INTEGER WAVELET TRANSFORM BASED APPROACH FOR HIGH ROBUSTNESS OF AUDIO SIGNAL TRANSMISSION

Transcription

1 Volume 116 No , ISSN: (printed version); ISSN: (on-line version) url: ijpam.eu INTEGER WAVELET TRANSFORM BASED APPROACH FOR HIGH ROBUSTNESS OF AUDIO SIGNAL TRANSMISSION 1 R.Punidha, 2 M.Sivaram 1 Professor/CSE, Bharathiyar Institute of Engineering for Women, Salem. 2 Assistant Professor/CSE, Rajagiri School of Engineering & Technology, Kakkanad. Abstract: Steganography is the practice of concealing a file, message, image, or video within another digital media. In this modern world sharing of any digital media becomes tremendously insecure. Steganography is an efficient method introduced for the purpose of data hiding. An audio speech signal is transmitted at the transmitter side. Reshuffling and secret arrangement of approximate coefficients of an audio signal are hidden in thecb component of the cover image using Integer Wavelet Transform (IWT). The wavelet used is Daubechies wavelet of level 2 (db2). Integer Wavelet Transform maps integer data set to integer data set. Embedding the audio signal within the cover file based on embedding procedure. At the receiver side using Inverse Integer Wavelet transform and extraction, the original signal is been retrieved. In the existing method, the haar wavelet process is weaved into steganography to conceal secret messages using Integer Wavelet Transform (IWT). The drawback of haar wavelet is that it is a primary wavelet. The simulation tool used in this project is MATLAB. The Integer Wavelet Transform (IWT) is mainly used for signaling techniques and the proposed work is also based on applying IWT to an audio speech signal. Keywords: Digital Signal Processing, Decoding, Encoding, Integer Wavelet Transform, Sampling, Wavelets 1. Introduction The growth of Internet is increasing at tremendous speed and it is directly proportional to the size of multimedia data. A digital data can be easily transmitted, received, duplicated or modified by using the internet. This rapid development of computer network and technology increases the communication in the form of audio, video and image. This leads to the problem of unauthorized data user and copyright breach, is a major concern when information is transmitted over networks, especially in the Web environment. Steganography is one of the techniques to overcome these types of problems. It is the process to plunge a data into other digital products such as images, audios and videos. This embedded data can later be detected or extracted from the multimedia by the receiver for many purposes. Steganography is a technique to hide an audio signal in an image. Since audio files contain large number of samples even for a small duration, the cover image has to be considerably large. Color images are suitable because of large hiding space. Since YCbCr approach is more secure than RGB approach, YCbCr approach is used [1]. The application chosen here is hiding an image inside an image using Daubechies Wavelet.The signal is splitted intohigh and low frequency parts. Using Discrete Wavelet Transform (DWT). The high frequency part contains information about the edge components, while the low frequency part is splits again into high and low frequency parts in [2]. A method to hide encrypted text in cover audio using Lifting Wavelet Transform (LWT) based on the coefficient values, number of bits used to hold secret data is chosen. Text is encrypted based on the message size and then hidden in cover audio [3].The complex information is decomposed into elementary forms at different positions and scales and subsequently reconstructed with high precision values.wavelet transforms allows the components of a non stationary signal to be analyzed. Wavelets allow filters to be constructed for stationary and non-stationary signals [4]. The encoding and decoding operation in the frequency domain is proposed. The text message is hidden in the Quick Response (QR-code) image. The Quick Response (QR-code) image is hidden into the Discrete Wavelet Transform (DWT). DWT performs well and the security of information is high [5].Audio steganography scheme based on sample comparison in Discrete Wavelet Transform (DWT) domain is that the selected coefficient of a segment value is changed by a threshold value depending upon the embedding cipher text bit in the cover image pixel values [6].Audio steganography coding is a technology of transforming stego speech into efficiently encoded version that can be decoded at the receiver side to extract a close representation of the uncompressed initial signal [7].To minimize the mean 295

2 squared distortion between the original and the watermarked images given a payload, the watermark is adaptively embedded in the cover image [8].Spatial domain image steganography has been used because of its compatibility to images. To increase the capacity of hidden data in a way that security could be maintained. The Most Significant Bit (MSB) of the randomly selected pixel has been used as an indicator [9]. The use of Discrete Wavelet Transform (DWT) will mainly address the capacity and robustness of the information hiding system features. It describes and implements steganography in the Wavelet domain. The hierarchical nature of the Wavelet representation allows multi resolution detection of the hidden messages, which is a Gaussian distributed random vector added to all the high pass bands in the Wavelet domain [10]. The urged work is that Integer Wavelet Transform based audio steganography. Integer wavelet transform is an emerging technique after DWT. During reconstruction of signal there will be no loss in the data whereas it acts as one of the drawbacks of DWT. Parametric analysis also shows higher results when compared to DWT. 2. Wavelet And Integer Wavelet Transform Wavelets can be used to extract information from many different kinds of data, but certainly not limited to audio signals and images. The set of complementary wavelets are useful in wavelet based compression and decompression algorithms where it is desirable to recover the original information with minimal loss. Integer wavelet transform is a type of wavelet transform which maps integer data set with another integer data set. The reason of choosing Integer Wavelet Transform (IWT) over Discrete Wavelet Transform (DWT) is that Integer Wavelet Transform maps the integer data set to another integer dataset whereas Discrete Wavelet Transform converts the floating point values to integer by truncation process. Truncation results in the loss of data. When the input data is in the form of integer, perfect reconstruction is possible while applying Inverse transform. As mentioned the difference between Integer Wavelet Transform (IWT) and Discrete Wavelet Transform (DWT), the LL (Low-Low) sub band in the case of Integer Wavelet Transform appears to be a close copy with smaller scale of the original image, while in the case of DWT resulting LL sub band is distorted. The Integer Wavelet Transform (IWT) has the important property that its coefficients have the same dynamical range as the original signal. This makes easier implementation considerations regarding the size of the variables to be used and the ranges to provide for in the coding algorithm. In the case of audio signal, the bordering effects are not too critical, since they manifest only at the beginning and end of the file, which leads to a marginal effect on the coding rate. For a particular Integer Wavelet Transform number of samples decides the decomposition level from time moments, in order to perform onelevel decomposition. A. Haar Wavelet The operation of Haar wavelet on data is by calculating the sum and difference of adjacent elements. The wavelet operates first on the adjacent horizontal elements and then on adjacent vertical elements. One nice feature of the Haar wavelet transform is that the transform and its inverse are equal. A Haar wavelet is the simplest type of wavelet. Like all wavelet transforms, the Haar transform decomposes a discrete signal into two sub signals of half its length.some of the advantages of Haar Wavelet are: 1) Haar Wavelet is a basic wavelet so it is conceptually simple. 2) A temporary array is not required since it is memory efficient. 3) It is exactly reversible without the edge effects that are a problem with other wavelet transforms. The limitations of Haar wavelet are: 1) In generating averages and coefficients for the next level and, the Haar transform performs an average and difference on a pair of values. 2) The algorithm shifts over by two values and calculates another average and difference on the next pair B. Daubechies Wavelet (db2) Daubechies wavelet is the family of orthogonal wavelets which are characterized by vanishing moments. Vanishing moments smoothenthe images. These are also called as Maxflat wavelets as their frequency responses have maximum flatness at frequencies 0 and π. The Daubechies wavelet transforms are defined in the same way as the Haar wavelet transform by computing running averages and differences via scalar products with scaling signals and wavelets. The only difference between the description on scaling signals and wavelets are defined. C. Problems in the existing work In the existing work the wavelet used is a Haar Wavelet which is a primary Wavelet and the wavelet is not continuous. There are many orthogonal wavelets which provide more security than the primary wavelet. SNR achieved is not sufficiently high. The work was not 296

3 exposed to different attacks to calculate its efficiency. In order to overcome the demerits of the existing work, the proposed work is been designed with different process. Objective is to improve Signal to Noise Ratio (SNR), Peak Signal to Noise Ratio (PSNR), Structural Similarity Index Metric (SSIM). The Parameters calculated is high when compared to audio steganography using DWT. 3. Hiding an a Audio in the C b Component of an Image Steganography methods proposed for audio signals are huge. The audio steganographic methods uses different techniques such as spread spectrum, echo hiding and support vector regression etc.,among the existing steganographic techniques Integer Wavelet Transform (IWT) for audio types shows the robustness of the hidden stego. It can also achieve good imperceptibility and high level of security. A. Encoding at the transmitter end RGB image is taken and the image is been converted into YCbCr format for better hiding. From the YCbCr image the three components are extracted namely Y, Cb and Cr components. The Cb component obtained is divided into four sub bands by applying Integer Wavelet Transform (IWT). The four sub bands are namely Low-Low (LL), low-high (LH), High-Low (HL), High-High (HH). The audio signal which is to be hidden is taken in the.wav format. Similarly IWT is applied on the audio file to divide the audio into approximate and detailed coefficients.the block diagram cited in Fig.1 sights a secured encryptionfor audio transmission. Wavelet Transform (IIWT) is applied on the Cb component to get the stego image. B. Embedding process In this process, two bits of the secret message are hidden in one byte of the cover image. Two bits from the secret are XOR ed with 5 th and 4 th bits of the cover byte to get encrypted secret bits. Suppose S1 and S0 are two secret bits, thens1 = S1 XOR b5 XOR b4 and S0 = S0 XOR b5 XOR b4, where b5 and b4 are 5 th and 4 th bits of the cover byte respectively. 3 rd and 2 nd bits of the cover byte are replaced by these encrypted secret bits. This type of dynamic encryption avoids the need for encryption key. The process of embeddingthe audio in the Cr component is also in the similar fashion C.Extraction process In the extraction process the input is given as a stego image and the output retrieved is the hidden audio. The block diagram shown in Fig.3 is the extraction process of the hidden audio. Stego image is read in RGB format. The RGB format stego image is converted to YCbCr format. The 2D Integer Wavelet Transform (IWT) is applied on the Cb component of the stego image to get the four sub bands. Extract the audio file from the 2 nd and 3 rd bit locations of the cover image. Applying Inverse Integer Wavelet Transform (IIWT) to get the approximate coefficients of the audio and considering the detail coefficient values of audio signal to be zero. The original audio is been retrieved from the stego image without any loss of information.the block diagram for embedding process is shown in Fig.2. Figure 1. Encoding at the transmitter end Approximate coefficients have been preferred because there will be no loss of data while retrieving. Approximate coefficients is been hidden into the Low- Low (LL) subbands of the image. Inverse Integer Figure 2. Embedding process The stego image is read and represented in YCbCr format. Obtain Integer Wavelet Transform (IWT) of Cbto get four sub bands LL, HL, LH, and HH. Extract the encrypted secret audio bits from the second and third bit 297

4 planes of LL sub band. In this method, two encrypted bits of the secret message are obtained from one byte of the stego image coefficient. Then decryption is done as follows: the two encrypted bits are XOR ed with 5 th and 4 th bits of the stego byte to get secret bits. Convert to decimal to get approximation coefficients of secret audio. Apply Inverse Integer Wavelet Transform (IIWT) for approximation coefficients and considering zeroes for detailed coefficients. The result is the secret audio. In the current endeavor, an audio file with.wav extension has been selected as host file. It is assumed that the least significant bits of that file should be modified without degrading the sound quality. Basically the file structure of the audio file should be known. Like most of the files.wav files have two basic parts, the header and the data. Normally in.wav files, the header is situated in the first 44 bytes of the file. Except the first 44 bytes, the rest of the bytes of the file are all about the data.the data is just one giant chunk of samples that represents the whole audio. While embedding the data, the section to be considered is the header section. That is because a minimal change in the header section leads to a corrupted audio file. A program has been developed in which the audio file is read bit by bit and stores them in adifferent file. The first 44 bytes should be left without any change in them because these are the data of the header section. D. Daubechies wavelet and Integer Wavelet Transform Daubechies wavelets are a family of orthogonal wavelets defining a discrete wavelet transforms and characterized by a maximal number of vanishing moments for some given support. The block diagram of the Extraction process is shown in the Fig.3. decomposition of the original image. This is based on the four magic numbers h0, h1, h2 and h3.daubechies wavelet is more complicated when compared to Haar wavelet. The standard Daubechies wavelet of level 2 is shown.the equations for standard daubechies wavelet are cited in the equations (1) and (2). an=s2n+ 3S2n+ 1 cn=s2n+ an 3 a 4. Results and Analysis The result shows that the audio is hidden secretly in the RGB image. The results which are as follows shows the input image in RGB format used is of dimension 512*512, audio signal of samples. The cover image and secret audio are the two inputs. The input image is chosen as a RGB image because the number of hidden sample is high.the audio file is of.wav format and the audio signal is not a real time audio. The audio plays for 2.52 sec and the signal when discretized is of samples.the RGB image is been converted to YCbCr format for more secured communication which is one of the model to represent the image in image processing. Extract the Cb component from the YCbCr model. 2D Integer Wavelet transform is applied on the Cb component and 1D Integer Wavelet transform is applied on the audio signal.the input image is of 512*512 dimensions and it s a Lena image from Matlab. The audio signal is of samples.the Cover image used is shown in the Fig.4.The audio signal to be hidden in the cover image is shown in the Fig.5. Figure 3. Block diagram of Extraction process This kind of 2D Discrete Wavelet Transform (DWT) aims to decomposethe image into approximation coefficients (ca) and detailed coefficient ch, cv and cd (horizontal, vertical and diagonal) obtained by wavelet Figure 4. Cover Image 298

5 Figure 5. Audio signal The Cb component of an image divides into four sub bands (LL, LH,HL,and HH). The audio divides into approximate and detailed coefficients. The approximate coefficients are hidden in the LSB of LL sub bands. The approximate coefficients are hidden because the signal when retrieved will not produce any loss in an image. Component of the cover image is shown in the Fig.6 and LL sub band of component of cover image is shown in Fig.7. HL, LH, HH sub bands of the Cb component are shown in the Fig Figure 7. LL sub band of component of input image Figure 8. HL sub band of component of input image. Figure 6. Component of an image Figure 9. LH sub band of component of input image 299

6 Figure 10. HH sub band of component of input image The audio signal is been embedded in the image. Read the stego image and by applying IWT to get four sub bands (LL,LH,HL,and HH). Extract the encrypted secret audio bits from the second and third bit planes of LL sub band and then perform decryption. The Stego image obtained is shown in the Fig.11. Figure 12. Stego image as an input image Figure 13. Retrieved signal Figure 11. Stego Image The approximation coefficients of secret audio should be converted to decimal to retrieve the original audio coefficients.obtain inverse IWT for approximationcoefficients and considering zeroes for detailed coefficients. The LL sub band contains information and the approximate coefficients to be hidden using IWT. During the extraction the stego image is given as the input and is shown in the Fig.12. The extracted output is shown in the Fig.13. The output shows that the audio has been hided successfully in the LSB of the LL sub band. The parameter calculations only show the result of losses. i. Parameters A. Peak Signal to Noise Ratio PSNR is defined as the Mean Square Error (MSE) and the Peak Signal to Noise Ratio (PSNR) used to compare image compression quality. PSNR metric is calculated using formula mentioned in the equation (3), PSNR=10 log (3)!" MAX is the maximum value of the original image pixel 512 for an RGB image. MSE is the mean square error between the original and stego images. MSE is calculated by using the formula is enumerated in the equation (4), MSE= % % 10 /0 O)i,j- D)i,j- (4) 300

7 O (i,j) is original pixel and D (i,j) is stego pixel. Greater the PSNR value indicates better quality of an image. It is expressed in terms of decibels. B. Signal to Noise Ratio SNR is defined as the ratio of signal power to noise ratio.snr is calculated by using the formula referred in the equation (5), SNR=10 log 3 % !" (5) where,mse= )x % :0 : y : -,xi is original sample and yi is the stego sample. C. Structural Similarity Index Metric (SSIM) Structural Similarity Index is used to check the quality of an image or image degradation. SSIM is calculated using the formula specified in the equation (6), SSIM= )= E >@AB=F )E > AB=- )>?A@?ABC- (6) Where, c1=)k1l- and c2=)k2l-, k1= 0.01, k2=0.03 are the constant values used.l is the dynamic range of the pixel values. SSIM is superior to traditional measures based on image quality metric such as Mean Square Error (MSE) and Peak Signal to Noise Ratio (PSNR). PSNR estimates the perceived errors, whereas SSIM considers image degradation as perceived change in structural information. The dynamic range of SSIM is between 0 and 1. Maximum value of 1 will be obtained for identical images. Cover image is of dimension 512*512. Audio file is of samples samples. When converting RGB to YCbCr component Cb component is of size 512*512.The SSIM value is PSNR is calculated for different types of noises and addition of filters. Noises such as Gaussian noise, Salt and Pepper Noise, Speckle noise, Poisson noise. Filters such as Median filter, Weiner filter, and spatial filter. PSNR calculation for noise attack on the stego image is shown in the graph in the Fig. 14.Table I refers to SNR, PSNR, SSIM metric calculation and Table II shows the PSNR, SNR comparison Figure 14. PSNR calculation for Noise Attack on the Stego Image Table 1. SNR, PSNR, SSIM Metrics Calculation Cover Image 512*512 Secret Audio Samples STEGO SSIM PSNR (db) lena.tif lena.tif lena.tif Extracted SNR in Db Table 2. Comparison Of Snr And Psnr Of Existing Work With Urged Work Parameters Calculated SNR(dB) PSNR(dB) Existing method Proposed Method 16565samples samples samples samples samples samples

8 The four sub bands are divided and each subband is of 128*128 pixels. Stego image is of size samples. The Table III represents File size after embedding. Table 3. Size of Files After Embedding File Audio RGB image Cb Component of cover image LL sub band Stego image 5. Conclusion File size samples 512*512 pixels 512*512 pixels 128*128 pixels samples The secret audio is been hidden in a RGB image using Integer Wavelet Transform (IWT). Discrete wavelet transform (DWT) maps the integer values by converting the floating point values to integer by truncation process. The truncation process results in the loss of data.integer Wavelet Transform (IWT) overcomes this drawback. The wavelet used is the Daubechies Wavelet of level 2 (db2).daubechies 2 (db2) provides a high level of security to protect the data over the network. The efficiency of data hiding (audio) is been improved by using Daubechies wavelet compared to Haar. One of the drawbacks of Haar wavelet is that the signals are not continuous. Efficiency is been calculated by using Signal to Noise Ratio, Peak Signal to Noise Ratio (PSNR), Structural Similarity Index Metric (SSIM). The parameters calculated shows that Daubechies wavelet shows higher results when compared to Haar. References [1] Hemalatha S., Dinesh Acharya. U., Renuka A., (2015), Wavelet transform based steganography technique to hide audio signals in image, Journal of Applied Signal Processing, Vol.2, No. 2, pp [2] Ajaya Shrestha, Arun K. Timalsina Image Steganography Technique Using Daubechies Discrete Wavelet Transform, Journal of Advances in Adaptive Data Analysis, Vol. 2, No. 1, pp [3] Chang L.C., Lie W.N., (2006), Robust and high quality time domain audio watermarking based on low frequency amplitude modification, IEEE Transactions on Multimedia, Vol. 8, No. 1, pp [4] Cvejic N., Seppanen T., (2004), Spread spectrum audio watermarking using frequency hopping and attack characterization, IEEE Journal on Signal Processing, Vol. 84, no. 1, pp [5] Chen B., Wornell G. W., (2001), Quantization index modulation methods for digital watermarking and information embedding of multimedia, Journal of VLSI Signal Processing System, Vol. 27, No. 7, pp [6] Huang J., Huang D., Shilpa Y.Q.,(2005), Efficiently self synchronized audio watermarking for assured audio data transmission, IEEE Transactions on Broadcasting, Vol. 51, No. 1, pp [7] Huanget N. E., (1998), The empirical mode decomposition and Hilbert spectrum for nonlinear and non-stationary time series analysis, in Proceedings of the Royal Society, Vol. 454, No. 1971, pp [8] Cox I. J., Kilian J., Leighton T., and Shamoon T., (1996), A secure, robust watermark for multimedia, in Proceedings of Lecture Notes in Computer Science, Vol. 1174, No.27, pp [9] Emmanuel S., Kankanhalli M. S., Wang. L., (2010), EMD and psychoacoustic model based watermarking for audio, IEEE International Conference on Multimedia and Expo, pp [10] Huang J., Kim H. J., Xiang. S., (2008), Audio watermarking robust against time-scale modification and MP3 compression, IEEE Journal on Signal Processing, Vol.88, No.10, pp [11] Khalilullah K. M. I., Islam M., Molla K.I., Zaman K., (2010), A robust digital audio watermarking algorithm using empirical mode decomposition, in Proceedings of Canadian Conference on Electrical and Computer Engineering, pp [12] Khalilullah K. M. I., Islam M., Molla K.I., Zaman K., (2010), A robust digital audio watermarking algorithm using empirical mode decomposition, in Proceedings of Canadian Conference on Electrical and Computer Engineering, pp [13] Jayasudha.S.,(2010), Integer Wavelet Transform based on Steganographic method using OPA [14] Algorithm, International Journal of Engineering and Science,Vol.2, No.4,pp [15] VikashRamesh, Kaushik, Premalatha Pandian Narayanan.,(2014), Steganography in audio signals using variable bit Replacement Method in DCT Domain, International Journal of Engineering Research & Technology,Vol.3,No.4,pp [16] Bhat V., Sengupta K. I., Das. A., (2010), An adaptive audio watermarking based on the singular value decomposition in the wavelet domain, in Proceedings of European Signal Processing Conference, pp [17] T.RajeshKumar, G.RSuresh, Examination of Militants utilizing NAM Microphone and Wireless Handset for Murmured Speech in view of Concealed Markov Model, International Innovative Research 302

9 Journal of Engineering and Technology, vol 02, no 04, pp ,2017. [18] T. Padmapriya and V.Saminadan, Handoff Decision for Multi-user Multiclass Traffic in MIMO- LTE-A Networks, 2nd International Conference on Intelligent Computing, Communication & Convergence (ICCC-2016) Elsevier - PROCEDIA OF COMPUTER SCIENCE, vol. 92, pp: , August [19] T. Padmapriya, V.Saminadan, Performance Improvement in long term Evolution-advanced network using multiple imput multiple output technique, Journal of Advanced Research in Dynamical and Control Systems, Vol. 9, Sp-6, pp: , [20] S.V.Manikanthan and K.Baskaran Low Cost VLSI Design Implementation of Sorting Network for ACSFD in Wireless Sensor Network, CiiT International Journal of Programmable Device Circuits and Systems,Print: ISSN X & Online: ISSN , Issue : November 2011, PDCS [21] S.V.Manikanthan and D.Sugandhi Interference Alignment Techniques For Mimo Multicell Based On Relay Interference Broadcast Channel International Journal of Emerging Technology in Computer Science & Electronics (IJETCSE) ISSN: Volume- 7,Issue 1 MARCH [22] Rajesh, M., and J. M. Gnanasekar. " Congestion Control Using Aodv Protocol Scheme For Wireless Ad-Hoc Network. " Advances in Computer Science and Engineering 16.1/2 (2016): 19. [23] Rajesh, M., and J. M. Gnanasekar. " Constructing Well-Organized Wireless Sensor Networks with Low-Level Identification. " World Engineering & Applied Sciences Journal 7.1 (2016). 303

10 304