Wavelet based Image Watermarking using Huffman Compression

2 downloads 0 Views 377KB Size Report
kinds of image compression techniques: Lossless and Lossy compression. ... Discrete Wavelet. Transform (DWT), Singular Value Decomposition (SVD),.
International Journal of Applied Engineering Research, ISSN 0973-4562 Vol. 10 No.69 (2015) © Research India Publications; httpwww.ripublication.comijaer.ht

Wavelet based Image Watermarking using Huffman Compression Technique Anum Javeed Zargar

Amit Kumar Singh

Department of Computer Science & Engineering, Jaypee University of Information Technology, Waknaghat, Solan Himachal Pradesh India

Department of Computer Science & Engineering, Jaypee University of Information Technology, Waknaghat, Solan Himachal Pradesh India

[email protected]

[email protected]

criteria used for enhancement. LSB substitution, spread spectrum and patchwork are some important spatial domain methods. These methods are less complex but are not robust against signal processing attacks. However, transform domain methods are more robust against the attacks. Discrete Wavelet Transform (DWT), Singular Value Decomposition (SVD), Discrete Cosine Transform(DCT)and Discrete Fourier Transform(DFT) are most important transform domain techniques. Recently, wavelet based watermarking methods have been achieved higher robustness. The main advantages of wavelet transform domain for watermarking applications are: 1) space frequency localization 2) multi-resolution representation 3) adaptability 4) multi-scale analysis and 5) linear complexity. The Wavelet transform provides both spatial and frequency resolutions [6].

Abstract—In this paper, an algorithm for digital image watermarking based on discrete wavelet transforms (DWT) and Huffman compression is proposed. In the embedding process, the host image is decomposed into first level DWT and the watermark image is compressed by Huffman compression before embedding into the selected sub-band of the cover. The compressed watermark is then embedded into the selected subband of the host image. The proposed method has been extensively tested against numerous known signal processing attacks and has been found to be robust and highly imperceptible. Further, the performance of the algorithm has been tested with Block truncation Coding (BTC) and fractal compression based watermarking method. The performance of the Huffman compression technique based technique is better than the BTC, fractal based compression techniques in terms of robustness and imperceptibility. Keywords— DWT; Huffman; BTC, Fractal compression; Medical image

Huffman compression based watermarking method has been proposed in [7-9] improves the quality and security of watermarked image, host image are almost identical to watermarked image and is difficult to differentiate between host and watermarked image. Huffman encoding of the host image keeps the image away from stealing, destroying and altering by any accidental users hence the proposed method is more robust against brute force attack. They are able to achieve 100 % accuracy against attacks and recovery up to 98 %. The main goal of using compression in any application is to reduce the number of bits significantly, while keeping the pixel resolution and the visual quality of the reconstructed image as close to the original image as possible [10].

I. INTRODUCTION A Watermark embeds an imperceptible signal to audio, video and image for variety of purpose including copyright protection and violation. However, the working principles of watermarking and compression seem to be different as perceptual data coding removes inherent redundancy during compression. Compression plays very important role in image processing when applications required sending large data on open channel [1]. Image Compression is a well known topic that codes picture into fewer amounts of data. There are two kinds of image compression techniques: Lossless and Lossy compression. Lossless image compression is error-free coding methods. It can decompress into one which is the original image. Lossless compression keeps all the information in image, the size of the compressed results is not reduced so much. Most common examples of lossless compression are Huffman compression, arithmetic compression, run length coding etc.

The proposed method based on DWT and Huffman compression techniques. In the embedding process, the host image is decomposed into first-level DWT and the compressed watermark image is embedded into the selected sub-band of the host image. The algorithm correctly extracts the embedded watermark. The method is robust against well known signal processing attacks without significant degradation of the quality of the watermarked image.

Lossy image compression techniques produce results with little distortion and image obtained from decompressing is not same as the original one [1]. BTC, Fractal compression, JPEG compression are most common techniques for lossy compression method. Image watermarking techniques can be divided into two domain methods [2]: spatial domain and transform domain watermarking [3-5]. Spatial domain is based on gray-level mapping where type of mapping depends on

II. PERFORMANCE METRICS Compression efficiency and complexity are the important metrics to evaluate the performance of a lossless image compression algorithm. Compression efficiency is evaluated by the compression ratio or by the bit rate [9]. Compression

1

International Journal of Applied Engineering Research, ISSN 0973-4562 Vol. 10 No.69 (2015) © Research India Publications; httpwww.ripublication.comijaer.ht

ratio is the size of the original to the compressed image, and bit-rate is the number of bits that are conveyed or processed per unit time, and is required by the compressed image [1114]. Robustness and imperceptibility are the key parameters to evaluate the performance of the watermarking algorithm. The PSNR and NC evaluate the perceptual quality of the watermarked image and the robustness of the extracted watermark respectively [15-18].

terms of PSNR, NC and compression ratio. For testing, gray scale images of size 512 512 as cover image [19] and logo image of size 256 256 as watermark image. The image watermark embedding method is based on DWT and Huffman. Figure 2 shows (a) brain image is taken as host image, (b) JUIT logo as watermark, (c) watermarked image and (d) extracted watermark is obtained by using extraction algorithm. For the testing, gain factor varied from 0.1 to 0.9.

III. PROPOSED ALGORITHM The watermark embedding and extraction method can be shown in Figure 1(a) and Figure 1(b) respectively. The proposed algorithm has two parts: (1) embedding and (2) extraction for image watermark as given below: A. Embedding Process In the embedding process, the host image is decomposed into first level DWT. Then any appropriate selected sub-band is taken and watermark image is compressed with the help of Huffman compression. The embedding between compressed watermark image and host image is done with the help of gain factor , which is varied accordingly from 0.1 to 0.7.

(a)

(b)

(c)

(d)

Figure 2: (a) Cover image (b) Watermark (c) Watermarked (d) Extracted watermark

Table 1 shows the comparison between PSNR and NC performance of the proposed method at different gain factor without any attack. In this table, gain factor k is varied from 0.1 to 9. The highest PSNR (52.56) has been obtained at gain factor 0.1. However, lowest PSNR (44.38) is obtained at gainfactor 0.9. The highest NC (1) value has been obtained at gain factor (0.9) and lowest NC (0.94) obtained at gain-factor 0.1. It is observed that larger the gain factor, stronger is the robustness and smaller the gain factor better image quality.

B. Extraction Process The reverse processing of the embedding process is used for the extraction of watermark image.

Table 2, shows PSNR and NC performance of the proposed method at different gain factor and using different sub-bands. PSNR value for different sub-bands are calculated at different gain-factors. The maximum PSNR value has been obtained with LH sub-band is 38.15 dB at gain-factor 0.1. However, the minimum PSNR value has been obtained with HH sub-band is 27.14 dB at gain-factor 0.3 Table 3 shows NC performance of proposed method against different attacks at gain-factor = 0.01. The highest value of NC is 0.88 for Gaussian attack and lowest value of NC is 0.75 for Contrast attack.

(a)

Table 4 shows PSNR and Compression Ratio of proposed method at LH sub-band at gain-factor 0.01. The highest value of compression ratio is 3.92 for MRI image. However, the lowest compression ratio is 1.78 for Brain image. These results shows that the Brain image take less time to compress than the other two images. Rate of compression for Brain image is better than Lung and MRI images. Table 1: PSNR and NC performance of the proposed method at different gain factor without any attack (b) Fig. 1: Watermark (a) Embedding and (b) Extraction process

IV. EXPERIMENTAL RESULTS AND ANALYSIS In this section, the performance of combined DWT and Huffman based watermarking method have been evaluated in

2

Gain Factor(k)

PSNR

NC

0.1

52.56

0.94

0.3

51.4

0.96

0.5

49.5

0.97

0.7

46.56

0.99

0.9

44.38

1

International Journal of Applied Engineering Research, ISSN 0973-4562 Vol. 10 No.69 (2015) © Research India Publications; httpwww.ripublication.comijaer.ht

Table 5 shows comparison between Huffman, BTC and Fractal based compression. It is observed that the increasing gain factor, PSNR decrease in case of all considered compression methods. Refereeing this Table again the Huffman based compression method provides better result than BTC and Fractal based compression technique. The highest value of PSNR using Huffman compression is 45.14 dB at gain factor 0.1. However, maximum PSNR has been obtained with BTC and Fractal compression methods are 34.17 dB and 29.17dB respectively. The compressed image will be exactly same as original image.

imperceptibility, robustness and capacity as compared to DWT applied individually. Due to the excellent properties of DWT, the DWT is suitable to identify the area in the cover image which is suitable for embedding watermark. Huffman based compression technique has less compression ratio and improved the robustness and perceptual quality of the watermarked image is better than BTC and fractal based compression techniques. The main properties of the proposed work can be identified as follows:(1) In the proposed method, we have embedded compressed watermark into different sub-bands of host image and desired results are obtained as watermark is compressed to embed more information which is useful in medical application.(2)Huffman coding is based on the particular frequency occurrence of a data item(pixel in images). It uses a lower number of bits to encode the data that occurs more frequently. Based on various applications such as medical applications, where the pixel of any data is more important, so any distortion in the received data may cause the wrong diagnoses both lossy and lossless have importance at their own places. Lossy compression is used to compress multimedia data such as audio, video and still images.

Table 2: PSNR and NC performance of the proposed method at different gain factor using different sub-band Gain Factor (K)

PSNR LH

HL

HH

0.1

LL 36.52

38.15

32.92

30.45

0.2

30.15

36.13

30.19

29.34

0.3

29.83

32.84

29.81

27.14

Table 3: NC performance of proposed method against different attacks at gain factor=0.01 Attacks NC Brightness attack 0.82 Rotation attack 0.81 Contrast attack 0.75 Salt and Pepper 0.76 Gaussian Noise 0.88

Acknowledgment The Author’s are sincerely thankful to the potential/ anonymous reviewer’s for their critical comments and suggestions to improve the quality of the paper.

Table 4: PSNR and Compression ratio of proposed method at LH sub-band at gain factor=0.01 Compression Different Image PSNR Ratio Brain Image

28.21

Lung Image

29.14

2.01

MRI Image

37.14

3.92

References [1] Da-Chun Wu and Wen-Hsiang Tsai, “Data hiding in image via multiple based number conversion and lossy compression”, IEEE transaction on consumer electronics, Vol. 44(4), pp. 1406-1412, 1998. [2] S. A Mustafa, E.-Sheimy and N, Tolba. “Wavelet packets-based blind watermarking for medical image managemen”t. The open biomedical engineering journal Vol. 4, pp. 93-98, 2010. [3] H.Peng, J. Wang and W.Wang “Image watermarking method in multiwavelet domain based on support vector machines”, Journal of Systems and Software Vol. 83(8), pp. 1470-1477, 2010. [4] N. Nikolaidis, and I. Pitas “Digital image watermarking: an overview”, IEEE International Conference on Multimedia Computing and Systems, Vol. 1, pp. 1-6, 1999. [5] G. Bhavnagar and B. Subramanian Raman , “A new robust reference watermarking scheme based on DWT-SVD” Computer Standards & Interfaces, Vol. 31(5), pp. 1002-1013, 2009. [6] A. K. Singh, B. Kumar, M. Dave and A Mohan, “Multiple Watermarking on Medical Images Using Selective DWT Coefficients”, Journal of Medical Imaging and Health Informatics, USA, Vol. 5(3), , pp. 607-614, 2015. [7] R.Das, and Th.Tuithung, “A Novel Steganography Method for Image Based on Huffman Encoding”, IEEE Conference on Emerging Trends and Applications in Computer Science (NCETACS), Shillong, pp.14-28, 2012. [8] W.Adiwijaya, P. N. Faoziyah, F. P. Permana, T. A. B. Wirayuda and U.N. Wisesty, “Tamper Detection and Recovery of Medical Image Watermarking using Modified LSB and Huffman Compression”, Second International Conference on Informatics and Applications (ICIA), Lodz, pp. 129-132, 2013. [9] A. Nag, S. Biswas, D. Sarkar and P. P. Sarkar, “A Novel Technique for Image Steganography Based on Block-OCT and Huffman Encoding”.

1.78

Table 5: Comparison between Huffman, BTC and Fractal based compression

Gain Factor(k)

BTC Compression

Huffman Compression

Fractal Compression

PSNR

NC

PSNR NC

PSNR

NC

0.1

34.17

0.90

45.14 0.95

29.14

0.84

0.4

31.15

0.94

43.24 0.97

28.87

0.87

0.7

30..43

0.96

41.14 0.98

27.87

0.90

V. CONCLUSIONS In this paper, a new approach for digital image watermarking has been proposed. The proposed watermarking technique using fusion of DWT and Huffman compression technique achieves better performance in terms of

3

International Journal of Applied Engineering Research, ISSN 0973-4562 Vol. 10 No.69 (2015) © Research India Publications; httpwww.ripublication.comijaer.ht International Journal of Computer Science and Information Technology, [15] A. K. Singh, M. Dave and A. Mohan, “Wavelet Based Image Watermarking: Futuristic Concepts in Information Security”, Vol. 2(3), pp. 48-58, June 2010 Proceedings of the National Academy of Sciences, India Section A: [10] A. P. Singh, and A. Mishra, “Wavelet Based Watermarking on Digital Physical Sciences. Vol. 84(3), pp. 345-359, 2014. Image” Indian Journal of Computer Science and Engineering Vol. 1(2), pp. 86-91, 2009. [16] A. Cheddad, J. Condell and K. Curran and Paul Mc Kevitt, “Digital [11] B. W. Tjokorda Agung and F. P. Permana, “Medical Image Image Steganography: Survey and Analysis of Current Methods” Watermarking with Tamper Detection and Recovery using Reversible Journal on Signal Processing Vol. 90(3), pp .727-752, 2010. Watermarking with LSB Modification and Run Length Encoding [17] M. Arsalan, Sana Malik and Asifullah Khan, “Intelligent reversible Compression, IEEE International Conference on Communication, watermarking in integer wavelet domain for medical images”,ACM Journal of Systems and Software, Vol. 84(4), pp. 883-894, April 2013. Network and Satellite, Bali, pp.167-171, 2012. [18] Gunjal, Baisa L. and R. R. Manthalkar, "An overview of transform [12] A. Sergio Alvarez Fulton, “Algorithms Notes on Lossless Data domain robust digital image watermarking algorithms" Journal of Compression and Huffman Coding”, USA, pp. 383-389. [13] Nilanjan Day, A. Bardhan Roy, and S. Dey “A novel approach of color Emerging Trends in Computing and Information Sciences Vol. 2(1), pp. image hiding using RGB color planes and DWT”, International Journal 37-42, 2010. TM of Computer Applications, Vol. 365(5), pp. 19 – 24, 2011. [19] MedPix Medical Image Database available at [14]S. M. Perumal and V. Vijaya Kumar, “A wavelet based digital http://rad.usuhs.mil/medpix/medpix.html. watermarking method using thresholds on intermediate bit values”, International Journal of Computer Applications, Vol. 15(3), pp. 29-36, 2011.

4