IJRIT International Journal of Research in Information Technology, Volume 2, Issue 2, February 2014, Pg: 28-37

International Journal of Research in Information Technology (IJRIT) www.ijrit.com

ISSN 2001-5569

A Robust Video Watermark Embedding and Extraction Technique Based on Random Frame Selection Amrinder Singh Electronics & Communication Engineering Guru Teg Bahadur Khalsa Institute of Engineering & Technology, Malout, Punjab, India [email protected] Sukhjit Singh Electronics & Communication Engineering Guru Teg Bahadur Khalsa Institute of Engineering & Technology, Malout, Punjab, India [email protected] Abstract Digital video watermarking is the enabling technology to prove ownership of copyrighted material, to solve the problem of piracy and to detect the originator of illegally made copies. In this paper, to solve the authentication problem an effective, imperceptible and robust non blind video watermarking algorithm is proposed which uses an encryption key to select the random frames of video in which watermark information is embedded uniformly throughout the video. To keep the algorithm imperceptible, blocks of watermark are embedded in the SVD-transformed video in a diagonal-wise fashion few blocks on the basis of higher entropy are selected and watermarked using LSB technique. The performance of algorithm is tested using MATLAB software on video of “rhinos” and watermark image of 512 X 512. The experimental results show that the proposed scheme is highly imperceptible, less time consuming, more secure and highly robust against frame dropping & other manipulations.

General Terms: Digital Watermarking, Security, Imperceptibility, Robustness Keywords: Video Watermarking, Entropy, PSNR, MSE, BER, SSIM 1. INTRODUCTION In the past duplicating art work was quite complicated and required a high level of expertise for the counterfeit to look like the original. However, in the digital world it is possible for almost anyone to alter digital data without and losing data quality. To overcome this issue, watermarking technology is used. Video watermarking is the process of embedding copyright information or verification messages in video bit streams. Video watermarking research received less attention than image watermarking due to it’s inherit difficulty, however, many algorithms have already been proposed .The information which is embedded is called watermark. It can be text or an image. Two types of digital watermarks may be distinguished, depending upon whether the watermark appears visible or invisible to the casual viewer. Visible watermarks can be a logo or text on frames of videos either in all frames or in just a few selected frames. If it is present in selected frames then it passes off without being noticed, due to high frame rate. Invisible watermarks or Hidden watermarks on other hand are present in the file in such a way that they cannot be sighted but have to be extracted. Watermarking algorithm should not affect the quality of video. It should also be robust to various signal processing operations i.e. watermark could not be damaged or degraded after any type of video manipulations. Watermarking algorithm can be blind or non blind. If the extraction process needed the original data for the recovery of watermark from watermarked video then it is said to be non blind scheme of watermarking. If watermark can be recovered from only watermarked video without any need of original data then it is called blind scheme of watermarking. This scheme applied to videos shows that it consumes very small time to embed the watermark information and it is highly imperceptible, exhibits high robustness against frame dropping & more secure scheme due to use of encryption key and random frame selection. Amrinder Singh, IJRIT

28

IJRIT International Journal of Research in Information Technology, Volume 2, Issue 2, February 2014, Pg: 28-37

2. PROBLEM STATEMENT As digital video-based application technologies grow, such as Internet video, wireless video, Video phones, and video conferencing, the problem of illegal manipulation, copying, distribution and piracy of digital video rises more and more. The problem of this paper research work is to solve the authentication problem and embed the watermark in such a way that it could not be removed or damaged from the video using the proposed algorithm of random frame selection through encryption key. The watermark is embedded in these selected frames. Encryption key used is decided by the owner of the video. And the random frames are selected by using the functions generated through this authentication key. These functions are designed such that avoiding the selection/clustering of frames in one chunk. Instead of the clustering of frames, the frames are selected uniformly from whole video. Then same watermark information is used to embed in all the selected frames to increase the probability of maintaining the watermark in manipulated watermarked video. For example if some unauthorized person tries to drop some frames of the video, then if some watermarked frames dropped from the video, and if only one watermarked frame is left behind in the video then the watermark information can be recovered from this frame only. The manipulations can be done with video either through frame dropping or through any other way by any unauthorized person for illegal copying the video. To preserve the quality of the video & keep the algorithm more imperceptible the entire video frame is not altered by embedding the watermark information. Instead of that the frame is divided into blocks of 8 X 8 and applies singular values decomposition technique on these blocks. And watermark information is embedded in these singular values.. As a result of using several watermarked blocks, several watermarks can be recovered. So if any attack affects the watermarked image, some of the watermarks will survive. This block-byblock method gives robustness against JPEG compression, cropping, blurring, Gaussian noise, resizing and rotation as the results will reveal. The watermark can either be a pseudo-random number, or an image. In this paper an image is used as watermark. After watermarking the frames, we insert them back in the video at their respective places to get the watermarked video. To extract the watermark from watermarked frames again same encryption key is required to find the watermarked frames. We set up a key identifier to give only three trials to the user. If the use tries extraction with more than 3 wrong keys then it is assumed that he is trying to find the watermarked frames by trying random keys. So at fourth try with wrong key the video is corrupted leaving no data behind.

3.

SVD Technique

The SVD mathematical technique provides an elegant way for extracting algebraic features from an image. The main properties of the SVD matrix of an image can be exploited in image watermarking. The SVD matrix of an image has good stability. When a small perturbation is added to an image, large variation of its SVs does not occur. Using this property of the SVD matrix of an image, the watermark can be embedded to this matrix without large variation in the obtained image

4. PROPOSED ALGORITHM In our proposed algorithm, Singular value decomposition technique is used for embedding watermark. The input video sequence is divided into its constituted frames. Then 10 random frames are selected through the functions generated using encryption key entered by owner of the video. Then the frame to be watermarked is divided into blocks of 8 X 8, and then rescales the watermark as video frame size and divides it into 8 X 8 blocks. and then watermark is embedded to the diagonal matrix (S matrix) of each block giving new matrices. An SVD is performed on each of these new matrices to get the SV matrices of the watermarked image blocks. Then, these SV matrices are used to build the watermarked image blocks. By combining these blocks again into one matrix of the original image dimensions, the watermarked image Fw is built in the spatial domain. The steps of embedding the watermark can be summarized as follows:

4.1 Watermark Embedding Algorithm Step 1: Extract all the frames N from input video file. Step 2: Enter 10 digit encryption key for random frame selection where each digit of key is 8bit. Step 3: Calculate an offset value using total number of frames in video for uniform selection of frames. off=N/10 Step 4: Using the ASCII values of 10 digits of the key entered in step 2 & offset value calculated in step 3 to generate 10 random functions to select 10 random frames from the video for watermarking. If the digits of key are, b, c, d, e, f, g, h, i, j then the 10 functions will be X1= (off*0) + (a+b) X1= (off*1) + (b+c) X1= (off*2) + (c+d) Amrinder Singh, IJRIT

29

IJRIT International Journal of Research in Information Technology, Volume 2, Issue 2, February 2014, Pg: 28-37

X1= (off*3) + (d+e) X1= (off*4) + (e+f) X1= (off*5) + (f+g) X1= (off*6) + (g+h) X1= (off*7) + (h+i) X1= (off*8) + (i+j) X1= (off*9) + (j+a) If addition of ASCII values of three digits is greater than the offset value, then offset value is subtracted from their sum to get a number which is less than offset value. These ten values of X1 to X10 represent the frame numbers. Frames with these frame numbers are selected for watermarking. Step 5: Select the blue component from the selected RGB frame in which the watermark is to be embedded. Step 6: Divide the blue component into non overlapping blocks Bij of size 8X8. Step 7: Perform SVD on each block of video frame (Bi matrix) to obtain the SVs (Si matrix) of each block. Where i =1, 2, 3,….., N, and N is number of blocks. Bi =UiSiVT Step 8: Rescale the watermark image as the same size of video frame and divide the watermark into the blocks of size 8X8 Step 9: Add the watermark image (W matrix) to the S matrix of each block. Di=Si+kW Step 10: Perform SVD on each Di matrix to obtain the SVs of each (Swi matrix). Di=UwiSwiV T Step 11: Use the (Swi matrix) of each block to build the watermarked blocks in the spatial domain. Bwi=UiSwiVT Step 12: Rearrange the watermarked blocks back into one matrix to build the watermarked video blue component frame in the spatial domain (Fw matrix). Step 13: Integrate this modified blue component with red and green components to get the watermarked RGB Frame. Step 13: Repeat step 5 to step 13 for all the selected frames for watermarking to get the watermarked frames. Step 14: Generate the checksum bits from the encryption key used in step 2 and store the checksum bits into the random pixels of the red component of frame 1. Set the first pixel value to zero in the red component of frame 2. Step 15: Develop the watermarked video using the modified frames by placing them to their respective position.

4.2 Watermark Extraction Algorithm Step 1: Extract all the frames N from watermarked video file. Step 2: Ask the user to enter the encryption key. Step 3: Generate the checksum from the key entered by the user in step 2. Step 4: Extract the checksum of original key stored in the red component of frame 1. Step 5: Compare both the checksums from step 3 & step 4. And increment the first pixel value in the red component of frame 2 every time checksum goes wrong. Step 6: When this pixel value reaches four then corrupt the video file by writing zero to all pixel values of video. And stop the extraction process. Step 7: If checksum matches then use the key entered in step 2 for finding the watermarked frames in the video. Follow step 4 of embedding process to find the watermarked frames. Step 8: Select the blue component of watermarked frame from which the watermark is to be extracted. Step 9: Divide the watermarked blue component into blocks having the same size used in the embedding process. Step 10: Performs SVD on each watermarked block (B*wi matrix) to obtain the SVs of each one (S*wi matrix). B*wi = Ui*S*wiVi*T Step 11: Obtains the matrices that contain the watermark using Uwi, Vwi, S*wi, matrices. D*I = UwiS*wi T Extract the possibly corrupted watermark (W* matrix) from the Di matrices. (D*i-Si)/k=W*i Step 12: Rearrange all the extracted watermark blocks to get the extracted watermark image Step 13: Rescale the extracted watermark image to the size of original watermark image

5. EXPERIMENTAL RESULTS AND PERFORMANCE EVALUATION

Amrinder Singh, IJRIT

30

IJRIT International Journal of Research in Information Technology, Volume 2, Issue 2, February 2014, Pg: 28-37

MATLAB 7.10.0 is used as the platform for implementing the proposed work & conducting experiments. The performance of the proposed video watermarking algorithm is evaluated using many colored videos containing different number of frames at various frame rates. But here results are discussed for a 7.6 seconds video clip of “rhinos” at a frame rate of 15fps constituting of 114 frames. The watermark used in our experiments was a binary image of 512 X 512. Encryption key used is “9876045095” based on which random frames are selected. A video frame, watermark image & corresponding watermarked frame is shown in figure 1.

Figure 1: Original Video Frame, Watermark Image & Watermarked Frame 5.1 Imperceptibility performance: To prove the proposed algorithm imperceptible, as a measure of quality of the watermarked video Mean Squared Error (MSE), Peak Signal to Noise Ratio (PSNR), Bit Error Rate (BER) and Structural Similarity Index Metric (SSIM) is calculated for all the watermarked frames. The values for these parameters for all the frames & their average values are tabulated in table 1. Figure 2, 3, 4, 5 shows the values of MSE, PSNR, BER and SSIM respectively for all the watermarked frames. Figure 6 shows the average values of these parameters. Higher average value of PSNR (67.77 dB), smaller values of MSE (0.0109) & BER (0.015) and value of SSIM (0.9997) closer to 1 shows the imperceptibility of proposed algorithm. 5.2 Security The proposed algorithm is more secure than the conventional algorithms due to the use of an encryption key for the selection of the frames to be watermarked. And at time of extraction process same encryption key is needed and if key is wrong then nobody can find the watermarked frames. And if someone tries for extraction with wrong key then he will be given only three chances of extraction, after that watermarked video will be damaged due to illegal processing and video will be of no use for that person. 5.3 Robustness Performance Similarity between the original watermark and the extracted watermarks after attacks from all the watermarked video frames is measured by computing correlation factor ρ. Random watermarked frame numbers are listed in table 1 & the extracted watermarks from respective frames are shown in figure 7. Original watermark & their corresponding correlation factors are also shown in figure 7. The proposed algorithm is robust to various attacks like Noise attacks, and blur attack than the conventional methods. The values of correlation factor between the extracted watermark and original watermark after these various attacks are tabulated in table 2 and attacked watermarked frames and extracted watermarks after attacks are shown in figures 9, 10, 11, 12. The proposed method is also robust to frame dropping attack because to destroy the watermark from watermarked frames, the watermark frames should be known. And the watermarked frames cannot be found out easily due to random & uniform frame selection for watermarking using the encryption key. Watermark is not embedded in the frames of one chunk but it is spread uniformly throughout the video to avoid the clustering of watermarked frames in one chunk. Also in proposed algorithm same watermark image is embedded in all the frames due to which if watermark is destroyed in some watermarked frames by any manipulation or some watermarked frames are dropped then it can be recovered from the others and probability of maintaining the watermark in manipulated watermarked video increases.

Amrinder Singh, IJRIT

31

IJRIT International Journal of Research in Information Technology, Volume 2, Issue 2, February 2014, Pg: 28-37

Figure 2: MSE Values for all the watermarked frames

Figure 3: PSNR Values for all the watermarked frames

Amrinder Singh, IJRIT

32

IJRIT International Journal of Research in Information Technology, Volume 2, Issue 2, February 2014, Pg: 28-37

Figure 4: BER Values for all the watermarked frames

Amrinder Singh, IJRIT

33

IJRIT International Journal of Research in Information Technology, Volume 2, Issue 2, February 2014, Pg: 28-37

Figure 5: SSIM Values for all the watermarked frames

Figure 6: Average Values of PSNR, MSE, BER & SSIM of All the Frames Table 1: Values of MSE, PSNR, BER, and SSIM for all the Frames & their average Watermar ked Frame Random Frame Number MSE PSNR

Frame 1

Frame 2

Frame 3

Frame 4

Frame 5

Frame 6

Frame 7

Frame 8

Frame 9

Frame 10

Avera ge Value

3

12

32

36

45

61

74

80

90

110

NA

0.0122 19 67.260 3

0.0119 63 67.352 3

0.0104 71 67.930 8

0.0107 25 67.826 7

0.0101 75 68.055 4

0.0115 81 67.493 2

0.0098 025 68.217 4

0.0094 719 68.366 4

0.0109 99 67.717 4

0.0115 12 67.519 3

0.0108 92 67.773 9

Amrinder Singh, IJRIT

34

IJRIT International Journal of Research in Information Technology, Volume 2, Issue 2, February 2014, Pg: 28-37

BER

0.0148 68

0.0148 47

0.0147 21

0.0147 43

0.0146 94

0.0148 16

0.0146 59

0.0146 27

0.0147 67

0.0148 11

0.0147 55

SSIM

0.9997 2

0.9997 3

0.9998 2

0.9998 2

0.9998 5

0.9997 5

0.9997 8

0.9997 5

0.9997 2

0.9997 6

0.9997 7

Figure 7: original watermark, extracted watermarks from all the 10 watermarked frames with frame number & their correlation factors

Figure 8: (a) Original Image (b) Watermark (c) Watermarked Image (d) Extracted watermark given correlation coefficient= 0.99023

Amrinder Singh, IJRIT

35

IJRIT International Journal of Research in Information Technology, Volume 2, Issue 2, February 2014, Pg: 28-37

Figure 9: (a) Original Image (b) Watermark (c) Gaussian noise attacked frame (d) Extracted watermark given correlation coefficient = 0.54851

Figure 10: (a) original image (b) watermark (c) Poisson noise attacked frame (d) Extracted watermark given correlation coefficient = 0.55214

Figure 11: (a) original image (b) watermark (c) Blur attack (d) Extracted watermark given correlation coefficient = 0.851

Figure 12: (a) original image (b) watermark (c) Rotation attack (d) Extracted watermark given correlation coefficient = 0.30355

Amrinder Singh, IJRIT

36

IJRIT International Journal of Research in Information Technology, Volume 2, Issue 2, February 2014, Pg: 28-37

6. CONCLUSIONS In this paper, a blind video watermarking algorithm is proposed in which random frames from the whole video frames are selected for watermarking using an encryption key. To preserve the quality of the video, a particular selected frame is divided into blocks and the blocks of high entropies are selected for watermarking. Then watermark information is embedded at LSB of each pixel of the selected block. The algorithm is evaluated in terms of imperceptibility, security, time consumption, data payload and robustness. To measure the imperceptibility of algorithm PSNR, MSE, BER & SSIM are computed. The calculated values of these parameters show the high imperceptibility of the algorithm. Also the algorithm is simple blind algorithm, less time consuming, more secure and highly robust against frame dropping & other manipulations.

7. REFERENCES [1]

L. Qiao and K. Nahrstedt, "Watermarking Schemes and Protocols For Protecting Rightful Ownership and Customer's Rights", Journal of Visual Commun. and Image Represent 9, pp.194– 210, 1998.

[2]

M. Arnold, M. Schumucker, and S. Wolthusen, “Techniques and Applications of Digital Watermarking and Content Protection”. Artech House, 2003.

[3]

Lama Rajab, Tahani Al-Khatib, Ali Al-Haj, “Video Watermarking Algorithms Using the SVD Transform” European Journal of Scientific Research, Vol.30 No.3, pp.389-401, 2009.

[4]

Manekandan. GRS, Franklin Rajkumar. V, “A Robust Watermarking Scheme for Digital Video Sequence using Entropy and Hadamard Transformation Technique”, International Journal of Computer Applications, Volume 41– No.18, pp.24-31, March 2012.

[5]

Angshumi Sarma, Amrita Ganguly, “An Entropy based Video Watermarking Scheme”, International Journal of Computer Applications, Volume 50 – No.7, pp.24-31, July 2012.

[6]

Jigar Madia, Kapil Dave, Vivek Sampat, Parag Toprani, “Video Watermarking using Dynamic Frame Selection Technique”, National Conference on Advancement of Technologies – Information Systems & Computer Networks (ISCON – 2012), pp.31-34, 2012.

[7]

Jassim Mohmmed Ahmed, Zulkarnain Md Ali, “Information Hiding using LSB technique”, International Journal of Computer Science and Network Security, VOL.11 No.4, pp.18-25, April 2011.

[8]

C.Sasi varnan, A.Jagan, Jaspreet Kaur, Divya Jyoti, Dr.D.S.Rao, “Image Quality Assessment Techniques on Spatial Domain”, IJCST Vol. 2, Issue 3, pp. 177-184, September 2011

[9]

R. A. Ghazy, N. A. El-Fishawy, M. M. Hadhoud, M. I. Dessouky and F. E. Abd El-Samie, “An Efficient Block -byBlock SVD-Based Image Watermarking Scheme.

Amrinder Singh, IJRIT

37

A Robust Video Watermark Embedding and Extraction ... - IJRIT

more secure and highly robust against frame dropping & other manipulations. ... watermarks can be a logo or text on frames of videos either in all frames or in just a few ... If the extraction process needed the original data for the recovery of.

362KB Sizes 1 Downloads 291 Views

Recommend Documents

A Robust Video Watermark Embedding and Extraction ... - IJRIT
The performance of algorithm is tested using MATLAB software on video of “rhinos” and watermark image of ... divided into blocks of 8 X 8 and applies singular values decomposition technique on these blocks. ... In our proposed algorithm, Singular

Secure Watermark Embedding through Partial Encryption
content distribution applications, the watermark embedder at the server side .... drawback that it requires a dedicated hardware installed base, cannot be eas-.

A Robust And Invisible Non-Blind Watermark for Network Flows
security monitoring. Generally, stepping stones are detected by noticing that an outgoing flow from an enterprise matches an in- coming flow. For example, in Figure ..... IPD database. Network. Figure 2. Model of RAINBOW network flow watermarking sys

Aoao Photo Watermark + Aoao Video Watermark Pro ...
Quite a few website visitors will find this page while checking any one of the major ... This is a video tutorial of how to download online videos with WonderFox DVD Video Converter quickly. ... Mobile apps; Multimedia & Design Software;.

Secure Watermark Embedding through Partial Encryption
age rights beyond the digital domain. In a forensic tracking system, each copy of the distributed content is watermarked with a unique transaction tag, which links that copy either to a particular user or to a specific device. When an unau- thorized

Robust Information Extraction with Perceptrons
First, we define a new large-margin. Perceptron algorithm tailored for class- unbalanced data which dynamically ad- justs its margins, according to the gener-.

Text Extraction Using Efficient Prototype - IJRIT
Dec 12, 2013 - as market analysis and business management, can benefit by the use of the information ... model to effectively use and update the discovered Models and apply it ..... Formally, for all positive documents di ϵ D +, we first deploy its

Text Extraction Using Efficient Prototype - IJRIT
Dec 12, 2013 - Various data mining techniques have been proposed for mining useful Models ... algorithms to find particular Models within a reasonable and ...

Robust Information Extraction with Perceptrons
... by “building” is the mention of an entity of type FACILITY and sub- ..... We call this algo- rithm the ..... 24. 90. 116. 5.6. 38.5. 2.4. 53.5. 88.0. 59.1. 70.7. PHYS. 428. 76. 298. 113. 8.7. 69.1. 6.2 .... C-SVC SVM type) takes over 15 hours

A Review: Study of Iris Recognition Using Feature Extraction ... - IJRIT
analyses the Iris recognition method segmentation, normalization, feature extraction ... Keyword: Iris recognition, Feature extraction, Gabor filter, Edge detection ...

A Review: Study of Iris Recognition Using Feature Extraction ... - IJRIT
INTRODUCTION. Biometric ... iris template in database. There is .... The experiments have been implemented using human eye image from CASAI database.

A Robust Solution for Collaborative Data Mining - IJRIT
His research interests include networking and cloud computing. ... He is Microsoft Certified System Engineer & CISCO Certified Network Administrator, ...

Robust Text Detection and Extraction in Natural Scene ...
Lecturer, Computer Engineering, Pune University/ VACOE Ahmednagar ... of the two classes (“'text”&”'Non Text”)by Considering both unary component ...

A Robust Solution for Collaborative Data Mining - IJRIT
... is the approach in which multiple data providers share their data for data mining tasks .... He is Microsoft Certified System Engineer & CISCO Certified Network ...

Wireless video transmission Evaluation using Lossless video ... - IJRIT
support streaming from live video sources, such as cable/satellite ..... Solomon code is widely used from optical media storages, such as DVD and CD, .... S. Priyadarshi, and J. Villasenor "H. 264 Parameter Optimizations for Internet Based.

Wireless video transmission Evaluation using Lossless video ... - IJRIT
open or closed-loop translating or even ... several parallel data stream . ..... the condition of transmission without degradation and error free sequence is possible ...

Perceptual Similarity based Robust Low-Complexity Video ...
block means and therefore has extremely low complexity in both the ..... [10] A. Sarkar et al., “Efficient and robust detection of duplicate videos in a.

Video Conference System Using Raspberry Pi - IJRIT
the early 1970s as part of AT&T's development of Picture phone technology. With the ... (such as phone and Internet), by reducing the need to travel, which is often carried out by aero plane, to .... There is no arrangement for audio input in Raspber

Perceptual Similarity based Robust Low-Complexity Video ...
measure which can be efficiently computed in a video fingerprinting technique, and is ... where the two terms correspond to a mean factor and a variance fac- tor.

low-delay robust video multicast
unicast server bitrate RS for (1) PAR-CT and (2) PAR-DT at different redundancy ..... the receivers by leveraging the presence of a dedicated server. In SRM, the ...

Adaptive Extraction of Information Using Relaxation Labelling ... - IJRIT
... web pages and are powered by a variety of forum software packages, but they always ... Full automatic extraction of structured data from web forums of conversations. .... Nigam K, Siegler M, Stockton R, and Tomokiyo T, “Deriving Marketing.

Robust Feature Extraction via Information Theoretic ...
function related to the Renyi's entropy of the data fea- tures and the Renyi's .... ties, e.g., highest fixed design breakdown point (Miz- era & Muller, 1999). Problem ...

Robust Feature Extraction via Information Theoretic ...
Jun 17, 2009 - Training Data. ▫ Goal:Search for a ... Label outliers: mislabeling of training data. ▫ Robust feature ... Performance comparison on MNIST set ...

Adaptive Extraction of Information Using Relaxation Labelling ... - IJRIT
Abstract-Internet forums are important services where users can request and exchange information ... 2.2 Web data extraction based on partial tree alignment:.