INTRODUCTION
In the recent years, a huge amount of digital information is circulating through the world by means of the rapid and extensive growth in internet technology; therefore there is a pressing need to develop several newer techniques to protect copyright, ownership and content integrity of digital media. Most of such data is exposed and can be easily forged or corrupted, consequently the need for intellectual property rights protection arises. This necessity arises because the digital representation of media possesses inherent advantages of portability, efficiency and accuracy of information content on one hand, but on the other hand, this representation also puts a serious threat on easy, accurate and illegal perfect copies of unlimited number. Unfortunately, the currently available formats for image, audio and video in digital form do not allow any type of copyright protection. Digital watermarking has been proposed as one of the possible ways to deal with this problem, to keep information safe.
Digital watermarking, an extension of steganography, is a promising solution for content copyright protection, it imposes extra robustness on embedded information. In other words, digital watermarking is the art and science of embedding copyright information in the original files. The information embedded is called watermarks.
Information hiding is a general practice encompassing a broad range of applications
in which the messages are embedded into the other media content for varying
purposes, while watermarking and steganography are two types of information
hiding. Steganography, which is derived from the Greek words means, covered
writing that hides the secret message into innocuous host content to achieve
covert communication (Lin and Delp, 1999). In order
to act as a successful camouflage to conceal the very existence of the secret
message, the host media content is usually chosen to have nothing to do with
the hidden information. Similar to Steganography, watermarking is also a procedure
of imperceptibly embedding the information, i.e., a digital watermark, into
the content. However, a digital watermark usually represents the ownership of
the content, the identity of the legitimate content user or other information
used to help protect the lost content. In other words, there exists a strong
relationship between the embedded digital watermark and the host content. Besides,
in order to achieve the intended functions, the existence of a digital watermark
is usually known to the users, in contrast to the fact that the hidden. between
the host media content and hidden information is a differentiating factor between
digital watermarking and steganography (Cachin, 1998).
The digital image watermarking techniques in the literature are typically grouped
into two classes (Yeung et al., 1998): the spatial
domain spatial domain and frequency domain watermarking techniques. Compared
to spatial domain techniques (Darmstaedter et al.,
1998) frequency domain techniques proved to be more effective with respect
to achieving the imperceptibility and robustness requirements of digital watermarking
algorithms Commonly used frequency domain transforms include the Discrete Cosine
Transform (DCT), Wavelet Transform (WT), the Discrete Contourlet Transform (CT)
and the Discrete Fourier Transform (DFT). However, DWT has been frequently used
due to its excellent spatial localization and multi resolution characteristics,
which is very close to theoretical models of the human visual system (Cabir
and Serap, 2007; HoungJyh et al., 1998; Inoue
et al., 1999). Recent improvements obtained by combining these frequency
domain transforms (AlHaj, 2007):
• 
Recently, to recognize key ideas, theories and conclusion and sets the
difference and similarities. In the following, we will summarize some of
such related works 
• 
Stephan (2005) embedding watermarking in still images
(BMP) true color, this method applies embedding watermark in large coefficients
and in high frequency subbands by using discrete wavelet transform. The
watermark in this method is capable of surviving against the JPEG2000 compression
and the watermark extracted using original image (nonblind watermark) 
• 
Lepik (2007) demonstrates that the Haar wavelet
method is a powerful tool for solving different types of integral equations
and partial differential equations. This method with less degree of freedom
and with smaller CPU time provides better solutions than classical ones 
• 
Cabir and Serap (2007) a new digital image watermarking
algorithm that combines the strengths of the moment based image normalization
and two dimensional discrete wavelet transform was proposed by the researchers.
Normalization provides robustness against geometrical degradations, whereas,
discrete wavelet transform achieves immunity for compression, linear and
nonlinear filtering by taking the properties of the human visual system
into consideration. This method is powerful to resist numerous image manipulations. 
• 
Wu et al. (2008) the authors in their work
propose a novel watermarking method to solve the problem of. For copyright
protection, their new method makes a difference by providing the user with
the power to process masses of digital image watermarking tasks using just
one private key 
• 
Tsai (2009) the authors studied novel visible
watermarking algorithm based on the content and contrast aware (COCOA) technique
with the consideration of Human Visual System (HVS) model. In order to determine
the optimal watermark locations and strength at the watermark embedding
stage, the COCOA visible watermarking utilizes the global and local characteristics
of the host and watermark images in the discrete wavelet transform (DWT)
domain 
• 
Leung et al. (2009) these researchers proposed
a selective curvelet coefficient digital watermarking algorithm. The selective
band provides an addition security feature against any physical tampering.
Their reported goal was to give an intensive study on the robustness of
watermarking using selective curvelet coefficients from a single band and
to find out the best band for embedding watermark. Wrapping of specially
selected Fourier samples is employed to implement Fast Discrete Curvelet
Transforms (FDCT) to transform the digital image to the curvelet domain 
• 
Bayram and Selesnick (2009) they developed an
over complete discrete wavelet transform (DWT) based on rational dilation
factors for discretetime signals. It was implemented using selfinverting
FIR filter banks. It is approximately shiftinvariant and can provide a
dense sampling of the timefrequency plane. This algorithm is based on matrix
spectral factorization 
In this study, we propose a robust method for digital watermarking and secure
copyright protection of digital images. The proposed watermarking system is
based on cascading two transforms; HWT and DWT. The system is proofs resist
against numerous image attacks. Furthermore, the method is easy to implement
and suitable for real time application. Adding a private key to the new technique
gives more robustness and security to the watermarked image against attacks.
WATERMARKING OVERVIEW
Digital watermarking means embedding information into digital material in such
a way that it is imperceptible to a human observer but easily detected by computer
algorithm. A digital watermark is a transparent, invisible information pattern
that is inserted into a suitable component of the data source by using a specific
computer algorithm. It is a signal added to digital data (audio, video, or still
images) that can be detected or extracted later to make an assertion about the
data (Amin et al., 2003).
Watermarking applications: Digital watermarking is described as a viable
method for the protection of ownership rights of digital image and other data
types. It can be applied to different applications including digital signatures,
fingerprinting, broadcast and publication monitoring, authentication, copy control
and secret communication (Cox et al., 1997, 2000;
Karzenbeisser and Perircolas, 2000).
Properties of watermarking system: When designing a watermarking system,
several properties must be observed, among which are the following (Kutter
and Hartung, 2000):
• 
Imperceptibilitythe watermark should be invisible not to degrade data
quality and to prevent an attacker from finding and deleting it 
• 
Readily detectablethe data owner or an independent control authority
should easily detect the watermark 
• 
Unambiguousretrieval of it should unambiguously and unequivocally identify
the owner of the data with a high degree of confidence 
• 
Robustdifficult to remove without producing a remarkable degradation
in data fidelity 
• 
Securityunauthorized parties should not be able to read or alter the
watermarking 
Watermarking techniques: There are many different watermarking techniques
and they range from the very simple to the complex (Das
et al., 2010). Obviously the type and the value of the content would
determine the watermarking technique to be used. For the image watermarking,
there are a number of schemes of varying robustness that have been implemented
(Haldar, 2008). These techniques have their strong and
weak points. Typically they fall into two categories: Spatial and Transform
domain (Yeung et al., 1998).
Spatial domain watermarking: Watermarking was the first scheme that
introduced works directly in the spatial domain. By some image analysis operations
(e.g., edge detection) it is possible to get perceptual information key, directly
in the intensity values of predetermined regions of the image (Paquet,
2001). Those simple techniques provide a simple and effective way for embedding
an invisible watermark into an original content but don't show robustness to
common alterations (Cox et al., 2002; Wolfgang
et al., 1999). One of the most famous spatial techniques is Least
Significant Bit (LSB) (Hanjalic et al., 2000).
One straightforward and rapid technique is based on the principle of generating
a pseudogenerated noise pattern and integrates it into specific chrominance
or luminance pixel values (Darmstaedter et al., 1998).
Such pseudorandom noise patterns consist of black (1), white (1) and neutral
values (0). The pseudo noise is generated with a secret key and algorithm. Additionally,
the process could be adjusted to the image components or feature vectors to
achieve a higher level of invisibility. In general, the watermark W(x, y) is
integrated into the image components I(x, y) by a factor that allows amplification
of the watermarking values in order to obtain the best results as shown in Eq.
1:
Frequency domain watermarking: It is also known as transforming domain
watermarking (Grans, 2003). Another way to produce high
quality watermarked content is by first transforming the original content (e.g.,
Image) into the frequency domain by the use of Fourier, Discrete cosine or wavelet
transform. With this technique, the marks are not added to the intensities of
the image but to the values of its transform coefficients (Robi,
2004), then inverse transforming the marked coefficients from the watermarked
image. The use of frequency based transforms allows the direct understanding
of the content of the image (Grans, 2003). Therefore,
characteristics of the Human Visual System (HVS) can be taken into account more
easily when it is time to decide the intensity and position of the watermarks
to be applied to a given image (Kunder and Hatzinkos, 2001).
Because high frequencies will be lost by compression or scaling, the watermark
signal is applied to lower frequencies, or better yet, applied adaptively to
frequencies containing important elements of the original image (Murtag,
2007). The following are some techniques of the frequency transform domain.
Discrete Fourier Transform (DFT): The scholar Joseph Fourier in 1822 produced what is known as Fourier analysis, which is a method to present periodic signals by using a series of sine and cosine.
The transformer transfers the signal from the space of time to space of frequency and vice versa and Fourier transform is mathematically defined as in Eq. 2:
But the problem is that the Fourier transform becomes inactive for the nonstationary
signals (variable frequency) because it does not provide us with information
on the frequency content over time (Dittmann, 2000).
The Discrete Cosine Transform (DCT): DCT is a real domain transform
which represents the entire image as coefficients of different frequencies of
cosines (Which are the basis vectors for this transform). The DCT of the image
is calculated by taking (8 x8) blocks of the image, which are then transformed
individually. DCT also forms the basis of JPEG image compression algorithm,
which is one of the most widely used image data storage formats. The DCT approaches
are able to withstand some forms of attack (Hsu and Wu,
1998; Dittmann, 2000).
The Wavelet Transform based techniques (WT): The wavelet transform provides
the time frequency transformation of a given signal (Paquet,
2001). Wavelet transform is capable of providing the time and frequency
information simultaneously, hence giving a timefrequency representation of
the signal.
The problem is the huge number of wavelet resulting from the use of all the gradations in the process of analysis and the reams of information, which also produced for the same reason and therefore the treatment process requires a very long time.
Two transformers are using unlimited number of gradations, rather than make
the conversion for all the gradations and are done by selecting time domains
in the signal (Kunder and Hatzinkos, 2001; Inoue
et al., 1999; Radomir and Bogdan, 2003). This
conversion produces a sufficient quantity of information, with less time of
accounting and maintaining the basic information of the depicted reference (i.e.,
without the loss of important information).
Attacks on digital watermarks Watermarking research has produced a wide
range of watermarking techniques that can be subdivided into various methodological
complexity levels. Each of these methods attempts to reduce vulnerability in
various attack scenarios. Attacks on digital watermarks can be mainly classified
into two major groups:
• 
Friendly and malicious attacks (Hanjalic
et al., 2000; Hartung et al., 1999) 
• 
Conventional image or data operations applied in the normal use of computer
technology can destroy the watermark information. Different operation of
the classical image processing field, such as scaling, color and gamma corrections
and so forth, can be identified at this point. Today, compression techniques
can also be placed in the field of classical operations, but often separated
as a single element in watermarking research. The friendly attack has two
common features. It is generally described as an unintentional event where
the user has no suppose and/or knowledge of the watermark and its embedded
procedure. The second type of attack, the malicious attack, occurs with
the intention of eliminating the information (Hanjalic
et al., 2000) 
THE PROPOSED SYSTEM
The proposed system generally consists of two independent, but complementary subsystems. The first subsystem is called the Haar Wavelet Transform (HWT) and the second subsystem is called the Discrete Wavelet Transform (DWT). The general structure of the proposed system is shown in Fig. 1.
Haar Wavelet Transform subsystems (HWT): The main tasks of the HWT subsystems:
• 
A standard decomposition of a 2D signal (image), this is done by performing
a one dimensional transformation on each row followed by a one dimensional
transformation of each column 
• 
Construct a function that Haar Transforms an image for differed levels 
Figure 2 shows the general structure of the HWT and shows its main parts. Essentially, each part is a complementary for the other parts and represents a task in the HWT Process.
This subsystem applies wavelet transform by using Haar Wavelet. The main function
of Haar Wavelet Transform can be explained through the following steps:
• 
Load an image (300x300 pixel) 
• 
Apply HWT Process. It consists of the following processes: 

Fig. 1: 
General structure of the proposed system 
Line transformation: This also is called row. It decompose the image
into rows using the flowing code:
Columns transformation: It decompose the image into columns using the
following code: another image:
Two, three, four layers wavelet transform: It decomposes the image into
two layers, three layers, or four layers. To understand how the Haar Wavelets
Transform works, let us consider the following simple example; suppose we have
one dimension we have image with a resolution of four pixels having values [8,
6, 3, 7].
Table 1: 
Decomposition to lower resolution 

Haar wavelet basis can be used to represent this image by computing a wavelet
transform. To do this, we find the average of two pixels together, results the
pixel values [7, 5]. Clearly, some information is lost in this averaging process.
We need to store some detail coefficients to recover the original four pixel
values from the two averaged values. In our example, 1 chosen for the first
detail coefficient, since the average computed is 1 less than 8 and I more than
and 1 more than 6. This single number is used to recover the first two pixels
of our original fourpixel image. Similarly, the second detail coefficient is2,
since 5+ (2) = 3 and 5(2) = 7. Thus, the original image is decomposed into
a lower resolution (twopixel) version and a pair of detail coefficients.
Regarding this process recursively on the averages gives the full decomposition shown in Table 1.
Thus, for the onedimensional Haar basis, the wavelet transform of the original fourpixel image is given by [6, 1, 1 and 2]. We call the way used to compute the wavelet transform by recursively averaging and differencing coefficients, filter bank.
We can reconstruct the image to any resolution by recursively adding and subtracting the detail coefficients from the lower resolution version.
Compression of 2D image with haar wavelet technique: It has been shown in the previous section how 1D image can be treated as sequences of coefficients. Alternatively, we can think of images as piecewise constant functions on the halfopen interval [0, 1]. To do so, the concept of a vector space is used. A onepixel image is just a function that is constant over the entire interval [0, 1]. Let V^{0 }be the vector space of all these functions. A two pixel image has two constant pieces over the intervals [0, 1/2] and [1/2, 1]. We call the space containing all these functions V^{1}. If we continue in this manner, the space V^{j} will include all piecewiseconstant functions defined on the interval [0, 1] with constant pieces over each of 2^{j} subintervals. Note that because these vectors are all functions defined on the unit interval, every vector in V^{j} is also contained in V^{j+1}. For example, we can always describe a piecewise constant function with two intervals as a piecewiseconstant function with four intervals, with each interval in the first function corresponding to a pair of intervals in the second. Thus, the spaces V^{j} are nested; that is, V^{0}⊂V^{1}⊂V^{2} this nested set of spaces V^{j} is α necessary ingredient for the mathematical theory of multiresolution analysis. It guarantees that every member of V^{0} can be represented exactly as a member of higher resolution space V^{1}. The converse, however, is not true: not every function G (x) in V^{1} can be represented exactly in lower resolution space V^{0}.
Now we define a basis for each vector space V^{j}. The basis functions for the spaces V^{1} are called scaling functions and are usually denoted by the symbol L. A simple basis for V^{j} is given by the set of scaled and translated box functions as shown in Eq. 3:
Where:
The wavelets corresponding to the box basis are known as the Haar wavelets, given by Eq. 4:
Where:
Thus, the DWT for an image as a 2D signal will be obtained from 1D DWT. we get the scaling function and wavelet function for 2D by multiplying two 1D scaling functions: Ø (xy) = Ø (x) Ø (y). The wavelet functions are obtained by multiplying two wavelet functions for wavelet and scaling function for 1D. For the 2D case, three exist there wavelet functions that scan details in horizontal Ψ(1) (xy)) = Ø (x) Ψ(y), vertical Ψ(2) (x.y) = Ψ(x) Ø (y) and diagonal directions; Ψ(3) (x.y) = Ø(x) Ψ(y). This may be represented as a four channel perfect reconstruction filter bank. Now; each filter is 2D with the subscript indicating the type of filter high pixels frequency (HPF) low pixels frequency (LPF) for separable horizontal and vertical components. By using these filters in one stage, an image is decomposed into resolution: horizontal (HL), vertical (LH) and diagonal (HH). The operations can be repeated on the low (LL) band using the second stage of identical filter bank.
Thus, a typical 2D Hear transform, used in image compression and can be represented as a four channel perfect reconstructions as shown in Fig. 3.

Fig. 3: 
Structure of 2D haar wavelet proposed systems 

Fig. 4: 
Structure of wavelet decomposition 
The WT (Wavelet Transform) separates an image into a lower resolution approximation
image (LL) as well as horizontal (HL), vertical (LH) and diagonal (HH) detail
components. The process can then be repeated to compute multiple scale wavelet
decomposition, as in the three scales WT shown in Fig. 4.
The transformation of the 2D image is a 2D generalization of the 1D wavelet transformed already discussed. This operation provides us with an average value and detail coefficients for each row. Next, these transformed rows are treated as if they were themselves an image and apply the 1D transform to each column. The resulting values are all detail coefficients except a single overall average coefficient. In order to complete the transformation, this process is repeated recursively only on the quadrant containing averages.
Now let us see how the 2D Harr Wavelet Transformation is performed. The image is comprised of pixels represented by numbers consider the 8x8 image taken from a specific portion of a typical image shown in Fig. 5. The matrix (a 2D array) representing this image is shown in Fig. 6.
Now we perform the operation of averaging and differencing to arrive at a new matrix representing the same image in a more concise manner. Let us look how the operation is done.

Fig. 6: 
2D arrays that representing Fig. 5 
• 
Averaging: ( 56+10)/2 = 33, (1+63)/2 = 32, (58+8)/2 = 33, (10+54)/2
= 32 
• 
Differencing: 5633 = 23, 132 =31, 5833 = 25 and 1032 =22 
So, the transformed now becomes (33 32 33 32 2331 2522). Now the same operation on the average values i.e., (32.5 32.5 0.5 0.5 2331 2522) is performed. Then we perform the same operation on the averages i.e. first two elements of the new transformed row. Thus the final transformed row becomes (32.5 0 0.5 0.5 3231 2522). The new matrix we get after applying this operation on each row of the entire matrix of Fig, 6 is shown in Fig. 7.
We get the final transformed matrix as shown in Fig. 8. This
operation on rows followed by columns of the matrix is performed recursively
depending on the level of transformation meaning the more iteration provides
more transformations. Knowing that the lefttop element of the Fig.
8 i.e., 32.5 is the only averaging element which is the overall average
of all elements of the original matrix and all the remaining elements are details
coefficients.
The point of the wavelet transform is that regions of little variation in the original image manifest themselves as small or zero elements in the wavelet transformed version. A matrix with a high proportion of zero entries is said to be sparse for most of the image matrices. Their corresponding wavelet transformed versions are much sparser than the original. Sparse matrices are easier to store and transmit than ordinary matrices of the same size.
This is because the sparse matrices can be specified in the data file solely in terms of locations and values of their nonzero entries.
It can be seen that in the final transformed matrix, we find a lot of zero
entries. From this transformed matrix, the original matrix can be easily calculated
just by the reverse operation of averaging and differencing, i.e., the original
image can be reconstructed from the transformed image without loss of information.

Fig. 7: 
Transformed array after operation 

Fig. 8: 
Final transformed matrix after one step 
Thus, it yields a lossless compression of the image. However, to achieve more
degree of compression, we have to think of the lossy compression.
Namely by using a suitable threshold, that is replacing by zeros all entries with small absolute value. But this means that the inverse transformation will not produce the original mage exactly.
The watermarking based on WT uses Eq. 5:
where, W_{i} denotes the coefficient of the transformed image into wavelet domain, x_{i} the bit of the watermarking to be embedded and αis a scaling factor. Figure 9 shows the embedding of a watermarking in the wavelet domain.
The following paragraph explains the embedding watermark in wavelet method;
input the original image (24bit), the embedding algorithm is described in the
following steps:
Step 1: 
Input the original image, anther image 
Step 2: 
The size of the image should be same 
Step 3: 
Decompose image by using Haar wavelet transform 
Step 4: 
Load the watermark into the suitable subband of the original image 
Step 5: 
Convert the watermark into a stream of bits (zeroes and ones) 
Step 6: 
The watermark will match the size of the matrix 
Step 7: 
Convert every image from RGB to matrix color format 
Step 8: 
Save watermarked color image 
Step 9: 
Display watermarked image 
Discrete wavelet transform subsystems: The second subsystem is the Discrete
Wavelet Transform (DWT).

Fig. 9: 
Embedding of a watermarking in the wavelet domain 
Figure 10 shows the general structure of the DWT and shows its main parts.
This subsystem consists of two parts, which are Hide and Extract.
Hide (embedding) part: To understand the embedding technique we will describe it through the following:
Let I be a color image with M*N pixels which consists of channels R, G and B. The three channels are divided into a set of n*n (n is odd) nonoverlapping subblocks. In general, the size of the subblock has an influence on the robustness of the watermark. Each watermark bit can be embedded into one subblock by modifying the values of the subblock’s middle pixel and the other pixels. Let us define m as the middle pixel value and u as the mean value of the other pixels. For example, it is supposed that the authors have a 3*3 subblock as shown in Fig. 11. In this case, P5 is the middle pixel value m and the mean value of the other pixels can be computed by μ (P1 + P2 + P3 + P4 + P6 + P7 + P8 + P9)/8. To control the balance between the robustness and image quality, the robustness coefficient value T must be used; different robustness coefficient values for channels R, G and B are used:
TR is the robustness coefficient value of channel R, TG is for channel G and TB is for channel B.
Basically, the visual effect to modify the R, G and B channels is different in terms of the human visual sensitivity. In addition, the B channel has the larger tolerance to be modified than that of other channels. For these reasons, the robustness coefficient value TB used in the scheme is the largest value. TR and TG are the secondly and minimal respectively. The other advantage to adopt varied robustness coefficient is that the whole survival rate of watermark bit can be enhanced.

Fig. 11: 
Subblock of size 3*3 
The Hide part process is described as follows:
• 
Load the original image which has a size larger than the size of second
image (embedded image) 
• 
Load the second image which has size smaller than the original image.
The second image may be a watermarked image generated by the first subsystem
(HWT) 
• 
Save image 
• 
Add password 
This part is giving more security for the watermarked image because it's embedded image, which differ than other techniques. Used in watermarking As well as, adding the password key gives additional security for the watermarked image.
The procedure of the embedding algorithm is as follows:
• 
Input original image, image watermarking 
• 
Divide channels R, G and B into a set of n*n nonoverlapping subblocks,
respectively 
• 
Set the robustness coefficient values TR, TG and TB for channels R, G
and B respectively. In the next step, T represents TR, TG and TB when channels
R, G and B are chosen to hide the watermark, respectively 
• 
Modify m (middle pixel) and μ (other pixels) for watermark embedding
in the following: 
No Modify
Else
No Modify
Else
• 
Add the password (owner PRK) to watermarking to hide part 
• 
Save watermarked image 
In Eq. 6 and 7, the μ value can be
adjusted by subtracting or adding the pixel values P1, P2, P3, P4, P5, P6, P7,
P8 and P9, respectively. When the first watermark bit has been embedded on channel
R, the same location on channels G and B are also chosen to embed the first
watermark bit.
Extract (recover) part: In the method of watermark Extraction in Wavelet, we need to input the watermarking image where the output is the original image. Watermark extraction needs to have some original data (original image). It is performed using Independent Component Analysis (ICA) which is applied to the bands of original and watermarked images and, extracted by the backward embedding formula. The procedures of an extraction after various attacks are realized in purpose to check the watermark robustness against attacks. The quality of the extracted watermark is calculated using the correlation coefficient. The advantages of ICA algorithm approach include storage of less information by the image’s owner and better quality of the extracted watermark in the case of attacks.
Independent Component Analysis (ICA) is a statistical and computational technique
for revealing hidden factors that underlie sets of random variables, measurements,
or signals. ICA defines a generative model for the observed multivariate data,
which is typically given as a large database of samples. In the model, the data
variables are assumed to be linear mixtures of some unknown latent variables
and the mixing system is also unknown.

Fig. 12: 
Extraction scheme using ICA 
The latent variables are assumed to be not Gaussian and mutually independent
and they are called the independent components of the observed data. These independent
components, also called sources or factors, can be found by ICA which is superficially
related to principal component analysis and factor analysis. ICA is a much more
powerful technique, however, capable of finding the underlying factors or sources
when these classic methods fail completely.
Following are the steps of Extract part process as shown in Fig.
12:
• 
Load the DWT image (generated in Hide part) 
• 
Add password (that used in Hide part) 
• 
Display and save all Hided images 
The extraction algorithm is described in the following steps:
Step 1: 
Input watermarked image 
Step 2: 
Add the password (owner PRK) to the watermarked image 
Step 3: 
Divide channels R, G and B into a set of n*n nonoverlapping subblocks,
respectively 
Step 4: 
Compute the middle pixel value m and the mean value μ of the subblock 
Step 5: 
Recover the watermark bit by comparing m with μ according to the
following statements: 
If m>μ: The watermark bit 1 is extracted
Else: The watermark bit 0 is extracted
Step 6: 
Read the next watermarked subblock and repeat Step 4 and 5 until all the
watermark bits are extracted 
Step 7: 
Display original image 
When the first watermark bit is extracted from channel R, the watermark bit
hidden on channels G and B can also be extracted using the proposed extracting
algorithm.
EXPERIMENTAL RESULTS
Here, we will demonstrate the steps of using the proposed system. After selecting
the watermark image shown in Fig. 13, the user should select
the Haar Wavelet Transform subsystem. Figure 14, 15
show the effect of choosing the line transform and the column transform processes.
Figure 1618 show the effect of 2Layer
WT, 3Layer WT and 4 Layer WT on the selected image, respectively.
The processes of HWT can be selected randomly and in any combination. After
selecting the desired HWT, the next step is selecting the watermarking option
form the main menu and chooses another image (using the same steps of open option)
Fig. 19 shows process of watermarking on the original image.

Fig. 15: 
Column transform 
The next step, the user should choose Discrete Wavelet Transform (DWT) process.
DWT process consists of two parts Hide and Extract.

Fig. 19: 
First watermarking process 

Fig. 20: 
Final result of hide part 

Fig. 21: 
The extracted watermarked image 
The hide (embedding) part: This part is to hide/embed an image (or HWT)
inside original image. It consists of the following steps:
• 
Load the original image 
• 
Load HWT image 
• 
Save resulted image 
• 
Add password 
• 
Hide button 
Figure 20 shows the result of the hide (embedding) part.
The Extract (recover) part: The task of this part is to extract/recover the
images that embedded in the Hide part. It consists of the following steps:
• 
Load the DWT image 
• 
Put the password 
• 
Extract the watermarked image 
Figure 21 shows the extracted watermarked image.
DISCUSSION
The robustness of the generated images based on the proposed system is tested using different types of attacks (Invert, Rotate, Crop and Scale). We create a program calls Image Attack that will be used to attack the watermarking image resulted from the proposed system. Figure 22 shows the general structure for our Image program attacks.
Figure 2326 show the 4 different types
of attacks (Invert, Rotate, Crop and Scale) applied to the watermarking image,
respectively.
After attacking the above image, we can extract the original image using the
Extract part of the DWT process. Figure 2730
show and proves that the watermarked image is extracted without any effects
of these attacks.
The proposed digital image watermarking algorithm is constructed by cascading
two different but complementary techniques: the Discrete Wavelet Transform and
Haar Wavelet Transform to provide robustness image to all attacking. The algorithm
is proofs resist against numerous image manipulations.

Fig. 22: 
Image attack structure 

Fig. 27: 
Extacted watermarked image as a result of the scale attack


Fig. 28: 
Extacted watermarked image as a result of the invert attack


Fig. 29: 
Extacted watermarked image as a result of the rotate attack


Fig. 30: 
Extacted watermarked image as a result of the crop attack

Furthermore, the method is easy to implement and suitable for real time applications.
Adding a private key to the new technique gives more robustness and security
to the watermarked image against attacks.
The robustness of the proposed system has been improved by using two powerful techniques HWT and DWT. The robustness of both techniques alone has been evaluated and tested by many researchers. For the sake of performance and comparison, we also evaluated the watermarking when DWTonly and HWTonly were used. The evaluated result shows better performance using the cascaded HWT and DWT. However, in comparison with other techniques, the wavelet transformation used frequently by many researchers due to its excellent spatial localization and multi resolution characteristics, which is very close to theoretical models of the human visual system.
The based embedding in the second algorithm allows watermark image to be hidden in this image that gives a clear superiority of the algorithm. This technology has been proposed to solve the problem of illegal manipulation and distribution of digital image. Therefore, DWT and HWT technique is used in our proposed system and because it is more robust against transmission and decoding errors, it is computationally efficient and can be implemented by using simple filter convolution.
DWTHWT are compared with HWT, both methods are the digital watermarking techniques
coming from Transform Domain category; therefore, they have some features in
common. However, they have some variant characteristics. Performance comparison
between DWTHWT and HWT is summarized in the next paragraphs. Moreover, we used
in this comparison the PSNR (Peak SignaltoNoise Ratio), to show the effects
of the attacked images and how it robustness against different types Imperceptibility
means that the perceived quality of the host image should not be distorted by
the presence of the watermark. We evaluated imperceptibility of the cascaded
HWTDWT algorithm by measuring PSNR. The PSNR in decibels (dB) is given below
in Eq. 8:
Here, MAX1 is the maximum possible pixel value of the image. When the pixels
are represented using 8 bits per sample, this is 255. More generally, when samples
are represented using linear with B bits per sample, MAX1 is 2B1. It is most
easily defined via the Mean Squared Error (MSE) which for two m*n monochrome
images I and K where, one of the images is considered a noisy approximation
of the other. The MSE is given in Eq. 9:

Fig. 31: 
HWT image without attack. PSNR value = 16.7838 db 

Fig. 32: 
HWT image after Invert and Noise attack. PSNR value = 12.5846
db 
Figure 3137 show the effect of different
types of attacks on the images that watermarked using HWTDWT and HWT. The results
are generated by using the PORCUPINE software, Version 1.2.2.
Therefore, from the above examples we can show that the imperceptibility of the proposed system has been improved by using two powerful techniques HWT and DWT. For the sake of performance and comparison, we also evaluated the watermarking when HWTonly were used. The evaluated results show better performance using the cascaded HWT and DWT.
Furthermore, comparing our proposed method (HWTDWT) with (DWTDCT) (AlHaj,
2007) exploits strength of two common frequency domains method;HWT and DWT,
to obtain higher efficiency and performance.

Fig. 33: 
HWTDWT image without attack PSNR value = 37.52231 db. PSNR
value = 28.3914 db 

Fig. 34: 
HWTDWT images After Invert and Noise attac 

Fig. 35: 
HWTDWT image without attack. PSNR value = 49.3914 db

The quality of each watermarks, which is extracted by exploiting the proposed
method, is superior to that of why DWTDCT method. In the following several
watermarking attacks, including inverting and noising, are simulated to investigate
the robustness of our watermarking method.

Fig. 36: 
HWTDWT images After Invert. PSNR value = 38.5894 db 

Fig. 37: 
HWTDWT image without attack. PSNR value = 50.3914 db 

Fig. 38: 
HWTDWT images after Noise. PSNR value = 39.2714 db 
Table 2: 
Comparing PSNR of AlHaj’s with the proposed method 

The experimental results for the cases of attacks related to Fig.
36 and 38 are shown it Table 2.
CONCLUSIONS
It is concluded from this study relating the proposed system that represents a new method (HWTDWT) constructed by cascading two different but complementary techniques for image protection by using watermarking technique. Such technique is considered one of the powerful and robust schemes in protection process. Wavelet transformation (HWTDWT) provides robust resistance to the protected image against manipulation and forgery attacks. Adding a private key (password) to the watermarking will increase the privacy and security, but by embedding watermark along with adding the private key (password) more protection in wavelet transform will result, leading to more resistance against attacks. A new technique has been proposed to solve the problem of illegal manipulation and distribution of digital image, i.e., HWT and DWT system. By comparing the response of this system with that of the DWTDCT, it (HWTDWT) shows the imperceptibility of our proposed method in illustrating through comparing different types of attacks on the images and the relevant generated values (PSNR).