Mohaisen and Morad Iraqi Journal of Science, 2016, Special Issue, Part A, pp:112-128 ____________________________ *Email: [email protected]112 Secure Information Using Steganography and Watermarking Hatem Nahi Mohaisen 1* , Ameer H. Morad 2 1 Department of Mathematics, College of Science, Baghdad University, Baghdad, Iraq 2 Al Khwarizmi college of Engineering, Baghdad University, Baghdad, Iraq Abstract This paper presents secure hiding of information which provides confidentiality and authentication based on techniques of video steganography and watermarking. The proposed system embeds both secret information and watermark image into cover video in random fashion using LSB method. For each embedding process, different sequences of frames are be selected randomly depending on initial seed value. The system includes two modules embedding, extraction. firstly, in embedding module, load the input arguments which are represented by cover video, secret information, watermark image, and initial seed, and finally, produced the stego-video which is the original cover video hides inside it the secret information, watermark image. In the extraction phase, the system receives the stego video in additional to the seed value and number of frames used in embedding phase. The extraction module uses both initial seed and number of frames to generate the same sequence of frames as in embedded phase, then extract secret information and watermark image. The authentication of the sender is tested by match between received and stored watermark. For evaluation of the proposed system and to be ensure that the original media (video) is not affected significantly due to embedding process, many measurements are being calculate from video before and after hiding process such as histograms, mean square error (MSE), peak signal to noise ratio (PSNR) and hiding time. The system is tested during embedding and extraction modules considering using several cover videos, secret information and seed values. Keywords: Multimedia contents security Steganography techniques, Watermarking, LSB Digital images and videos, Peak Signal-to-Noise Ratio (PSNR), Mean Square Error (MES). لمائيةمة ا والع الختزستخدام فن امعلومات با سرية ال محيسن حاتم ناهي1 ,* اد أمير حسين مر2 1 لرياضيا قسم ا ت امعة بغداد, بغداد, العرعلوم, جاية ال , كل ق2 رزمي, جامعةية هندسة الخوا كل اق بغداد, بغداد, العرصة: الخ نق الختزد على فن الة والتي تستنصالسرية والتي تزودنا بات امعلوماء اللبحث أمنية إخفا دم في هذا المائيةمة ا والعلفيديو للمائيةمة ات السرية والعمعلوما من ال من ك المقترح يضلنظام . ا إلىوي غطاء فيديستخدام طريقةئية با بطريقة عشوا عمليةكل أهمية. لقل البت اطر مختلفة من اسلختيار سين, يتم ا تضم بشكل عشوائيدا على قيمةعتما اولية البذرة ا.(Seed number) ى وحدتين التضميننظام عل ال يشمل, ص. أوستخ واتي تمثلهات ال المدختضمين تحمل في وحدة الت السرية,معلوماوي, الء الفديلغطا ا والبذرة المائيةمة ا صورة العصلي ومخفيء الغطاوي المضمن الذي هو اء الفديلغطا ا ينتج ابتدائية, وأخيرضافةلمضمن بام الفيديو امستلم الص, يستلستخة. في مرحلة المائي والصورة ات السريةمعلوماه ال في داخلص تستخدم كستخ التضمين. وحدة امستعملة في مرحلةطر ال البذرة وعدد ا إلى قيمة مولية ن البذرة اISSN: 0067-2904
17
Embed
Secure Information Using Steganography and Watermarking · 2016-10-16 · The system includes two modules embedding, extraction. firstly, in embedding module, load the input arguments
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Mohaisen and Morad Iraqi Journal of Science, 2016, Special Issue, Part A, pp:112-128
LSB Digital images and videos, Peak Signal-to-Noise Ratio (PSNR), Mean Square
Error (MES).
سرية المعلومات باستخدام فن االختزال والعالمة المائية
2أمير حسين مراد*, 1حاتم ناهي محيسن ق, كلية العلوم, جامعة بغداد, بغداد, العراتقسم الرياضيا1
بغداد, بغداد, العراق كلية هندسة الخوارزمي, جامعة2 الخالصة:
دم في هذا البحث أمنية إخفاء المعلومات التي تزودنا بالسرية واألصالة والتي تستند على فن االختزال نقغطاء فيديوي إلى. النظام المقترح يَضمن كال من المعلومات السرية والعالمة المائية للفيديو والعالمة المائية
تضمين, يتم اختيار سالسل مختلفة من األطر البت األقل أهمية. لكل عمليةبطريقة عشوائية باستخدام طريقة يشمل النظام على وحدتين التضمين (Seed number). البذرة األوليةاعتمادا على قيمة بشكل عشوائي
الغطاء الفديوي, المعلومات السرية, في وحدة التضمين تحمل المدخالت التي تمثلها واالستخالص. أوال,البتدائية, وأخيرا ينتج الغطاء الفديوي المضمن الذي هو الغطاء األصلي ومخفي صورة العالمة المائية والبذرة ا
في داخله المعلومات السرية والصورة المائية. في مرحلة االستخالص, يستلم المستلم الفيديو المضمن باإلضافة ن البذرة األولية مإلى قيمة البذرة وعدد األطر المستعملة في مرحلة التضمين. وحدة االستخالص تستخدم كال
ISSN: 0067-2904
Mohaisen and Morad Iraqi Journal of Science, 2016, Special Issue, Part A, pp:112-128
113
تولد نفس السلسلة من األطر, كما في مرحلة التضمين, ثم تنتزع المعلومات السرية وصورة وعدد األطر للمرسل نعمل تطابق بين العالمة المائية المستلمة والمخزونة. لتقييم النظام للتحقق من أصالة ا. العالمة المائية
العديد من المقاييس جرى حسابها ملحوظ بسبب عملية التضمين,ر بشكل المقترح وان الفيديو األصلي لم يتأثعلى الفيديو قبل وبعد التضمين مثل المدرج اإلحصائي, متوسط مربع الخطأ, نسبة ذروة اإلشارة للتشويش
وات مختلفة, يووقت اإلخفاء. تم اختبار النظام خالل عملية التضمين واالستخالص حيث طبق على عدة فيد .ةالبذور األوليوقيم معلومات سرية,
1. Introduction
The art of security of sensitive information has received much attention in the recent years as
security system has become a big concern in internet area. Development of many techniques and
mechanisms to maintain most of the major requirements for security services such as: confidentiality,
authentication, nonrepudiation, and integrity of information when transferred on internet. Now day,
still there are many challenges to gain pure secure media to exchange information.
Figure-1, shows the techniques and mechanisms which are being used in security system, special
focus on information hiding techniques. The proposed hiding system presented here revolves around
steganography in digital images (video) with watermarking (imperceptible) which be indicated by
shadow blocks in the figure. Many researches are published [2 - 11] to develop mechanisms and
algorithms to hiding secret information into carrier media such as text, audio, digital images and video.
Table-1 summarizes the methods and tools being used in each of these researches.
Figure 1- The different disciplines of security system, special focus on information hiding techniques.
Table 1- Comparison between set of researches done in the same field
research steganography watermark LSB DWT DWT+DCT MES PNSR Image Video
1
2
3
4
5
6
7
8
9
10
The aim of our research work is to develop a system uses steganography technique to hiding the
information into video portion to achieve confidentiality and uses embedded watermark image to
video for purpose of authentication of sender and integrity of the information. The system is use LSB
method. The embedding process of information is done into random sequence of video frames.
Security Systems
Information Hiding Cryptography
Robust
Fragile
Video
Steganography Watermarking
Technical Linguistic
Visible
Imperceptible
Fingerprint
Images
Audio
Text
Mohaisen and Morad Iraqi Journal of Science, 2016, Special Issue, Part A, pp:112-128
114
In each selected frame, the portion of information is hiding in the first bit of pixels and in other
hand the watermark is embedded randomly in one of next three bits of pixel bytes. Since video are
stored in computer using different format, each of which has its own features. The most popular video
type is AVI and the frames (images) type is BMP type.
2. The proposed system structure
The implementation video steganography and watermarking system consists of two modules:
hiding module and extraction module, each of these units contains of many submodules. To start up
with hiding module, it is hiding the information in the specific frame and position bit in the pixel, with
a watermark hiding in another bit in same pixel in such a way that the hidden information is
imperceptible to human eye. The hidden technique should be kept the perceptible information of a
video without change the information and watermark, which must founded by an extraction algorithm.
Embedding the information and watermark requires compatibility of the length of message with frame
capacity and the dimensional of watermark with the dimensional of frames.
Hiding Module: The hiding module algorithm, shown in Figure-2, consists of many sub-modules to
produce stego in which it combines of (cover video + secret information+ watermark). Each sub-
module is described by using following algorithms:
Preparing of the video, algorithm-1, load the video portion and convert it into a set of frames and
save these frames in folder to be ready to use by embedded process.
Hide of Secret Information, algorithm-2, enters the information and converts its characters into
binary according to ASCII code. Compute the size of the information (in bits). Randomly choose a
sequence of frame. Then start embedding the information bits into first least significant bit (b1) of each
frame pixels of the frame as shown in Figure-3a.
Algorithm 1- Preparing of the Video
1. Open window to choosing video file name and the format (AVI).
2. Read the video.
3. Convert the video into numbers of frames(images).
4. Constricting the speed of frame in video (i.e. frame per second (fps)).
5. Read the numbers of frame in video
6. Store these frames in specific folder by format (.BMP).
Algorithm 2-Hide of Secret Information
1. Prepare the information (secret data) from text file name with format .txt, or
typing it directly by using keyboard from main interactive page of the system.
2. Convert the information to binary; replace each character or symbol by its
ASCII code. 3. Measure the size of the information data (in bits). Determine the number of
frames that needs to hide the information according to its size. 4. Randomly choose frame from the frame's folder. 5. Hide bit by bit of a portion of information data in the first lest significant bit
(LSB) of each frame pixels. 6. Repeat steps (4 and 5) unit complete hiding of all information data.
Mohaisen and Morad Iraqi Journal of Science, 2016, Special Issue, Part A, pp:112-128
115
Figure 2- Hiding Process Algorithm
Hiding of Watermark, algorithm-3, the watermark image is type (bmp) format and is the same size
of video frame. Firstly convert it into binary image and the embedded each bits of watermark into one
of three least significant bits (b4 b3 b2) of each frame pixels. Here the embedded is done in two
methods. In first method, use pseudo random numbers to determine randomly where the watermark
bits are embed to one of bits (b4 b3 b2) as shown in Figure-3b .In second method, Figure-3c, divide the
each frame horizontally into three equal regions. Watermark bits are hiding into (b4) of pixels in the
first region and into (b2) of pixels in the second region and finally, in the third region, hides watermark
bits into (b3) of pixels.
a b c
Figure 3- a- Select b1 for embedded message bit. b- Randomly select one bit of (b2, b3, or b4) for embedded
watermark. c- Predetermine bits.
Mohaisen and Morad Iraqi Journal of Science, 2016, Special Issue, Part A, pp:112-128
116
Pseudorandom Number Generation, algorithm-4, one of main contribution for this research is the
process of selection the set of frames, from the folder, are done randomly. These frames will be used
later for hides both secret data and watermark image. In addition, the hide of watermark image into
one of bits (b4 b3 b2) is also done randomly. The selection method is full randomizing fashion to makes
the task of disclosure of the secret information more difficult. For each embedded process, different
sets of frames are selected depending on initial seed value. The seed value is one of input arguments
are given at the beginning of stenography process. The pseudorandom number generation is defining
by eq. (1).
x(i+1)=(a*xi + b)mod n (1)
Where xi is previous random number, and x0 is initial seed, n is number of frames imply in the video
portion, i is number of needed frame for embedded process, and a, b are constant.
Evolution, algorithm-5, PSNR value is used only to compare the performance of loss coding scheme.
Calculation of PSNR is depends on the mean square error (MSE). The MSE is average of the square
error of the two frames (differences in pixel values), and also, calculate histogram error.
3. Extraction Module
At receiver side, extraction process, as in Figure-4 is done in the same method of hiding process
but in reverse. The modules accept stego and initial seed with number of used frames. Firstly, as in
algorithm-4, generate random sequence of numbers, and then determine the set of frames in which
embedded the information and watermark. Extracting secret information and watermark image is
follow algorithms-3 and 2, only here instead of hiding, the process is extracting. In the end the
extracted information is appear on front page of MATLAB or save it in text file.
Algorithm 3- Hiding of Watermark
1. Choose watermark image with format of (*.bmp) and has same
dimensional with frame.
2. Convert the watermark image to binary image.
3. Randomly choose frame from the frame's folder.
4. Using the LSB technique to hidden bit by bit of watermark image
into one of bits (b4 b3 b2). For frame pixels.
Algorithm 4- pseudorandom Number Generation
1. Read the number of frames is implied in the cover video which is
representing n (modulo) in eq.(1).
2. Read the require numbers of frames to hidden process i
(subscription in eq.(1).
3. Use the pseudo random generator eq. (1) with given seed (x0)
value to produce random sequence of frames.
Algorithm 5- Evolution
1. Read the original frame and the same frame after embedding process.
2. Calculate MSE, PSNR, and MAE values for both frames in step-1.
3. Find histograms for both original and embedded frames (HO and HE).
4. Calculate histogram error=((HO-HE) /h*w)^2
Mohaisen and Morad Iraqi Journal of Science, 2016, Special Issue, Part A, pp:112-128
117
Figure 4- Extraction Process Algorithm.
4. Testing and Results
All algorithms and interactive front pages for the proposed hiding system are being implemented
using MATLAB software. Figure-5 shows the front pages for the hiding system steps. Load the video
portion to the system as in Figure-5a. Enter the secret information and seed value and then determine
number of needed frames as shown in Figure-5b. Figure-5c shows the process of embedding the secret
information and watermark image into frames. Finally, reconstruct the video in original order of
frames to produce stego_video and also, calculate PSNR and MSE and draw histograms for each
frame before and after embedding process. In extraction process is as in Figure-5d, first extract X and
S values, then extract both secret information and watermark image.
Mohaisen and Morad Iraqi Journal of Science, 2016, Special Issue, Part A, pp:112-128
118
Figure 5- Hiding process steps. a- Load video, split into frames and save in folder.
b- Input secret information and seed value. c- Generate random sequence of frames, embedded information and
watermark image. Calculate PSNR, MSE, and embedding time.
d- Extraction process.
Experimentally the system is tested using four different video portions with properties listed in
Table-2 and Figure-6 show some samples of frames in each of videos. For all tests, use unique
watermark image which is appeared in Figure-7.
a
d
b
c
Mohaisen and Morad Iraqi Journal of Science, 2016, Special Issue, Part A, pp:112-128
119
Table 2- Four video portions
Sample video name Duration Frame rate No. of frames Width Height
Bird 11 sec 30 355 256 240
Cbw3 6 sec 40 250 256 240
Drop 6 sec 30 182 256 240
Flame 3 sec 35 110 256 240
Bird
Cbw3
Drop
Flame
Figure 6- Samples of frames in each of videos.
1. In the first experiment, the proposed system is tested with compare to second system in which
the hiding of information and watermark are embedded in fixed predetermine bits shown in Figure-3c.
Loading Bird video portion and split it into 355 frames. Also, load watermark image showing in
Figure-7, convert it into binary image.
Embedding the information and binary watermark image into frame number (68) uses random
distribution method and predetermine method are shown in Figure-8. The results, listed in Table-3,
indicate that somewhat the random distribution is better than predetermine distribution even the last
one is more speedily, but still in random distribution the frames little affected by embedded process,
also, the confidentiality for proposed system is better. For proposed system, in each attempt of hiding
process, the secret information and watermark image are distributed randomly on the frame, i.e. the
distributions is not linearly and differ for each frame, but in second system the distribution of secret
information watermark image are the same for each trial of hiding process for all frames.
Figure 7- Watermark image
Table 3- Hiding process evaluation comparison for both cases
After embedded process
Random distribution Predetermine
distribution
PSNR 36.6983 36.593
MSE 13.9067 14.249
MAE 0.240 0.242
Histogram Error 0.0011587 0.00140
Hiding time (sec) 0.6552 0.5616
Mohaisen and Morad Iraqi Journal of Science, 2016, Special Issue, Part A, pp:112-128
120
Figure 8- Frame-68 for Bird video. a- The image and histogram for original frame 68. b- The image and
histogram for frame 68 after embedded information and watermark using random distribution. c- The image and
histogram for frame 68 after embedded information and watermark using predetermine distribution.
1. The proposed system is tested by loading Bird video portion and split it into 355 frames. for
experimental purpose, let the number of needed frames to embedding the secret message is (4 frames).
Then enter initial seed number (14), and randomly generate the sequence of 4 numbers (14, 211,248,
and 300) as shown in Figure-9.Then run the embedded process to embed both secret message and
binary watermark image. The result as shown in Figures-9c, d, e, and f.
b
a
c
Mohaisen and Morad Iraqi Journal of Science, 2016, Special Issue, Part A, pp:112-128
121
Figure 9- Embedding process using cover Bird video, with initial seed (14) and four frames. a- Enter initial seed.
b- watermark image. c- Embedded process in frame (14). d-Embedded process in frame (211). e- Embedded
process in frame (248). f-Embedded process in frame (300).
The minimum and maximum of, PSNR (36.802-36.05), MSE (13.5726-14.0614), MAE (0.242171-
0.24834), histogram error (0.00107776-0.00118636), and embedded time (0.6240-0.6708) Sec
respectively. Also, the histograms, before and after embedding process, for each frame are shown in
Figure-10.
a
c
b
f e
d
f e
d c
a
Mohaisen and Morad Iraqi Journal of Science, 2016, Special Issue, Part A, pp:112-128