image steganographic techniques with improved ... - IEEE Xplore

7 downloads 0 Views 507KB Size Report
domain [2][3], have certain robustness against other multimedia processing. This paper, intends to implement two methods of image steaganography, namely ...
IEEE-International Conference on Recent Trends in Information Technology, ICRTIT 2011 MIT, Anna University, Chennai. June 3-5, 2011

IMAGE STEGANOGRAPHIC TECHNIQUES WITH IMPROVED EMBEDDING CAPACITY AND ROBUSTNESS J.Anita Christaline1 , D. Vaishali2

Assitant Professors – ECE, SRM University, Chennai. India. 1 [email protected] , [email protected]

those in transform domain. Information hiding algorithms in time domain or spatial domain have high capacity and relatively lower robustness. In contrast, the algorithms in transform domain, such as DCT [1], Wavelet domain [2][3], have certain robustness against other multimedia processing. This paper, intends to implement two methods of image steaganography, namely the filter method (time domain) for high capacity text communication and the wavelet method (transform domain) for highly secured image communication. While the first method provides high embedding capacity, the second method focuses on high robustness.

Abstract Secured data transmission over computer networks can be achieved through steganography. In specific, Image Steganography entails the opportunity of hide any secret information into images. This paper presents the implementation of two image steganographic techniques in MATLAB. The first is a filter method to embed text information into image and new methods have been demonstrated to increase the information embedding capacity in the same domain. The second method is the wavelet transform method which proves to be more secured than any other method of image steganography. Keywords – Information hiding, steganography, masking and filtering method, transform method, stego image, carrier, key.

II. IMAGE STEGANOGRAPHIC APPROACHES There are many techniques for hiding information or messages in images in such a manner that the alterations made to the image are perceptually indiscreminable. The most common image steganographic approaches include : • Least significant bit insertion (LSB). • Masking and filtering. • Transform techniques.

I. INTRODUCTION Recent years have witnessed the rapid development of the Internet and telecommunication techniques. In such a scenario, information security is becoming more and more important. Applications such as covert communication, copyright protection, etc, stimulate the research of information hiding techniques. Traditionally, cryptography was used to realize secured communication. In cryptography, important information is not protected as the cipher text message can arouse suspicion. This necessitated the need for hiding information rather than modifying the information. The secret information is initially hidden in a host data, such as digital image, video or audio, etc. and then transmitted secretly to the receiver. Generally, information hiding techniques can be classified under two categories, namely, information hiding techniques in time domain or spatial domain and

978-1-4577-0590-8/11/$26.00 ©2011 IEEE

A. Masking and Filter method of Image Steganography: The masking and filtering method is usually restricted to 24 bit images. This method uses specific masking algorithms or mathematical formulae to select specific pixels from the cover image. The chosen pixels can be used to embed the secret information after due pre-processing. Thus the embedded secret information looks as an integral part of the cover image [4]. By using proper filter (mathematical expression) to select the pixels, the security of

97

IEEE-ICRTIT 2011

4D dictionary of words to work with a specific word compression and retrieval logic. E. The Dictionary Approach for increased embedding capacity: This method has two parts, namely the word compression logic in the sender side and the word retrieval logic in the receiver side. Word compressing logic Input: Text information. Output: Compressed text. 1. Read the first (or next) word. 2. Keep the first and the last letter intact. 3. Delete the alternate letters from the remaining word. 5. Repeat the above for all words. Word Retrieval logic Input: Compressed text. Output: original text 1. Read the first (or next) compressed word. 2. Keep the first and the last letter intact. 3. Calculate the number of letters in the word 4. Perform the search in the 4 – D Dictionary. 5. Identify the word. 6. Repeat for all compressed words.

the image steganograpophic method is enhanced and the embedding capacity can be increased by choosing many pixels. B. Algorithm for Embedding Logic in Filter method: Input : Secret Text message, Carrier image. Output : Embedded (Stego) Image. 1. Choose a set of pixels from the carrier image using a mathematical function (KEY). 2. The gray / red level values of these pixels are processed. 3. Convert secret text information into bit stream (ASCII coding). 4. If the first bit in the text bit stream is zero, then the first pixel value is not modified. 5. If the first bit happens to be one, then the first pixel is made to be odd by adding one to the pixel value. 6. Repeat above logic till the text bit stream expires. 7. The modified pixels are put back into the original cover image to get the Stego Image. C. Algorithm for Debedding Logic in Filter method: Input : Embedded (Stego) Image, Key. Output : Secret text message. 1. Identify the specific pixels in stego image by using the same mathematical function (KEY). 2. Examine the pixel values. If the pixel value is odd, then the embedded text bit was a one. 3. If the pixel value is even, then the embedded text was a zero. 4. Based on the above interpretation, retrieve the text bit stream. 5. Convert bit stream into ASCII and then to the respective character to get the secret text. D. Algorithm to Increase the Embedding Capacity: Embedding capacity is the maximum number of characters that can be embedded in an image. It depends on size of the image, the key used to select pixels, type of coding. Thus ways to increase embedding capacity could be to use an user defined coding (either 5 or 6 bit coding) rather than the usual 8 bit ASCII. Also some word compression logic can be used to reduce the number of characters per word to increase the embedding capacity. This paper intends to implement 5 bit user coding to increase the embedding capacity and a

F. The 5 bit User Coding for increased embedding capacity: Text Coding logic Input: Compressed Text Output: Binary Stream based on User Coding. 1. Read the first (or next) word. 2. Convert it into binary based on the user coding. 3. Construct the binary stream. Text decoding logic Input: Binary Stream Output: Compressed Textual Data 1. Read the first (next) 5 bits from text bit stream. 2. convert to equivalent character based on User Coding. III. WAVELET BASED IMAGE STEGANOGRAPHY Wavelet transform is used to convert a spatial domain into frequency domain [5]. The use of wavelet in image stenographic model lies in the fact that the wavelet transform clearly separates the high frequency and low frequency information on a pixel by pixel basis. Discrete Wavelet Transform (DWT) is preferred over Discrete Cosine Transforms (DCT) because image in low frequency at various levels can offer corresponding resolution needed. The DWT decomposition results into four classes or

98

Image Steganographic Techniques with Improved Embedding Capacity and Robustness

MATLAB GUI showing Embedding Logic

band coefficients, namely, approximate band (LL), Vertical Band (LH), Horizontal band (HL), and diagonal detail band (HH). The approximation band consists of low frequency wavelet coefficients, which contain significant part of the spatial domain image. The other bands also called as detail bands consists of high frequency coefficients, which contain the edge details of the spatial domain image[6]. It is therefore expected that use of the DWT will allow independent processing of the resulting components without significant perceptible interaction between them, and hence makes the process imperceptibility marking more effective.

MATLAB GUI showing comparison of original carrier image and embedded stego image

A. Embedding logic for DWT image steganography: Input : Carrier image (KEY) , Secret image. Output : Stego image. 1.Apply DWT ( db4) on to the carrier image to get the decomposed carrier image (Four bands) 2.Apply DWT (db4) on to the secret Image to get the decomposed secret image (Four bands) 3. Apply wavelet fusion of DWT coefficients of the carrier and secret images. 4. Take Inverse DWT with the four sub bands of the fused image to get the Stego image.

MATLAB GUI showing the retrieval in receiver end

B. Debedding logic for DWT image steganography: Input : Stego image, carrier image (KEY) Output : Secret image. 1. Apply DWT (db4 ) on to the received Stego image to get the four sub bands ( wavelet coefficients ) 2. Apply DWT (db4) on to the carrier image (KEY) , to get the four sub bands ( wavelet coefficients ). 3. Subtract DWT coefficients of carrier image from that of the Stego image to get the DWT image of the secret image. 4.Apply Inverse DWT to get the secret image.

MATLAB GUI showing the implementation of Dictionary and user coding to increase embedding capacity

MATLAB GUI showing the retrieval with Dictionary and user coding in receiver end

IV. EXPERIMENTAL RESULTS A. Experimental results for Filter method:

99

IEEE-ICRTIT 2011

Retrieved image

B. Experimental results for Wavelet method: Original secret image to be transmitted

V. PERFORMANCE EVALUATION A. Filter Method For the filter method, the performance is evaluated by the Comparison of maximum embedding capacities as in following tables.

Carrier image

COMPARISON OF DATA FOR FILTER METHOD

Image File Name

Peppers. png

rice.png Stego image that is transmitted

Key used

No. of Pixels Selected

50x50

88

Max. no. of Characters that can be Embedded. ASCII 5 bit UC 11

10x10

2028

253

405

Alternate pixels All pixels

98304

12288

19660

196608

24576

39321

50x50

36

4

7

10x10

676

84

135

Alternate pixels All pixels

32768

4096

6553

65536

8192

13107

COMPARISON FOR DIFFERENT CODING METHODS FOR HIGH EMBEDDING CAPACITY

Type of Steganography 8 -Bit Steganography 6 – Bit Steganography 5 – Bit Steganography 8 – Bit Steganography with Dictionary Compression 6 – Bit Steganography with Dictionary Compression 5 – Bit Steganography with Dictionary Compression

100

17

% Increase in Embedding capacity 0 ( ref.) 25 % 38 % 37 % 62 % 75 %

Image Steganographic Techniques with Improved Embedding Capacity and Robustness

[2] Nedeljko Cvejic, Tapio Seppanen. A Wavelet Domain LSB Insertion Algorithm For High Capacity Audio Digital Signal Processing, Steganography,

B. Wavelet Method Performance evaluation for the DWT ( db4) method has been done in terms of MSE, PSNR and Entropy [7] of the images.

workshop 2002 and the 2nd signal processing education workshop. Proceedings of 2002 IEEE, 10th, 13-16, Oct, 2002, Pages: 53-55.

COMPARISON OF MSE

Images

Type

Bust (secret) mask (carrier) Woman (secret) mask (carrier)

JPEG

JPEG

Size

256X256

256X256

MSE Between Carrier and stego image 1.000

1.000

[3] NI Ronggong, RUAN Qiuqi. Embedding Between secret and retrieved image

Information

2.32 e -25

Bust (secret) mask (carrier) Woman (secret) mask (carrier)

Color

Images

Using

Wavelet,

[4]Sahoo1 and R. K. Tiwari, Designing an Embedded Algorithm for Data Hiding using Steganographic Technique by File Hybridization, IJCSNS International Journal

-2.65 e 27

of Computer Science and Network Security, VOL.8 No.1, January 2008. [6] H S Majunatha Reddy, & K B Raja, HIGH CAPACITY AND SECURITY STEGANOGRAPHY USING DISCRETE WAVELET TRANSFORM, International

COMPARISON OF PSNR AND ENTROPY

images

into

TENCON’02, Proceedings, 2002 IEEE Region 10 Conference on Computers, Communications, Control and Power Engineering, Volume:1, 2831 Oct, 2002, pages: 589-601, Vol.1.

PSNR Stego image 21.45

Retrieved image 24.45

Entropy Stego image 0.0

Journal of Computer Science and Security (IJCSS), Volume (3): Issue (6)

Retrieved image 0.0

[5] Bo Yang and Beixing Deng, Steganography In Gray

22.17

26.036

0.0012

Images

Using

Wavelet,

http://online.redwoods.cc.ca.us/instruct/darnold/ maw/catmap.htm

0.0012

[7] FRANK Y. SHIH , Image Processing and Pattern Recognition, Fundamentals and Techniques, John Wiley & Sons, Inc., Hoboken, New Jersey., 2010, PP.71.

VI. CONCLUSION AND FUTURE SCOPE This paper has implemented two different steganographic methods. The filter method shows that the embedding capacity can be increased and the wavelet method shows that the security of data is highlightened. In future, this paper can be extended to improve the robustness of the filter method by improving the KEY used for pixel selection. For the wavelet method, the future work may include considering different mother wavelets to enhance the robustness. Also various class of images can be considered which will improve the MSE and PSNR. REFERENCES [1] Chiou-Ting Hsu, Ja-Ling Wu. Hidden Digital Watermarks In Images, IEEE Transaction on Image Processing, 8(1): 56-58,January 1999.

101