue

3311

Page 4: A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES · 2018-09-29 · A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES VELPURU MUNI SEKHAR1, CH SRAVAN2, P BUDDHA REDDY3, K ANVESH4

[11] has proposed a new NDH technique with projectionhistogram modification. This NDH [11] technique focuses oncapacity, visual quality and robustness of NDH and RDHtechniques. Zhang et al., [12] also proposed a new NDH techniquewith Least Significant Bit (LSB) substitution. This NDH [12]technique provide strong robustness with low computationalcomplexity. Po Y chen et al., [13] gave new direction to currentNDH technique with frequency domain DWT and quantizationcoefficients. This work [13] proved that DWT plus quantizationhas more embedding capacity with good visual quality thentraditional data hiding techniques. Sagar et al., [8] also proposeda new NDH technique with frequency domain non-zero DCTcoefficients. This NDH [8] technique focuses on embeddingcapacity with good visual perception while enable compression.Moreover, Muni Sekhar V[4,5,6] has combined sagar [8] and Po Ychen [13] with adaptive quantization and frequency domainnon-zero coefficients data hiding. This works [4,5,6] focuses onlarge embedding capacity with good visual quality, comparisonanalysis of DCT non-zero coefficient embedding technique [8] andDWT plus adaptive quantized non-zero embedding coefficients,comparison analysis of DWT plus quantization embedding [13]and DWT plus adaptive quantized non-zero embedding. Thiswork NDH [5], analysed the embedding capacity and visualquality in different adaptive quantized intervals. This work [6]applied DWT plus adaptive quantized data hiding strategy tosteganography application for covert communication and analysedthe performance with RDH techniques.

This paper presents survey of NDH techniques to enableresearchers to given sight into near-reversibility. Moreover, thispaper also categorized NDH techniques and gave each basicstructure, block diagram and performance analysis. The NDHtechniques are classified into three categories such as

1. Histogram Modification (HM) based NDH techniques [11]

2. Quantization based NDH techniques [4,5, 6,8, 13]

3. LSB substitution based NDH techniques [12]

All three categories are different with way of embedding andsame with input and output.

4

International Journal of Pure and Applied Mathematics Special Issue

3312

Page 5: A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES · 2018-09-29 · A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES VELPURU MUNI SEKHAR1, CH SRAVAN2, P BUDDHA REDDY3, K ANVESH4

This survey report organized as shown in figure 3, where eachrectangle in the NDH consider a subsection in NDH techniques.The rest-of-the-paper is organized as follows, section 2, discussesdifferent near reversible data hiding and their methodology withrequired block diagram and tables. Then section 3, compare theresults of different data hiding scheme that are discussed inprevious section. Finally, section 4 concludes with summary of alltechniques.

Figure 3: organization of paper

2 NEAR-REVERSIBLE DATA

HIDING TECHNIQUES

Near Most of the Data hiding applications allows a reasonablelevel of modification after extraction process this reasoning leadsto the new category of research called NDH. In NDH, original

5

International Journal of Pure and Applied Mathematics Special Issue

3313

Page 6: A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES · 2018-09-29 · A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES VELPURU MUNI SEKHAR1, CH SRAVAN2, P BUDDHA REDDY3, K ANVESH4

cover-content is extracted after extraction process withmoderately acceptable level.

Histogram Modification based NDHNi et al., [14] developed a novel RDH technique based on HM.HM embedding technique identify pair of peak and zero points isselected from the histogram of the cover-content as shown figure4. Only pixels with values between peak and zero points undergomodification during embedding process. However, the data hidingcapacity is restricted to the number of pixels there in the peakpoint in a histogram of the cover-content. However, Tang andHuang [11] have proposed an enhanced a near-reversible dataembedding scheme based on projection histogram manipulations.The problems addressed by NDH [11] are robustness. Most of theapplication allows a reasonable level of modification afterextraction process this reasoning leads to the new category ofresearch called NDH. Over tradition data hiding techniques NDHare performing better in terms of embedding capacity, imagefidelity, and robustness. Image Lena original, histogram modifiedand stego-images shown below figure 5. Embedding process isillustrated in figure 5 and histograms of example image (Lena)figure 6 (a) and its histogram modified are shown in figure 6 (b).

Embedding of secret data into cover-content in HM basedNDH as shown in figure 4. It is generally consists of three steps 1)Generating Histogram from Image, 2) Identifying location toembed secret data, and 3) finally, embed secret data intocover-content.

Histogram can be as following equation

h(x)#{1imn;xx}x[0, 255] (3)

All HM techniques data hiding will be performed on Histogramonly as shown in figure 5(b). Embedding on histogram is shown inequation (4) [11, 14].

Aij(x, y) =

{A(x, y), ifS(k) = 0&Aij(x, y) = p.

1, ifS(k) = 1&Aij(x, y) = p.(4)

Quantization based Data Hiding SchemesThe tern Near-reversibility first introduced by Barni et al.., [3] byusing quantization approach. And, Sagar G et al.., [8] was

6

International Journal of Pure and Applied Mathematics Special Issue

3314

Page 7: A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES · 2018-09-29 · A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES VELPURU MUNI SEKHAR1, CH SRAVAN2, P BUDDHA REDDY3, K ANVESH4

Figure 4: Abstract Embedding process of Histogram Modificationmethod

Figure 5: (a) Lena Image (b) Lena Histogram

7

International Journal of Pure and Applied Mathematics Special Issue

3315

Page 8: A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES · 2018-09-29 · A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES VELPURU MUNI SEKHAR1, CH SRAVAN2, P BUDDHA REDDY3, K ANVESH4

extended the work of Barni et al.., [3], Tang Y-L [11] and Zhang B[12] with quantization and DCT compression based data hidingfor MPEG4 videos. Moreover, Po-Yueh Chen [16] was introducednew paradigm called adaptive quantization for better quality withbetter image compression compared to traditional uniformquantization. Then, Muni Sekhar V el al.., [4,5,6] combined thenone-zero embedding and adaptive quantization based on DWTto produce better stego-image quality and compression ratio.Moreover, experiment carried out in different quantizationintervals with respective PSNR and embedding capacity. Theseexperiments proved that by increasing number of quantizationinterval embedding capacity and image quality both together areincreasing [4]. The process of embedding in quantization basedNDH as shown in figure 6:

Figure 6: quantization based data embedding process

LSB Substitution based Data Hiding schemeThe LSB substitution is oldest data hiding scheme among the

all. The LSB substitution techniques are widely using techniqueirrespective of data hiding schemes [16]. Po-Yueh Chen [16]extended the work of Van Schyndel [15] from LSB substitution toMLSB substitution. LSB substitution pixel representation figure 8play major role.

The basic concept of LSB substitution is to embed theconfidential data at the rightmost bits (bits with the smallestweighting). Hence, the embedding procedure does not affect much

8

International Journal of Pure and Applied Mathematics Special Issue

3316

Page 9: A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES · 2018-09-29 · A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES VELPURU MUNI SEKHAR1, CH SRAVAN2, P BUDDHA REDDY3, K ANVESH4

Figure 7: Gray scale pixel bit positions

of the original pixel value. The mathematical representation forLSB method is:

I ′(x, y)I(x, y)I(x, y)mod2km (5)

In above equation I(x,y) and I’(x,y) represents the pixel valuesof original and stego-images at (x,y) location. Where m representsthe secret data bit embedding at (x,y) location. And, k representsthe number of LSB bits to substitute. Mathematically the extractedmessage is represented as:

mI ′(x, y)mod2k (6)

3 RESULTS ANALYSIS OF

NEAR-REVERSIBLE DATA

HIDING SCHEMES

Results of various techniques are analyzed based on followingmeasures such as Peak Signal Noise Ratio (PSNR), Mean SquareError (MSE), embedding capacity and Normalization CorrelationCoefficient (NCC). PSNR = 10log10

2552

MSE

MSE = 1M∗NM∗N (aij, bij

Comparison of DCT plus Quantization [8], DWT plusAdaptive quantization [4, 5, 6], Multiple LSB [16] and HistogramModification [14] data hiding schemes:

9

International Journal of Pure and Applied Mathematics Special Issue

3317

Page 10: A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES · 2018-09-29 · A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES VELPURU MUNI SEKHAR1, CH SRAVAN2, P BUDDHA REDDY3, K ANVESH4

Figure 8: Embedding capacity analysis for various images

Embedding Capacity vs. Computational Complexity:All three categories of approaches have their own advantages anddisadvantages. If data hiding application required compressionwhile data hiding, then Quantization mechanisms are bestsuitable. If data hiding application required less computationalcomplexity with less embedding capacity, then histogrammodification is better suitable approach. If data hidingapplication required less computational capacity with moreembedding capacity, then LSB substitution is better suitableapproach as shown in figure 9.

Visual quality vs. Computational Complexity: All threecategories of approaches have their own advantages anddisadvantages. If data hiding application required compressionwhile data hiding, then Quantization mechanisms are bestsuitable. Among all quantization based technique adaptivequantization based techniques is best suitable for goodcompression ratio and visual quality from table 2. If data hidingapplication required less computational complexity with goodvisual quality, then histogram modification is better suitableapproach. If data hiding application required less computationalcapacity with good visual quality, then LSB substitution is bettersuitable approach as shown in figure 11.

Compression ratio of various quantization methods are shownin table below.

10

International Journal of Pure and Applied Mathematics Special Issue

3318

Page 11: A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES · 2018-09-29 · A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES VELPURU MUNI SEKHAR1, CH SRAVAN2, P BUDDHA REDDY3, K ANVESH4

Figure 9: embedding capacity analysis for various images

Figure 10: PSNR analysis for various images

11

International Journal of Pure and Applied Mathematics Special Issue

3319

Page 12: A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES · 2018-09-29 · A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES VELPURU MUNI SEKHAR1, CH SRAVAN2, P BUDDHA REDDY3, K ANVESH4

Figure 11: PSNR analysis for various images

12

International Journal of Pure and Applied Mathematics Special Issue

3320

Page 13: A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES · 2018-09-29 · A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES VELPURU MUNI SEKHAR1, CH SRAVAN2, P BUDDHA REDDY3, K ANVESH4

DWT plus adaptive quantization method is more dominatecompared to DCT plus quantization scheme with respectivecompression ratio, embedding capacity and visual quality.

4 CONCLUSIONS

This report classified near-reversible techniques into three groupssuch (1) quantization (Compression), (2) Histogram modificationand (3) LSB substitution primarily based information hidingschemes. From literature, it is clear the content-aware data hidinghas good performance over fixed location based data hiding andadaptive quantization has good compression ratio as well as goodembedding capacity with good visual quality. a) Near-reversibletechniques enable good compression ratio with good visual andembedding capacity over reversible data hiding. b) NDHperforming better in the sense of RDH techniques with Histogrambased techniques and LSB substituting techniques. c) NDHtechniques are in semi- fragile in nature.

References

[1] C.W. Honsinger, P.W. Jones, M. Rabbani, J.C. Stoffel,Lossless recovery of an original image containing embeddeddata, U.S. Patent No. 6,278,791 (2001).

[2] Jiri Fridrich, Miroslav Goijan, Rui Dub (2001) InvertibleAuthentication Proceedings of SPIE Vol. 4314, pp-197-208.

[3] Barni M, Bartolini F, Cappellini V, Magli E, Olmo G. (2002)Near-lossless digital watermarking for copyright protection ofremote sensing images. In: IEEE international geoscience andremote sensing symposium, 2002. IGARSS 02., vol 3, pp14471449.

[4] V.Muni Sekhar, K.V.G.Rao, K.SambasivaRao and M.Gopichand “Comparing the Capacity, NCC and Fidelityof Various quantization Intervals on DWT”,InternationalConference on Innovations in Computer science &Engineering(ICICSE-2015), published in Journal of Advances

13

International Journal of Pure and Applied Mathematics Special Issue

3321

Page 14: A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES · 2018-09-29 · A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES VELPURU MUNI SEKHAR1, CH SRAVAN2, P BUDDHA REDDY3, K ANVESH4

in Intelligence Systems and Computing, Springer ISBN978-981-10-0417-9, August 2015.

[5] Muni Sekhar V, K. Venu gopal Rao, N. Sambasiva Rao (2015)Enhanced Adaptive Data hiding in DWT. In IOSR Journal ofComputer Engineering Vol 17(2) pp 30-40 DOI: 10.9790/0661-17263040.

[6] Muni Sekhar V, K. Venu gopal Rao, N. Sambasiva Rao (2015)Reference Cover Image Steganography. In ACST Vol. 8(1) pp-17-28.

[7] J. Feng, I. Lin, C. Tsai, Y. Chu, Reversible Watermarking :Current Status and Key Issues, International Journal 2 (3)(2006) 161170

[8] Sagar Gujjunoori and B. B. Amberker (2012) A DCT BasedNear Reversible Data Embedding Scheme for MPEG-4 VideoProceedings of the Fourth International Conference on Signaland Image Processing 2012, pp-69-80.

[9] Asifullah Khan, Ayesha Siddiqa, Summuyya Munib, and SanaAmbreen Malik (2014) A Recent Survey of ReversibleWatermarking Techniques Information sciences 2014,DOI:10.1016/j.ins.2014.03.118, 1-37.

[10] Ahmed F, Moskowitz IS (2006) A semi-reversible watermarkfor medical image authentication. In: 1st Transdisciplinaryconference on distributed diagnosis and home healthcare.D2H2, pp 5962.

[11] Tang Y-L, Huang H-T (2007) Robust near-reversible dataembedding using histogram projection. In: In IIH-MSP 2007,vol 02, pp 453456.

[12] Zhang B, Xin Y, Niu X-X, Yuan K-G, Jiang H-B (2010) A nearreversible image watermarking algorithm. In: Internationalconference on machine learning and cybernetics (ICMLC), vol6, pp 28242828, july 2010.

[13] Po-Yueh Chen and Hung-Ju Lin (2006) “A DWT BasedApproach for Image Steganography” International Journal ofApplied Science and Engineering, Vol. 4(3), pp-275-290.

14

International Journal of Pure and Applied Mathematics Special Issue

3322

Page 15: A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES · 2018-09-29 · A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES VELPURU MUNI SEKHAR1, CH SRAVAN2, P BUDDHA REDDY3, K ANVESH4

[14] Z. Ni, Y. Shi, N. Ansari, W. Su, Reversible data hiding, IEEETransactions on Circuits and Systems for Video Technology,16 (3) (2006) 354362.

[15] Van Schyndel, R. G., Tirkel, A. Z., and Osborne, C. F.(1994) “A digital watermark”, IEEE International. Conf.Image Processing, 2: 86-90.

[16] Po-Yueh Chen and Jia-Yu Chang (2013) “An AdaptiveQuantization Scheme for 2-D DWT Coefficients” InternationalJournal of Applied Science and Engineering, Vol. 11(1), pp-85-100.

[17] V Muni Sekhar, K.V.G.Rao, N. Sambasiva Rao, Ch SravanKumar (2017) “A Reversible RIE based Watermarkingscheme” IEEE 7th International conference on IACC, 5-7January, 2017, Hyderabad, India.

15

International Journal of Pure and Applied Mathematics Special Issue

3323

Page 16: A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES · 2018-09-29 · A SURVEY ON NEAR-REVERSIBLE DATA HIDING TECHNIQUES VELPURU MUNI SEKHAR1, CH SRAVAN2, P BUDDHA REDDY3, K ANVESH4

3324