A Robust Blind Video Watermarking Scheme based on Discrete

Wavelet Transform and Singular Value Decomposition

Amal Hammami

1 a

, Amal Ben Hamida

1 b

and Chokri Ben Amar

1, 2 c

REsearch Groups in Intelligent Machines, University of Sfax, National Engineering School of Sfax, Sfax, 3038, Tunisia

Department of Computer Engineering, College of Computers and Information Technology, Taif University, Saudi Arabia

Keywords: Blind Video Watermarking, Discrete Wavelet Transform, Singular Value Decomposition, Robustness.

Abstract: The outgrowth in technological world has massively promoted to information fraud and misappropriation by

the ease of multimedia data content regeneration and modification. Consequently, security of digital media is

considered among the biggest issues in multimedia services. Watermarking, consisting in hiding a signature

known as watermark in a host signal, is one of the potential solutions used purposely for media security and

authentication. In this paper, we propose a robust video watermarking scheme using Discrete Wavelet

Transform and Singular Value Decomposition. We embed the watermark into the mid frequency sub-bands

based on an additive method. The extraction process operates following a blind detection algorithm. Several

attacks are applied and different performance metrics are computed to assess the robustness and the

imperceptibility of the proposed watermarking. The results reveal that the proposed scheme is robust against

different attacks and achieves a good level of imperceptibility.

1 INTRODUCTION

In day-to-day life, the tremendous improvement in

computer technology field is leading to many

problems for the multimedia industry. Indeed, many

data files, such as video files, can be easily copied,

distributed and tampered without compromising the

quality of the data. Thus, a broad range of approaches

has already been proposed to secure video sequences

content. Cryptographic techniques based on secret

key are used to encrypt data in order to secure the

visual content (Omar and Shawkat, 2018; Vikrant and

Shubhanand, 2016). Nevertheless, it is pointed out

that these techniques come together in hands with

some ambiguities. In fact, their main drawbacks

derive from the non-preservation of the videos

original formats. To tackle this problem, video

watermarking techniques provide a promising

security solution. A video watermarking scheme

includes predominantly two steps: the embedding

process which is the process of incorporating an

imperceptible data (a binary image, a bits sequences,

etc.), known as watermark, into cover video and the

https://orcid.org/0000-0002-7728-6620

https://orcid.org/0000-0002-3164-5456

https://orcid.org/0000-0002-0129-7577

detection process which is the process of extracting

the inserted data from the watermarked video

(Asikuzzaman, and Pickering, 2018). Digital

watermarking system has been a very interesting

research area in many applications such as copyright

protection, data authentication, etc (Tuan and Duan,

2015; Charfeddine et al, 2014). Hence, each

watermarking system should have its intrinsic

properties regarding the given application (Ben

Hamida et al., 2011; Zhang et al., 2012; Ahuja and

Bedi, 2015; Tarhouni et al., 2018; Jyothika and

Geetharanjin, 2018; Koubaa et al, 2012). Generally,

there are three important requirements considered in

the most practical video watermarking systems (Asim

et al., 2015; Arti and Ajay, 2017). The first one is the

imperceptibility. It refers to the visual quality of the

watermarked video, which should perceptually be as

close to the original video as possible. The second

requirement is the robustness. It denotes the ability of

watermark to sustain unintentional and intentional

attacks. Capacity, which is defined as the amount of

the secret information concealed in the host video, is

the third required property. Imperceptibility,

Hammami, A., Ben Hamida, A. and Ben Amar, C.

A Robust Blind Video Watermarking Scheme based on Discrete Wavelet Transform and Singular Value Decomposition.

DOI: 10.5220/0007685305970604

In Proceedings of the 14th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2019), pages 597-604

ISBN: 978-989-758-354-4

597

robustness and capacity are inversely related. Hence,

it is very important to maintain a good trade-off

between all the watermarking properties (Hedayath et

al., 2016). The watermarking techniques are

commonly divided into two main classes according to

the embedding domain criterion; the spatial domain

techniques and the frequency domain ones (Arti and

Ajay, 2017; Amit and Navdeep, 2016; Asim et al.,

2015). In the first class, the watermark is inserted by

modifying the pixel values of the host video directly.

For the second class, the host video frames undergone

a domain transformation technique and the

watermark is inserted into selective coefficients from

these frames afterwards. The spatial domain-based

techniques present low computational complexity.

Nevertheless, they are not reliable in the presence of

different image processing methods and have low-bit

capacity. However, the watermarking techniques in

the frequency domain are comparatively more

resistant to common distortions including noise

addition, lossy compression and rotation. Besides,

they can effectively achieve the compromise between

imperceptibility and robustness requirements of

digital watermarking techniques. With much interest

to this, we propose in this work a blind and robust

video watermarking scheme in the frequency domain

based on a cascade of two transformations; Discrete

Wavelet Transform (DWT) and Singular Value

Decomposition (SVD).

The organization of the remainder of this paper is

as follows. Section 2 provides a review on existing

frequency domain video watermarking techniques.

The proposed watermarking scheme is described in

Section 3. Experimental results as well as a

comparison with others techniques are given in

Section 4. The last section summarises and concludes

this work.

2 RELATED WORK

In the literature, a variety of video watermarking

techniques has been proposed and can be categorized

into different classes based on different criteria such

as working domain, watermark visibility, watermark

robustness, etc. As we previously mentioned, in

frequency domain watermarking the embedding

process is preceded by the application of some

transformation methods to the cover video frames.

Fast Fourier Transform (FFT), Discrete Cosine

Transform (DCT), Discrete Wavelet Transform

(DWT) as well as Non-subsampled Contourlet

Transform (NSCT) and Singular Value

decomposition (SVD) are the most used methods in

this domain (Wali et al, 2010; Guedri et al, 2011;

Othmani et al, 2010). In this section, we will focus on

digital video watermarking in the transform domain.

Some video watermarking schemes can operate

using one single transformation technique. Indeed, a

DWT based watermarking algorithm has been

presented in (Mostafa and Ali, 2016). This video

watermarking scheme satisfies the imperceptibility

requirement exploiting the human visual system

characteristics. In fact, blocks with highest motion

vector magnitude are selected to hide the watermark

after being transformed using the discrete wavelet

transform. Another watermarking technique using the

discrete wavelet transform has been proposed in

(Dolley and Manisha, 2018). In this method, the

watermark is embedded only on scene-changed

frames chosen using a scene-change detector

algorithm. Therefore, those frames are converted to

the grey-scale and decomposed using the three level

DWT. Then the watermark bits are inserted in the low

frequency components using an additive method. The

authors in (Tuan and Duan, 2015) proposed a blind

approach based on discrete cosine transform. The

luminance component Y extracted from the host

video frame is decomposed into DCT 8x8 blocks. The

watermarks bits are subsequently hidden on several

randomly chosen DCT coefficients using even-odd

quantization algorithm.

On the other hand, video watermarking schemes

can also perform based on a combination of domain

transformation techniques. An example of hybrid

video watermarking technique based on non-

subsampled contourlet transform and singular value

decomposition is introduced in (Narasimhulu, 2017).

In this approach, both of the original video frame and

the watermark are decomposed by the non-

subsampled contourlet transform. Then, the singular

values of watermark are incorporated in those of

original video using an additive non-blind algorithm.

In (Naved, 2016), a robust video watermarking

technique using four distinct domain transformation

techniques is proposed based on DWT, SVD, FFT

and DCT. The watermark embedding process is done

using an additive method. A third example of

watermarking schemes involving two transformation

techniques is introduced in (Jeebananda and Prince,

2016). Here, authors proposed a video watermarking

in four approaches all based on discrete wavelet

transform and singular value decomposition. The four

methods differ from each other by the number of

frames to be used for the embedding process. In the

third approach, authors used a scene change detector

in order to select the embedding medium. In fact, the

insertion occurs in the singular value corresponding

VISAPP 2019 - 14th International Conference on Computer Vision Theory and Applications

598

to the low frequency sub band extracted from only

detected scene change frames.

3 PROPOSED APPROACH

The proposed system is a frequency domain video

watermarking using both discrete wavelet transform

and singular value decomposition. Using these two

domain transformation techniques allows improving

the watermark robustness by exploiting their

complementary properties. The proposed system

consists of two stages: embedding process and

extraction process. The block diagrams of these

processes are shown in figure 1 and figure 2

respectively and explained separately in the two

following sub-sections.

3.1 Embedding Process

As presented in figure 1, the video is split into

sequences of N consecutive frames. The number N

that defines the video sequence size is experimentally

determined. It is used as a secret key to improve the

security level of the watermarking technique. Within

each video sequence, the same watermark is inserted

into all frames in order to enhance the robustness

against frame dropping attack. A different binary bits

message is used as a watermark for every video

sequence. All frames in each sequences are subject

for the following process.

Since pixel values are highly correlated in RGB

color space, the RGB frame is converted into YUV

space. Taking into account that Human Visual

System cannot notice the changes in regions of high

luminance, only the luminance component Y is

selected for the watermark incorporation. This

component is subdivided in 4*4 non-overlapping

blocks. Then, one level Discrete Wavelet Transform

is applied to transform each block to the frequency

domain. This operation generates four sub-bands:

LL1, HL1, LH1 and HH1. The high frequency

component HH1 contains the least significant parts of

the video frame, a potential information loss can

occur during compression. The low frequency sub-

band LL1 contains the most significant information

of the video frame; its modification degrades the

visual quality. So, the mid frequency components

LH1 and HL1 are selected as the best locations for

watermarking to meet the trade-off between the

robustness and the imperceptibility requirements.

The chosen sub-bands are decomposed using

the singular value decomposition into three

independent matrices U, S and V. Only S matrix is

implied into the embedding process. Indeed, a slight

modification in the singular values does not yield a

large visual alteration in the host video. Besides, they

exhibit attractive properties such as rotation

invariance, transposition invariance and translation

invariance.

Using blocks of size (4x4) to embed the

watermark bits enables to obtain S matrix containing

few non-zero values. Hence, the order relationship

between the different singular values can be

rigorously respected.

Next, the watermark is inserted in the S matrices

corresponding to the two sub-band HL1 and LH1.

The embedding process operates according to the

following formulas:

If W = 0

(0,0)S

= S(0,0) +

(1)

(1,1)S

= S(0,0)

(2)

Else

(0,0)S

= S(1,1) +

(3)

(1,1)S

= S(1,1)

(4)

With W is the watermark bit to be inserted,

the modified version of the original matrix S,

and

are two factors allowing balancing

imperceptibility and robustness. These values are

computing using the bellow equations (5) and (6).

S(1,1)S(0,0)





(5)

S(1,1)S(0,0)





(6)

According to (5) and (6),

and



are

proportional to the S coefficients, which allows

avoiding the perceptual distortion and increasing the

watermark robustness. The values of α and β

employed in our work will be explained in the sub

section 4.1.

In order to obtain the watermarked luminance

component Y, the inverse SVD operator is applied on

the modified matrix

followed by the inverse

DWT. This watermarked Y is merged with the

unchangeable U and V components and converted to

RBG color to obtain the final watermarked video

frame. The above-described process is repeated to all

frames in each sequence to finally construct the

watermarked video.

A Robust Blind Video Watermarking Scheme based on Discrete Wavelet Transform and Singular Value Decomposition

599

Figure 1: Block diagram of the proposed watermarking embedding process.

3.2 Extraction Process

The extraction process as depicted in figure 2 is the

reverse of the embedding process. The detection

algorithm is blind. Therefore, the scheme does not

require the original non-watermarked video during

the extraction of concealed watermark.

The watermarked video is partitioned into

sequences of N frames. Then all frames of each

sequence are processed as described below to extract

the correspondent watermark. At first, the considered

video frame is converted from the RGB to YUV color

space. The luminance component Y is extracted and

decomposed into 4x4 blocs. Next, the discrete

wavelet decomposition is performed on each block,

obtaining respectively LL1, LH1, HL1 and HH1.

After applying the singular value decomposition to

both LH1 and HL1 sub-bands, watermark bits are

extracted from

ext

matrices based on the following

rules:

(0,0)S

ext

(1,1)S

ext

βα



ext

= 0

(7)

Else

ext

= 1

(8)

With

ext

is the extracted matrix S,

ext

is the

extracted watermark bit,

and

are computed

using (5) and (6) equations mentioned in the sub-

section 3.1

Figure 2: Block diagram of the proposed watermarking

extraction process.

4 EXPRIMENTAL RESULTS

To verify the robustness and the imperceptibility of

the proposed video watermarking algorithm,

simulation experiments are conducted on different

standard videos. The considered input videos are

presented in table 1.

The watermarks used in this work are binary

sequences. For every frame in each video sequence,

the maximum capacity is equal to the total number of

4x4 blocks resulting from the decomposition applied

VISAPP 2019 - 14th International Conference on Computer Vision Theory and Applications

600

on the corresponding luminance component Y. For a

256×256 frame, the maximum possible size of the

inserted message is 4096 bits, which indicates the

large capacity of proposed watermarking technique.

The number N that defines the size of each video

sequence, obtained after splitting the host video, is

fixed after evaluating the robustness of the

watermarking technique by varying the N value. It

has been notified experimentally that for each video

the adequate choice for this parameter is the frame per

second (FPS) value.

Table 1: Details of the used videos for simulation.

4.1 Imperceptibility Results

The imperceptibility requirement is examined using

the Peak Signal to Noise Ratio (PSNR), and the

Structural Similarity index (SSIM) (Asim et al., 2015;

Kadu et al., 2016). The Peak Signal to Noise Ratio is

used to find out the quality of the watermarked video

with respect to human view. This measure is defined

as:

PSNR =















MSE

log10

(9)

Where L is the maximum pixel value in the

corresponding frame and MSE is the mean square

error.

The SSIM measures the similarity between two

images. It is calculated using the following formula:

SSIM =

d)σc)(σμ(μ

d)c)(2σμ

(2μ

xyy





(10)

Where µ

and σ

are respectively the average and

the variance of the intensities available in the original

frame, µ

and σ

are respectively the average and the

variance of the intensities available in the original

frame and σ

is the covariance of original and

watermarked frames, c and d are two variables used

to stabilize the division.

For each tested video, PSNR and SSIM are

determined by computing the average of PSNR and

SSIM values of all video frames.

To deduce the suitable values for (α;β) used in

equations (5) and (6), we measure the visual

similarity between the original video and the

watermarked one. For this purpose, we calculate the

Peak Signal to Noise Ratio (PSNR) by varying (α;β)

values. The graph shown in figure 3 reveals that the

couple (2;4) achieves the highest values of PSNR for

all used videos. Therefore, α=2 and β=4 are the

approved values for the performances evaluation.

Figure 3: PSNR values obtained for different values of

(α;β).

Figure 4 displays the first frames of two host

videos and their corresponding watermarked

versions. It can be seen that the watermark is

completely transparent and there is no visual

distinction between frames of non-watermarked and

watermarked videos. The obtained PSNR values,

which are illustrated in figure 5, vary between

33.9109 dB and 48.6773 dB. It indicates that the

watermarked videos have a good visual quality.

Likewise, the SSIM values are approaching towards

1, as illustrated in figure 6, which proves the high

similarity between the host videos and the

watermarked ones. Hiding the watermark into the S

component of SVD applied to the mid frequency sub

bands of DWT presents a reasonable choice that

enables to obtain this good level of imperceptibility.

4.2 Robustness Results

The robustness requirement is scrutinized computing

the normalized correlation (NC) (Asim et al., 2015;

Kadu et al., 2016).

A Robust Blind Video Watermarking Scheme based on Discrete Wavelet Transform and Singular Value Decomposition

601

Figure 4: (a) original frame of the video stefan.avi (b)

watermarked frame of the video stefan.avi (c) original

frame of the video foreman.avi (d) watermarked frame of

the video foreman.avi.

Figure 5: PSNR values of watermarked videos.

Figure 6: Obtained SSIM values.

The NC assesses the similarity between the

original watermark and the watermark extracted from

the attacked frame. It formula is provided below:

NC =

  

 

  

 

j)(i,W'j)(i,W

j)(i,j)W'W(i,

(11)

Where W(i,j) and W'(i,j) are respectively the

original watermark bit and the extracted one and m

and n are the watermark dimensions.

To test the performance of the proposed

watermarking scheme in term of robustness, different

distortions and attacks are firstly performed on the

watermarked videos. Secondly, the NC is calculated

after the extraction process.

Figure 7 exhibits the NC values of attacked

videos. According to this figure, our watermarking

scheme is robust to salt and pepper and Gaussian

noise attacks. In fact, the NC reaches 0.99364 and 1

after applying respectively salt and pepper and

different white Gaussian noise mean. This high

correlation is guaranteed by involving the discrete

wavelet transform, which is resilient to noise adding.

Moreover, it is noticed that the proposed scheme

is immune to rotation attack. In fact, by varying the

rotation degrees the watermark is successfully

extracted. The obtained NC values against this attack

are up to 0.99976. This robustness is achieved due to

the use of the singular value decomposition SVD that

is invariant to geometric attacks and especially to

rotation attack.

Besides, results in this figure demonstrate the

robustness against median filter attacks and cropping

attacks. Indeed, the NC corresponding to cropping

Figure 7: Obtained NC values against different attacks.

VISAPP 2019 - 14th International Conference on Computer Vision Theory and Applications

602

attack is superior than 0.94 and attains 0.99988.

Concerning the median filter attack, the obtained NC

values vary between 0,941314 and 1.

Experimental results prove that the proposed

scheme is compression resilient by providing NC

values up to 0.99987. This high level of robustness is

reached thanks to the embedding of the watermark in

the mid frequency sub bands of the discrete wavelet

transform.

The redundant embedding of a full watermark in

every frame of each video sequence reinforces the

resilience of the proposed scheme against the frames

dropping attack. Indeed the minimal NC value after

removing 32% of the frames in every sequence is

0.95335.

4.3 Comparison with Existing Methods

In this section, we will compare the performance of

our proposed technique with three existing ones;

(Dolley and Manisha, 2018), (Narasimhulu, 2017)

and (Jeebananda and Prince, 2016).

According to NC results presented in table 2, it is

noticed that our algorithm and the technique proposed

in (Jeebananda and Prince, 2016) withstand Gaussian

noise attack. In fact, our proposed scheme efficiently

resists to this attack with the highest value of NC that

is equal to 1. However, the method presented in

(Dolley and Manisha, 2018) has a poor robustness to

Gaussian noise attack. Besides, both our technique

and (Jeebananda and Prince, 2016) one successfully

survive the median filter attack and our technique

shows the better NC value that is 1. All approaches

cited in the table 2 demonstrate their robustness

against salt and pepper and rotation and the best NC

values are provided by our technique.

Table 2: Comparison of robustness between the proposed

approach and other video watermarking methods.

Moreover, it is proven that our scheme provides

high level of robustness to MJPEG compression

compared to the techniques proposed in (Jeebananda

and Prince, 2016) and (Dolley and Manisha, 2018) by

offering a NC value that reaches 0.99987. Regarding

cropping attack, our method shows better robustness

than (Narasimhulu, 2017) with a NC value equal to

0.99976. On the other hand, the method presented in

(Jeebananda and Prince, 2016) and the proposed one

have a similar resilience to frames dropping attacks.

5 CONCLUSION

In this paper, a blind robust video watermarking

scheme based on discrete wavelet transform and

singular value decomposition was proposed. The S

component of the singular value decomposition

(SVD), which is applied to the mid frequency sub

bands of the discrete wavelet transform (DWT), is

used in the embedding process in order to achieve the

best trade-off between the imperceptibility and the

robustness requirements. Normalized Correlation

Coefficient (NC), as well as Peak Signal to Noise

Ratio (PSNR) and Structural Similarity index (SSIM)

are computed to scrutinize the proposed technique

performance. Experimental results prove that the

proposed scheme successfully sustains several

attacks namely geometrical, image processing and

compression. Comparing it with others techniques,

the proposed one shows high level of robustness. In

term of imperceptibility, the quality of the video is

maintained. Hence, it can be concluded that the

proposed approach is efficiently suitable for

applications, which require more robustness than

imperceptibility. Thereby, the future works will focus

on exploiting the developed technique for video

authentication goal in video surveillance context.

ACKNOWLEDGMENTS

The research leading to these results received funding

from the Ministry of Higher Education and Scientific

Research of Tunisia under the grant agreement

number LR11ES48.

REFERENCES

using blind video watermarking algorithm based on

MPEG-2 structure. In International Conference on

A Robust Blind Video Watermarking Scheme based on Discrete Wavelet Transform and Singular Value Decomposition

603

Computing, Communication & Automation, Noida,

2015, pages 1048-1053.

Amit, V., and Navdeep, K. G. (2016). Analysis of

watermarking techniques. International Journal of

Computer Science and Technology (IJCS), 7(1):153-

156.

Arti, B., and Ajay, K. (2017). Digital video watermarking

techniques: a review. International Journal

Engineering and Computer Science (IJECS),

6(5):21328-21332.

Asikuzzaman, M. and Pickering, M. R. (2018). An

overview of digital video watermarking. IEEE

Transactions on Circuits and Systems for Video

Technology, 28(9): 2131-2153.

Asim, N., Yasir, S., Nisar, A., and Aasia, R. (2015).

Performance evaluation and watermark security

assessment of digital watermarking techniques. Science

International, 27(2):1271-1276.

Ben Hamida, A., Koubaa, M., Ben Amar, C., and Nicolas,

H. (2011). Hierarchical traceability of multimedia

documents. In IEEE Symposium on Computational

Intelligence in Cyber Security (CICS), Paris, 2011,

pages 108-113.

Charfeddine, M., El'Arbi, M., and Ben Amar, C. (2014). A

new DCT audio watermarking scheme based on

preliminary MP3 study. Multimedia Tools and

Applications, 70(3):1521-1557.

Dolley, S., and Manisha, S. (2018). Robust scene-based

digital video watermarking scheme using level-3 DWT:

approach, evaluation and experimentation.

Radioelectronics and Communications Systems,

61(1):1-12.

Guedri, B., Zaied, M., and Ben Amar, C. (2011). Indexing

and images retrieval by content. In International

Conference on High Performance Computing and

Simulation (HPCS), Istanbul, 2011, pages 369-375.

Hedayath, B. S., Gangatharan, and N., Tamilchelvan, R.

(2016). A survey on video watermarking technologies

based on copyright protection and authentication.

International Journal of Computer Applications

Technology and Research (IJCAT), 5(5):295-303.

Jeebananda, P., and Prince G. (2016). An efficient video

watermarking approach using scene change detection.

In 1st India International Conference on Information

Processing (IICIP), Delhi, 2016, pages 1-5.

Jyothika, A., and Geetharanjin, P. R. (2018). A robust

watermarking scheme and tamper Detection using

integer wavelet transform. In 2nd International

Conference on Trends in Electronics and Informatics

(ICOEI), Tirunelveli, India, 2018, pages 676-679.

Kadu, S., Naveen, C., Satpute, V. R., and Keskar, A. G.

(2016). Discrete wavelet transform based video

watermarking technique. In International Conference

on Microelectronics, Computing and Communications

(MicroCom), Durgapur, 2016, pages 1-6.

Koubaa, M., Elarbi, M., Ben Amar, C., and Nicolas, H.

(2012). Collusion, MPEG4 compression and frame

dropping resistant video watermarking. Multimedia

Tools and Applications, 56(2): 281-301.

Mostafa, S., and Ali, A. (2016). A Multiresolution video

watermarking algorithm exploiting the block-based

motion estimation. Journal of Information Security

Engineering (JIS), 7(4):260-268.

Narasimhulu, C. V. (2017). A robust hybrid video

watermarking algorithm using NSCT and SVD. In

IEEE International Conference on Power, Control,

Signals and Instrumentation Engineering (ICPCSI),

Chennai, 2017, pages 1495-1501.

Naved, A., (2016). A robust video watermarking technique

using DWT, DCT, and FFT. International Journal of

Advanced Research in Computer Science and Software

Engineering (IJARCSSE), 6(6):490-494.

Omar, G., and Shawkat, K. G. (2018). A survey on

cryptography algorithms. International Journal of

Scientific and Research Publications (IJSRP),

8(7):495-516.

Othmani, M., Bellil, W., Ben Amar, C., and Alimi, A. M.

(2010). A new structure and training procedure for

multi-mother wavelet networks. International Journal

of Wavelets, Multiresolution and Information

Processing (IJWMIP), 8(1):149-175.

Tarfouni, N., Charfeddine, M., and Ben Amar, C. (2018). A

new robust and blind image Watermarking Scheme In

frequency domain based on optimal blocks selection. In

International Conference in Central Europe on

Computer Graphics, Visualization and Computer

Vision, Plzen, 2018, pages 78-86.

Tuan, T. N., and Duan, D. N. (2015). A robust blind video

watermarking in DCT domain using even-odd

quantization technique. In International Conference on

Advanced Technologies for Communications (ATC),

Ho Chi Minh City, 2015, pages 439-444.

Vikrant, M., and Shubhanand, S. H. (2016). A survey on

cryptography techniques. International Journal of

Advanced Research in Computer Science and Software

Engineering (IJARCSSE), 6(6):469-475.

Wali, A., Ben Aoun, N., Karray, H., Ben Amar, C., and

Alimi, A. M. (2010). A new system for event detection

from video surveillance sequences. Lecture Notes in

Computer Science (LNCS), 6475(2):110-120.

Zhang, R., Wang, H., and Wang, Y. (2012). "A novel image

authentification based on semi-fragile watermarking. In

Fifth International Joint Conference on Computational

Sciences and Optimization, Harbin, 2012, pages 631-

634.

VISAPP 2019 - 14th International Conference on Computer Vision Theory and Applications

604