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

net/publication/339435251

Creating Color Image Signature Based On Laplacian Equation

Article · February 2020

DOI: 10.30630/joiv.3.2.223

CITATIONS READS
7 1,148

5 authors, including:

Bilal Zahran Ziad Alqadi

Al Balqa Applied University Al-Balqa' Applied University
35 PUBLICATIONS   772 CITATIONS    338 PUBLICATIONS   3,888 CITATIONS   

SEE PROFILE SEE PROFILE

Belal Ayyoub Muhammed Mesleh

Al-Balqa' Applied University Al-Balqa' Applied University
39 PUBLICATIONS   625 CITATIONS    4 PUBLICATIONS   98 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

A technique of hiding secrete text in wave file View project

Scientific Hypothesis as a source of innovation and discovery View project

All content following this page was uploaded by Belal Ayyoub on 22 February 2020.

The user has requested enhancement of the downloaded file.

VOL 3 (2019) NO 2

e-ISSN : 2549-9904
ISSN : 2549-9610

INTERNATIONAL JOURNAL ON INFORMATICS VISUALIZATION

Creating Color Image Signature Based On Laplacian Equation

Jamil Al-Azzeh#, Bilal Zahran#, Ziad Alqadi#, Belal Ayyoub#, Muhammed Mesleh#
#
Computer Engineering Department, Al Balqaa Applied university, Amman, Jordan
E-mail: azzehjamil@gmail.com, zahranb@bau.edu.jo, Natalia_maw@yahoo.com, belal_ayyoub@hotmail.com, meslehmuh@gmail.com

Abstract— Colour image retrieval, identification and recognition play an important role in vital computerized applications, so
creating an image signature which can be used as a key to identify the image is an essential and important problem to be solved. In
this paper a new method of image signature extraction will be proposed. The proposed method will be implemented and tested and
the experimental results will be compared with other method results to prove the efficiency of the proposed method by mean of
creating a unique signature which will require a minimum memory space and a smaller time to extract the signature.

Keywords— LBP, CSLBP, image signature, extraction time, Laplacian operator.

the correct features and select the appropriate classification

I. INTRODUCTION algorithm for these features. Different classification results
Image feature is a simple image pattern, based on which for the same image can be obtained with different feature
we can describe what we see on the image. For example, cat extraction algorithms [1-30]. Therefore, feature selection is
eye will be a feature on a image of a cat. The main role of the one of the most important step for classification. The
features in computer vision (and not only) is to transform purpose of feature extraction algorithms is to identify
visual information into the vector space. This give us features that can best represent the image and contain fewer
possibility to perform mathematical operations on them, for parameters. With the specified features, the image can be
example finding similar vector (which lead us to similar expressed meaningfully using fewer parameters. A faster and
image or object on the image). successful classification can be made with fewer
There are many algorithms for feature extraction, most computational loads by eliminating unimportant parameters
popular of them are LBP, CSLB, most of this algorithms [4]. Low-level features and high-level features are usually
based on image gradient. removed from the images. Low-level features are simpler
Today we will use KAZE descriptor, because it shipped in features in the image and computational load is less.
the base OpenCV library, while others are not, just to However, the classification success is low for complex
simplify installation. images. High-level features are more complex and have
Image processing techniques have an important role in the more computational load. The choice of which features to
interpretation of medical images and for automatic diagnosis. use varies depending on the problem. For this reason, there
Especially in recent years, the development of whole-slide are many feature extraction algorithms with different
imaging techniques and the increase in cancer cases have approaches in the literature
attracted the attention of many researchers to automatic
histopathological image analysis [1-30]. The whole-slide
images have a very high resolution and it takes quite a long
time to be examined by experts. Computer aided automatic
image processing methods are presented to facilitate this
exhaustive process. These methods help the expert to decide
on the analysis of the image, and in some cases assume the
role of decision maker [1-30]. Image features are used in the
automatic classification of images and in the decision-
making process. Many features such as texture differences,
shape differences, light fluctuations, color changes in the
image provide useful information for classification
algorithms. The most important point here is to determine Fig. 1 Color image and RGB matrices

94
 RGB color image [1, 29] is a combination of three colors,
red, green and blue, and it can be represented by 3D matrix
as shown in figure (1), one 2D matrix for each color.

Fig. 2 Color image components

Fig. 4 Color as 2D matrix
Figure (2) shows the color panels projecting of the color
components in color image are represented in following way: Feature extraction has a long history and a lot of feature
R=RGB (:,:, 1); G=RGB (:,:,2); B=RGB (:,:,3); extraction algorithms based on color, texture and shape have
As shown in Figure 3, the RGB color model assembles been proposed. Feature selection is a critical issue in image
the primary lights of Red, Green, and Blue together in analysis. In spite of various techniques available in literature,
various combinations to produce a broad range of colors. it is still hard to tell which feature is necessary and sufficient
Red and Green lights are combined together to pro-duce to result in a high performance system. Color is the first and
Yellow light. The RGB color model is termed as an additive most straightforward visual feature for indexing and retrieval
color model in which the combination of the Red, Green, of images. The first order (mean), the second order (variance)
and Blue primary lights produces White light. The RGB and the third order (skewness) color moments have been
color model is used in various tech-neologies producing proved to be efficient in representing color distribution of
color images, such as conventional photography and the images [2]. An approach that lies between subdividing the
display of images in electronic systems. Examples of the images and relying on fully segmented images was proposed
RGB input devices include image scanners, video games, by Stricker and Dimai [3]. They worked with 5 partially
and digital cameras as well color model is an additive color overlapping, fuzzy regions. The texture is very important cue
model for displays. The Cyan, Magenta, Yellow, and Key in region based segmentation of images. Texture features
Black (CMYK) color model is a subtractive color model for play a very important role in computer vision and pattern
printing. The Red, Yellow, and Blue (RYB) color model is recognition [4]. Texture analysis has a long history and
designed for mixing colors with paints. texture analysis algorithms range from using random field
models to multiresolution filtering techniques such as the
wavelet transform [5]. Due to resemblance between
multi-resolution filtering techniques and human visual
process, Gabor and Wavelet Transform techniques are often
used for texture characterization through the analysis of
spatial frequency content [6].The first two approaches have
been explored more thoroughly than shape based approaches.
Shape representation and description is a difficult task. This
is because when a 3-D real world object is projected onto a
2-D image plane, one dimension of the object Information is
lost[7].
Image recognition systems, and image retrieval systems
[3, 4], searching, browsing and retrieving color image from
a database of images requires a key or image signature
which basically formed from some features extracted by a
certain method and saved to be used later as an image key
Fig. 3 Visual summary of color models. for image retrieval or recognition.
The Red, Green, and Blue (RGB) Color image features extraction has received a lot of
attention in recent years. The basic idea is to first detect
It is very easy to extract each color matrix, and here we interest regions that are covariant to a class of
can manipulate each color as a grey image as shown in transformations [5, 6, and 7]. The methods of feature
figure (4): extraction are based on calculating local binary pattern
(LBP), for each pixel, these calculations as shown in figure

95
 (5) formed an array of 256 elements, each element value figure (9) show an example of how calculate CSLBP
points to the number of repetition of color value in the image, operator for a pixel
and here this array can be used as a key or a signature to 100 105 200 0 1 1
retrieve or recognize the image.
40 104 110 1
70 80 215
Binary = 0011
Decimal=3
So add 1 to the repletion of the value 3

Fig. 9 CSLBP calculate operator for example

II. THE PROPOSED METHOD

The proposed method we consider involves the Laplacian
operator,[9. 10, and 11]

(eq. 1)
Approximating the partial derivatives with centered
second differences gives the 5-point discrete Laplacian as
shown in figure (10)
Fig. 5 LBP calculations

figure (6) show an example of how calculate LBP operator for

a pixel

100 105 200 0 1 1

40 104 110 0 1 (eq.2)
70 80 215 0 0 1
Binary = 0000111
Decimal=15
So add 1 to the repletion of the value 15

Fig. 6 LBP calculate operator for example

Center-symmetric local binary pattern (CSLBP) is also an

efficient method of image features extraction, and features
have proven to be robust against illumination changes, they
are very fast to compute, and do not require many
parameters to be set [8]. CSLBP computes a binary pattern
for each pixel and it reduces the feature array to 16 elements,
thus will reduce the memory space to store the key, it also
Fig. 10 5-point discrete Laplacian operator
reduces the time needed for calculations required to generate
the feature array, CSLBP calculations are similar to LBP To take the benefits of Laplacian operator we have to
calculations as shown in figure (7).

calculate the first and the second operators

using eq.2.
Based on Laplacian operators we can use the proposed
method of image features extraction and it can be
implemented applying the following steps:
1) Get the color image.
2) Initialize the 4 elements features array (FA) to zeros.
3) Reshape 3D color matrix to 2D matrix.
4) For each pixel in the image do the following:

a) A0=
Fig. 8 LBP and CSLBP calculations.

96
 9 432000 64466 50569 45900 268249 0.0995
10 151353 31273 22450 20959 74967 0.0376
b) A1=
Average 940510 0.2201
c) If A0>=0 let c0=1 else let c0=0.
d) If A1>=0 let c1=1 else let c1=0.
e) R=c0+c1*2. The image signature is very sensitive to any changes in the
f) FA(R+1)= FA(R+1) +1. image, so we can consider a signature as a unique value and
5) Save FA. it can be used for image retrieval or image recognition.
Table (2) shows that even if we change on pixel in the
III. IMPLEMENTATION AND EXPERIMENTAL RESULTS original image the signature will be affected and change
The following matlab code was written to implement the accordingly.
proposed method which uses Laplacian operator:
TABLE 2
CHANGES IN THE ORIGINAL IMAGE WILL BE REFLECTED IN IMAGE
SIGNATURE

Original Signature Signature Signature

image after after after
signature changing 1 changing 1 changing 1
pixel pixel pixel
55061 55062 55061 55061
45104 45104 45106 45106
37108 37107 37109 37109
131397 131397 131394 131394

- CSLBP method generates an image signature with 16

values, thus it needs more memory space and more
extraction time to deal with the image signature, the
following matlab code was written to implement
CSLBP method of image signature extraction. From
the results shown in table 3 we see that each image
signature requires 128(16*8) bytes of memory space,
Different color images with different types and sizes were while the proposed method signature requires 32 bytes
used and for each image the signature and extraction time (4*8).
were calculated.
Table (1) shows the implementation results.
From the obtained results shown in table (1) we can raise
the following facts:
- The number of elements in the features array (image
signature) was reduced to 4, thus the memory space
which will be used to store the images signatures will
be also reduced.
- The proposed method can be used for any type of
images and for images with any size.
- The signature extraction time is small and it is around
0.2201 seconds in average.
TABLE 1
SIGNATURES AND EXTRACTION TIMES OBTAINED BY IMPLEMENTING THE
PROPOSED METHOD

Extraction
Image Size Signature time(seco
# {Pixels) nds)
1 270948 55061 45104 37108 131397 0.0635
2 151875 18640 23584 15027 92828 0.0397
3 49152 7121 6270 5699 29042 0.0153
4 1125600 165902 145839 131100 678349 0.2663
5 540000 75906 67073 60331 333694 0.1297
6 3396069 395295 301166 259191 2432369 0.7789
7 2359296 201452 153677 129269 1868246 0.5594
8 928800 247859 175997 162563 338089 0.2108

97
 Table (3) shows some samples of images signatures REFERENCES
obtained by implementing CSLBP method.
- The proposed method is more efficient comparing with [1] Akram A. Moustafa, Ziad A. Alqadi Color Image Reconstruction
CSLBP method and the experimental results shown in Using A New R'G'I Model Journal of Computer Science 5 (4): 250-
table (4) show that the average signature extraction 254, 2009 Science Publications.Dr. Rushdi S. Abu Zneit,Dr. Ziad
AlQadi,Dr. Mohammad Abu Zalata, A Methodology to Create a
time for the proposed method was reduces to 1.1790 Fingerprint for RGB Color Image, IJCSMC, Vol. 6 , Issue. 1, January
times comparing with CSLBP method 2017, pg.205–212.
[2] Modified Least Significant Bit (MLSB),MA Zaher, Computer and
TABLE 3 Information Science 4 (1), 60, 2010
COLOR IMAGES SIGNATURES USING CSLBP METHOD [3] 25. Jamil Al -Azzeh,S. F. Yatsun, A.A. Cherepanov, I.V. Lupehina4
and V.S. Dichenko; Computer simulation of vibration robot created
Signat Signat Signat Signat Signat for the wall movement; Research Journal of Applied Sciences.; 2014 ,
ure of ure of ure of ure of ure of Issue: 9, Page No.: 597-602 T. Ahonen, A. Hadid, M. Pietikainen,
“Face description with local binary patterns: application to face
image 1 image 2 image 3 image 4 image 5 recognition” IEEE Transactions on Pattern Analysis and Machine
44920 28694 9089 211947 94738 Intelligence 28 (12) (2006) 2037–2041.
7301 3808 1884 70998 23013 [4] Jamil S. AL-Azzeh: Distributed Mutual Inter-Unit Test Method For
D-Dimensional Mesh-Connected Multiprocessors With Round-Robin
2474 839 911 35009 14684 Collision Resolution: Jordanian Journal of Computers and
7897 3677 1814 59995 24361 Information Technology April 2019.
4395 2670 1421 30977 15849 [5] Jamil Al-Azzeh, Bilal Zahran, Ziad Alqadi, Belal Ayyoub,
1441 652 639 17522 7167 Muhammed Mesleh : A Novel Based On Image Blocking Method To
Encrypt-Decrypt Color JOIV: International Journal on Informatics
4728 2210 1377 27596 14504 Visualization, 2019
29455 19944 5932 98374 51057 [6] Jamil S. Al Azzeh, Abdelwadood Mesleh ,Sergiy Gnatyuk and
40941 33681 5747 102697 53709 Anastasiia Abakumova, Evaluation Method for SDN Network
Effectiveness in Next Generation Cellular Networks : International
6346 3505 1352 31735 15473 Journal of Communication Networks and Information Security
1982 819 671 17718 7132 December 2018.
6284 3638 1626 36351 18119 [7] Jamil S. AL-Azzeh: Improved testability method for mesh-connected
VLSI multiprocessors: Jordanian Journal of Computers and
9594 3643 1802 56873 23831 Information Technology August 2018.
3201 934 951 34502 14022 [8] Jamil Al-Azzeh, Bilal Zahran, Ziad Alqadi, Belal Ayyoub And
10099 3937 1967 66963 25675 Mazen Abu-Zaher: A Novel Zero-Error Method to Create a Secret
Tag for an Image; Journal of Theoretical and Applied Information
87612 37428 10949 221933 133670
Technology 15th July 2018.
[9] Jamil Al-Azzeh, Bilal Zahran and Ziad Alqadi: Salt and Pepper
TABLE 4 Noise: Effects and Removal; International Journal on Informatics
EXTRACTION TIME COMPARISONS Visualization July 2018.
[10] Jamil AL-Azzeh, Oleksandr Kovalenko , Oleksii Smirnov Anna
Image # Size Extraction Extraction Speed Kovalenko , Serhii Smirnov : Qualitative risk analysis of software
{Pixels) time(seconds time(seconds up of development ; Asian Journal of Information Technology July 2018.
[11] Bilal Zahran , Jamil Al-Azzeh ,Ziad Alqadi, Mohd-Ashraf Al
) ) Using Zoghoul : A Modified Lbp Method To Extract Features From Color
Using Using Laplace Images : Journal of Theoretical and Applied Information Technology
Laplace CSLBP operator May 2018.
operator method [12] Jamil AL-Azzeh, Information Technologies for Supporting
Administrative Activities of Large Organizations; DESIDOC Journal
1 270948 0.0635 0.0769 1.2110 of Library & Information Technology, Vol. 38, No. 3, May 2018.
2 151875 0.0397 0.0446 1.1234 [13] Jamil S. AL-Azzeh: A Distributed Multiplexed Mutual Inter-Unit in-
3 49152 0.0153 0.0174 1.1373 Operation Test Method for Mesh-Connected VLSI Multiprocessors;
Jordan Journal of Electrical Engineering; 2017 Volume 10, Number
4 1125600 0.2663 0.3094 1.1618 5.
5 540000 0.1297 0.1526 1.1766 [14] Jamil S. AL-Azzeh: Fault-Tolerant Routing in Mesh-Connected
6 3396069 0.7789 0.9292 1.1930 Multicomputer based on Majority-Operator-Produced Transfer
7 2359296 0.5594 0.6471 1.1568 Direction Identifiers; Jordan Journal of Electrical Engineering
Volume 3, Number 2, April 2017.
8 928800 0.2108 0.2537 1.2035 [15] Jamil S. AL-Azzeh, Mazin Al Hadidi, R. Odarchenko,S. Gnatyuk, Z.
9 432000 0.0995 0.1198 1.2040 Shevchuk :Analysis of Self-Similar Traffic Models in Computer
10 151353 0.0376 0.0444 1.1809 Networks; International Review on Modelling and Simulations;
October 2017 Volume 10, Number 5.
Average 940510 0.2201 0.2595 1.1790 [16] Jamil Al Azzeh, Ziad Alqadi Qazem, M. Jabber: Statistical Analysis
of Methods Used to Enhanced Color Image Histogram; XX
IV. CONCLUSIONS International Scientific and Technical Conference; Russia May 24-26,
2017.
An efficient method of color image signature (features) [17] Mazen Abuzaher, Jamil AL-Azzeh: JPEG Based Compression
was proposed, implemented and tested. The obtained Algorithm; International Journal of Engineering and Applied
experimental results showed that the proposed method Sciences Volume 4, Number 4, 2017
[18] Mazin al hadidi, Jamil s. Al-azzeh, oleg p. Tkalich,roman s.
reduces the memory space needed to store the signature, Odarchenko,sergiy o. Gnatyuk and yulia ye. Khokhlachova2: Zigbee,
reduces the time needed to generate the signature, creates a Bluetooth and Wi-Fi Complex Wireless Networks Performance
unique signature which can be used as a key to identify the Increasing; International Journal On Communications Antenna And
Propagation, vol 7 No 1
image, no matter what the image size and what the image
[19] Jamil Al Azzeh, Daniel Monday Afodigbokwu, Denis Olegovich
type. Bobyntsev, Igor Valerievich Zotov: Implementing Built-In Test in

98
 Analog and Mixed-Signal Embedded-Core-Based System-On-Chips; Computers and Software (I.RE.CO.S.), Vol. 11, N. 2 ISSN 1828-
Asian Journal of Information Technology, Medwell Journals ,2016. 6003 (impact factor = 6.14). February 2016.
(SJR indicator = 0.11). [25] Jamil Al Azzeh, Bidirectional Virtual Bit-slice Synchronizer: A
[20] Jamil Al Azzeh, Hussein Alhatamleh, Ziad A. Alqadi, Mohammad Scalable Solution for Hardware-level Barrier Synchronization.
Khalil Abuzalata : Creating a Color Map to be used to Convert a Research Journal of Applied Sciences, Engineering and Technology,
Gray Image to Color Image; International Journal of Computer 11(8): 902-909. Maxwell Scientific Publication Corp November
Applications (0975 - 8887).Volume 153 - No2, November 2016. 2015.
[21] Jamil Al-Azzeh: Analysis of Second Order Differential Equation [26] Jamil Al Azzeh, Michael E. Leonov. Dniitriy E. Skopm. Evgeny A.
Coefficients Effects on PID Parameters International Journal on Titenko, Isor V Zotov; The Organization of Built-in Hardware-Level
Numerical and Analytical Methods in Engineering (IRENA) Vol 4, Mutual Self-Test in Mesh-Connected VLSI Multiprocessors;
No 2 2016. International Journal on Information Technology (I.RE.I.T.) Vol. 3,
[22] Dmitriy Skopin and Jamil Al-Azzeh; Automated Demodulation of Praise Worthy Prize, March 2015.
Amplitude Modulated Multichannel Signals with Unknown [27] Jamil Al Azzeh, Dmitriy B. Borzov2, Igor V. Zotov3 and Dmitriy E.
Parameters Using 3D Spectrum Representation Research Journal of Skopin'; an approach to achieving increased fault-tolerance and
Applied Sciences, Engineering and Technology, Maxwell Scientific availability of multiprocessor-based computer systems" ; Australian
Publication June 05, 2016 Journal of Basic and Applied Sciences. Apr. 2014
[23] Mazin Al Hadidi, Jamil S. Al-Azzeh, R. Odarchenko, Sergiy [28] Jamil Al -Azzeh,S. F. Yatsun, A.A. Cherepanov, I.V. Lupehina4 and
Gnatyu,k and A. A bakumova Adaptive Regulation of Radiated V.S. Dichenko; Computer simulation of vibration robot created for
Power Radio Transmitting Devices in Modern Cellular Network the wall movement; Research Journal of Applied Sciences.; 2014 ,
Depending on Climatic Conditions. Contemporary Engineering Issue: 9, Page No.: 597-602.
Sciences, Vol. 9, 2016, [29] AL-Azzeh Jamil, Review of Methods of Distributed Barrier
[24] Mazin Al Hadidi, Jamil S. Al-Azzeh, B. Akhmetov, O. Korchenko,S. Synchronization of Parallel Processes in Matrix VLSI Systems,
Kazmirchuk, M. Zhekambayeva: Methods of Risk Assessment for International Review on Computers and Software (IRECOS), Praise
Information Security Management International Review on Worthy Prize, Part A, vol. 8, no. 4, pp.42- 46, April 2013.

99

View publication stats