See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/315783334

Comparison of Discrete Cosine Transforms (DCT), Discrete Fourier Transforms

(DFT), and Discrete Wavelet Transforms (DWT) in Digital Image Watermarking

Article in International Journal of Advanced Computer Science and Applications · February 2017
DOI: 10.14569/IJACSA.2017.080232

CITATIONS READS

13 2,406

1 author:

Rosa Andrie
Politeknik Negeri Malang
124 PUBLICATIONS 650 CITATIONS

SEE PROFILE

All content following this page was uploaded by Rosa Andrie on 05 April 2017.

The user has requested enhancement of the downloaded file.

 (IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 8, No. 2, 2017

Comparison of Discrete Cosine Transforms (DCT),

Discrete Fourier Transforms (DFT), and Discrete
Wavelet Transforms (DWT) in Digital Image
Watermarking
Rosa A Asmara Reza Agustina Hidayatulloh
Information Technology Department Information Technology Department Information Technology Department
State Polytechnics of Malang State Polytechnics of Malang State Polytechnics of Malang
Malang, Indonesia Malang, Indonesia Malang, Indonesia

Abstract—Digital Image Watermarking is used recently to method. It is also proved that even the image is printed and
secure the image by embedding another digital image. It is scanned to return the format in digital, the watermark image
typically used to identify ownership of the copyright of the signal. can also be extracted smoothly.
Frequency domain transformation methods used widely in
Digital Image Compression and Digital Image Watermarking. Discrete Cosine Transform (DCT) are popular among
They reduce the weakness of classics digital image watermarking science and engineering application, from image
such as Least Significant Bit (LSB) methods which is more noise- watermarking, steganography, and lossy compression for audio
tolerant. Popular transformation method used are Two and image. Cosine function is used rather than sine function
Dimensional Discrete Cosine Transform (2D DCT), Two due to the critical for compression, fewer cosine functions are
Dimensional Discrete Fourier Transforms (2D DFT), and Two needed to approximate the typical signal. Cosine functions also
Dimensional Discrete Wavelet Transform (2D DWT). This paper express a particular choice of boundary condition in differential
will show the comparison result of those three transformation equations. DCT is similar to the Discrete Fourier Transform
method. The experiments are comparison analysis of image (DFT), but using only real numbers. DCT are equivalent of
watermark quality using Peak Signal to Noise Ratio (PSNR), DFT of roughly twice the length, operating on real data with
color converting, image resizing, image optical scanning and the even symmetry and in some variants the input or output data
noise-tolerant of the image watermarked by giving Gaussian are shifted by half a sample.
noise.
The remainder of this paper is organized as follows:
Keywords—Digital Image Watermarking; 2D Discrete Cosine Section 2 explained the 2D DCT, 2D DFT, 2D DWT and
Transform (2D DCT); 2D Discrete Fourier Transform (2D DFT); PSNR for analyzing the watermarked image. Section 3
2D Discrete Wavelet Transform (2D DWT); Least Significant Bit describes our experiment and a technique for image
method (LSB); Digital Signal Processing watermarking in frequency domain information. Section 4
presents the experiments results. Conclusion and future work
I. INTRODUCTION
are given in the final section.
Digital Image Watermarking is used recently to secure the
image by embedding another digital image. Komatsu and II. PROPOSED METHOD
Tominaga are the first person using the term Digital
A. Digital Image Watermarking
Watermarking [1]. It is typically used to identify ownership of
the copyright of the signal. The information are embedded in Digital image watermarking is one of the steganography
image is called a “digital image watermark”. The information branches, which is a technic to hide information in a digital
where the watermark is to be embedded is called a “host media. The purpose is to protect important information [8]. The
image” [2,3]. Traditional method for Digital Image information inserted in digital image watermarking can be in
Watermarking used Least Significant Bit (LSB). Many text, image, or audio format file. There are some criteria to
researcher proposed the LSB method with some improvement create good image watermarking:
and analysis to create better digital image watermark results  Imperceptibility: Good Watermark is invisible by
[4,5,6]. The method for LSB will be explained in detail in human eye. One cannot distinguish well between
Section 2. original and watermarked image.
LSB method for digital image watermarking has a
 Robustness: Good Watermark has to be resisted with
weakness which cannot handle simple noise. Image
file manipulation such as file compression, image
watermarked also will loss the watermark information if some
noising, color converting, and image resizing.
image processing is implemented such as Image Resizing and
Image Cropping. Some Frequency Domain Transformations  Security: Watermarked file can only be detected by the
method is implemented to handle such weakness in traditional file owner or authorized ones.

245 | P a g e
www.ijacsa.thesai.org
 (IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 8, No. 2, 2017

 Recovery: Watermarked file must be converted back to 2M1 + 1. Let F(k1,k2) denote the 2D discrete Fourier Transform
original file. Main purpose of watermarking is for (2D DFT) of the image. F(k1,k2) are given by
copyrighting file owner, which can be used for file
( )
authorization. ( )
∑ ( ) ( )
B. Least Significant Bit Method [3]
Least significant bit (LSB) is very simple and less
computation cost compare to the other method of Where k1 = -M1,…,M1, k2 = -M2,…,M2, ,
transformations. Figure 1 shows the insertion method in LSB.
For every bit starting from MSB of watermark image pixel is , and the operator ∑ denotes
inserted to the LSB of some image source pixel. The results ∑ ∑ ( ) is an amplitude component, and
will be the addition of 1 if the watermark bit is 1 and same as ( )
is a phase component.
previous value if the watermark bit is 0. Human eyes will be
difficult to distinguishing the image source before and after The 2D Inverse Discrete Fourier (2D IDFT) of ( ) is
watermarking. However, this method faces some image given by
processing implementations such as Image Resizing and Image
Cropping due to the pixel value modification. The PSNR of the ( ) ∑ ( )
LSB 1 bit substitution is 55.8784 [4]. Where ∑ denotes ∑ ∑

E. 2D Discrete Wavelet Transform (2D DWT)

Discrete wavelet transform (DWT) represents an image as a
subset of wavelet functions using different locations and scales.
1 1 1 1 1 1 1 1 It makes some decomposition images. Any decomposition of
an image into wavelet involves a pair of waveforms: the high
frequencies corresponding to the detailed parts of an image and
1 1 0 1 0 1 1 0 the low frequencies corresponding to the smooth parts of an
image. DWT for an image as a 2-D signal can be derived from
a 1-D DWT. According to the characteristic of the DW
1 1 0 1 0 1 1 1 decomposition, an image can be decomposed to four sub-band
images through a 1-level 2-D DWT, as shown in Fig. 2. These
Fig. 1. LSB method of Digital Image Watermarking
four sub-band images in Fig. 4 can be mapped to four sub-band
C. 2D Discrete Cosine Transform (2D DCT) elements representing LL (Approximation), HL (Vertical), LH
(Horizontal), and HH (Diagonal) respectively.
DCT in image processing is first introduced by Ahmed,
Natarajan and Rao [7]. DCT is similar to DFT, but using only
real numbers. DCT turn over the image edge to make the
image transformed into other form of even function. This is
one of linear transformations in digital signal processing. 2D
DCT is defined as:
( )
( ) ( )
( ) ( )∑ ∑ ( ) [ ] [ ] [1]

The corresponding inverse discrete cosine transformation

(2D-IDCT) is defined as:
( )
( ) ( )
∑ ∑ () ( ) ( ) [ ] [ ]
[2]
The 2D-DCT can not only focusing on transform the main
information of original image into the smallest low-frequency
component, but also it can cause the image blocking effect
being the smallest, which can realize the good compromise
between the information centralizing and the computing
complication, thus it obtains the wide spreading application. Fig. 2. 1-Level Decomposition 2D DWT

D. 2D Discrete Fourier Transform (2D DFT) The discrete Wavelet Transform will decompose a given
Consider one N1 x N2 image, f(n1,n2), where we assume signal into other signal known as the approximation and detail
that the index range are n1 = -M1,…,M1 and n2 = -M2,…,M2, coefficients. A given function f(t) can be expressed through the
for mathematical simplicity, and hence N1 = 2M1 + 1 and N2 = following representation:

246 | P a g e
www.ijacsa.thesai.org
 (IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 8, No. 2, 2017

() Original RGB Image

Converting RGB to YCbCr Divide Y level to 8x8 DFT Transform in Y Level
Color Level Subblock Subblock
∑ ∑ ( ) ( ) ∑ ( ) ( )
[4] Image Watermarking in
Image to be inserted Convert to binary image DFT Low Frequency
Where: ( ) is the mother wavelet and ( ) is the scaling Subblock Image

function. a(L,K) is called the approximation coefficient at scale

Converted YCbCr
L and d(j,K) is called the detail coefficients at scale j. The watermarked image to
Combine Y Level to
become YCbCr Image
Combine Subblock to one
Y level image
IDFT Transform
approximation and detail coefficients can be expressed as: watermarked RGB

Fig. 3. Watermark image steps in frequency domain transform

( ) ∫ () ( ) [5]
√
The usage of DCT, DFT, and DWT in image watermarking
( ) ∫ () ( ) [6] start by dividing image to 8x8 pixel sub-block. These sub-
√
block will consist of 64 coefficient (1 DC-zero frequency
Based on the choice of the mother wavelet φ(t) and scaling coefficient and 63 AC coefficient in low frequency, middle
function θ(t), different families of wavelets can be constructed. frequency, and high frequency. Figure 4 shows the DCT
frequency division.
F. Mean Square Error (MSE) and Peak Signal to Noise Ratio
(PSNR)
Peak signal-to-noise ratio (PSNR) is a ratio between
maximum possible power of a signal and the power of LF
corrupting noise that affects the fidelity of its representation.
PSNR is usually expressed in terms of the logarithmic decibel
scale. The signal is an original data, and the noise is the error
from watermark system. MF
PSNR usually defined via the mean squared error (MSE).
Given a noise-free m x n image I and noisy approximation K, HF
MSE is defined as:
∑ ∑ ( ( ) ( )) [7]
The PSNR (in dB) is defined as:

( )
Fig. 4. DCT Frequency components

( ) LF represents the low frequency, MF represents middle

√ frequency, and HF represents sub-block highest frequency. In
( ) ( ) DCT transform, LF can be found in every edge and corner of
Where, MAXI is the maximum possible pixel value of the image sub-block. In DCT and DFT, watermark bit image will
image. When the pixels are represented using 8 bits per be put in LF. In DWT, watermark bit image will be put in LH
sample, this is 255. and HL. Figure 5 shows the diagram block for DCT
watermarking process. This diagram block also represents for
III. EXPERIMENTS DFT and DWT.
Figure below shows the general watermark image steps in
frequency domain transforms.

247 | P a g e
www.ijacsa.thesai.org
 (IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 8, No. 2, 2017

Start detransformed using inverse-DCT, inverse-DFT, and

RGB Image
inverse-DWT to get Y-component watermarked image.
Watermark Index
i,x - Y-component is combined again with Cb-component
and Cr-component to get YCbCr watermarked image.
Convert RGB Image to
YCbCr Image
- Last proses is YCbCr watermarked converted again to
Y Component = imgLuminance
RGB Space to get RGB watermarked.

Divide imgLuminance
IV. RESULTS
into 8x8 subBlock
We use 32x32 binary image resolutions as an image
watermark. This image will embed on image with 65536 total
Watermark image = indexWtrmrk[x] pixel amount, since 1 pixel of binary image will embed on 8x8
image sub-block. Figure 5(a) is used as an image watermark
and figure 5(b) as an original image.
X=0

For i = 0
i<subBlock amount
i++

DCT Coefficients

If amount If amount If amount

indexWtrmrk > i indexWtrmrk[x] == 255 indexWtrmrk[x] == 0
(a) (b)
DCTCoeff[7,7] + 30 DCTCoeff[7,7] - 30
Fig. 6. (a) Image Watermark, (b) Original Image

DCT Inverse x++

Table 1 shows the results of embedding image watermark
to the LF, MF, and HF of original image.
Last subBlock ?
TABLE I. PSNR RESULTS OF IMAGE WATERMARK IN DCT, DFT, AND
DWT
Convert YCbCr Watermarked
to RGB Watermarked DCT DFT
Freq.: Embed DWT PSNR
PSNR PSNR
Pixel Coordinate (dB)
(dB) (dB)
End
LF: [w0, h1] 40,24 32,2 33,78
Fig. 5. Diagram blocks of DCT image watermarking 33,87 (HL)
MF: [w3, h3] 39,96 35,18
33,87 (LH)
- User input host image, RGB image in which HF: [w7, h7] 40,82 32,18 33,87
watermark image will be inserted. Watermark image
TABLE II. PSNR RESULTS OF GAUSSIAN NOISE ATTACK IN WATERMARK
will use binary image provided. DCT, DFT, AND DWT
- RGB image will be converted to YCbCr color space, Y 20 % Gaussian Noise to the watermarked image
color space (luminance) component is the only PSNR PSNR
Frequency PSNR DWT
component where the watermark image will be DCT DFT
inserted, and thus the Y-component transformed using LF 19,36 19,04 18,90
DCT, DFT, and DWT. Y-component is used because 19,01 (HL)
MF 19,26 19,00
18,95 (LH)
human perception more sensitive to the light intensity
HF 19,29 19,11 19,15
and the result for watermarked image will
imperceptible. Chrominance blue (Cb) and TABLE III. PSNR RESULTS OF COMPRESSION ATTACK IN WATERMARK
chrominance red (Cr) component will not transform DCT, DFT, AND DWT
and will be used only for Inverse Transform.
Compression test using RIOT Application
- Y-component is divided into 8x8 sub-blocks. Level
Extraction Extraction DFT Extraction
DCT DWT
- Each 8x8 sub-blocks then transforms using DCT, DFT,
and DWT. Image watermark is a binary image with 0 25 %
and 1 pixel value. If the pixel value is 1 in the
watermark image, then the sub-block host image in
index [7, 7] will be add with 30, otherwise will be 50 %
subtract with 30. This process is then repeated until the
last watermark image pixel. The Y-component then

248 | P a g e
www.ijacsa.thesai.org
 (IJACSA) International Journal of Advanced Computer Science and Applications,
Vol. 8, No. 2, 2017

TABLE IV. PSNR RESULTS OF IMAGE CONTRAST MANIPULATION IN watermark in Low Frequency. It has 19.36 dB PSNR
WATERMARK DCT, DFT, AND DWT
value, as show in table 2.
HF Watermarked on Image Contrast Manipulation
Image after Contrast PSNR PSNR PSNR  Compressing image using RIOT application for all
Manipulation DCT DFT DWT three transformation, DWT shows the best results as
shown in table 3.
Using image contrast manipulation, DCT shows the best
27,57 21,48 20,32 results as shown in table 4.
REFERENCES
[1] Bender, W., Gruhl, D., Morimoto, N. and Lu, A(1996).: Techniques for
data hiding. IBM Systems Journal, vol. 35, nos. 3&4.
[2] Saraju Prasad Mohanty(,January 1999 )“Watermarking of Digital
Images”, Submitted at Indian Institute of Science Bangalore, pp. 1.3 –
50% Contrast Addition 1.6,.
[3] Katzenbeisser, S. and Petitcolas, F(1999).: Information hiding
V. CONCLUSION techniques for steganography and digital watermarking. Artech House
Books.
Frequency domain transformation methods are used widely
[4] Puneet Kr Sharma and Rajni(2012).: ANALYSIS OF IMAGE
in Digital Image Compression and Digital Image WATERMARKING USING LEAST SIGNIFICANT BIT
Watermarking. It reduces the weakness of classics digital ALGORITHM, International Journal of Information Sciences and
image watermarking such as Least Significant Bit (LSB) Techniques (IJIST) Vol.2, No.4, July 2012.
methods which is more noise-tolerant. Popular transformation [5] Abdullah Bamatraf, Rosziati Ibrahim and Mohd. Najib Mohd.
method used are Two Dimensional Discrete Cosine Transform Salleh(2011).: A New Digital Watermarking Algorithm Using
(2D DCT), Two Dimensional Discrete Fourier Transforms (2D Combination of Least Significant Bit (LSB) and Inverse Bit, Journal of
Computing Vol.3, Issue 4, April 2011, ISSN 2151-9617.
DFT), and Two Dimensional Discrete Wavelet Transform (2D
DWT). This paper proposed the comparison between those [6] Deepshikha Chopra, Preeti Gupta, Gaur Sanjay B.C., Anil Gupta (2012).:
Lsb Based Digital Image Watermarking For Gray Scale Image, IOSR
three transformation methods. The experiments are image Journal of Computer Engineering (IOSRJCE) Vol.6, Issue 1, pp 36-41,
watermark quality analysis using Mean Square Error (MSE), Sep-Oct 2011, ISSN 2278-0661.
Peak Signal to Noise Ratio (PSNR), and the noise-tolerant of [7] Ahmed, Natarajan, and Rao (1974) : On Image Processing and a Discrete
the image watermarked by giving Gaussian noise in it. The Cosine Transform, IEEE Trans. On Computer C-23(1): 90-93
experiments shows result of:
AUTHOR PROFILE
 DCT Transformation embed on High Frequency is the Rosa A. Asmara: Lecturer of Information Technology Department in
best for image watermarking. It has 40.82 dB PSNR State Polytechnics of Malang, Indonesia. Research interest are Image
values, as shown in table 1. Processing and Computer Vision, Pattern Recognition, and Artificial
Intelligence.
 Adding 20% Gaussian noise to the watermarked image, Reza Agustina, Hidayatulloh: Undergraduate student in Information
the best Transformation for Gaussian Attack is DCT Technology Department, State Polytechnics of Malang, Indonesia.

249 | P a g e
www.ijacsa.thesai.org
View publication stats