IRJET- Image based Information Retrieval

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 06 Issue: 05 | May-2019 www.irjet.net p-ISSN: 2395-0072
© 2019, IRJET | Impact Factor value: 7.211 | ISO 9001:2008 Certified Journal | Page 6011
Image Based Information Retrieval
Pooja Basak1, Prachi Dhankar2, Saurav Gupta3, Palak Verma4
1,2,3,4 Student, Department of Computer Science & Engineering, Inderprastha Engineering College, U.P, INDIA
-------------------------------------------------------------------------***-------------------------------------------------------------------------
Abstract - The use of digital images have drastically
increased. There is a high requirement of a system which
manipulates these images and delivers relevant information.
This paper focuses on the significance of monument
recognition. The use of Content Based Image Retrieval is
inculcated to match the similar features of the images. Such
a system can find its use in tourism applications where there
is a need to find the monument‘s information using the
image of the monument. Images can be technically divided
into classes based on its features. The prominent features of
an image are color, texture and shape. This research paper
uses an algorithm which is based on shape feature
extraction. Few image samples are taken and their region
wise similarity is mathematically compared.
Keywords: Digital images, recognition, similar
features, Content Based Image Retrieval, tourism
1. INTRODUCTION
Image processing is a process of manipulating the images
for different purposes such as image quality improvement,
noise reduction, object detection, object recognition and
content based image retrieval. With growth in
digitalization the use of digital images have grown rapidly
in all the fields such as in forensic investigation, medical
image analysis. There is yet another field which can find
image processing a useful tool for recognition and
information retrieval purposes. This research paper uses
Content Based Image Retrieval (CBIR) as a core concept
for similar feature matching.
1.1 Querying with the image
There is an immense scope of such systems because
textual annotation of every image manually is not possible
so a system which works with an image as input is
necessary. This can be bought in use when querying is
specifically through image. There may arise a situation
when recognition is possible only with the visual contents
of the image and there is no associated text with it in such
cases querying by image finds its use. The most common
example of this is finger print recognition system where
identity of the person is revealed by just the scanned
images of the fingers. Different image processing systems
consider different aspects of images for purposeful
computations. Images are stored for comparison or
computations in an image data base which have fields
according to the application architecture i.e. what many
aspects of image and related information to the image is to
be stored for manipulation.
1.2 Content Based Image Retrieval
CBIR is also called as Query By Image Content (QBIC). The
features of the query image are manipulated and
compared for similarity matching with the already stored
image
“Content-based" means that the search will analyze the
actual contents of the image. Various visible features
which make an image different are color, texture and
shape.
Steps for CBIR include:
1) Query image: It is the image which is given as input for
similarity matching
2) Image database: It consists of n number of images
depending on the choice of the user.
3) Feature extraction: It consists of visual information
(feature vector) gathering of the image and saves these
features along with the image in the image database.
4) Image matching: These feature vectors are used to
compare the query image and the other images
The feature vectors stored in database are compared with
the feature vectors of the query image
5) Resultant retrieved images: It returns the similar
images and related information based on the closeness of
the feature vector values from the image database.
1.3 CBIR Techniques
1.3.1 Color retrieval: Each image has a certain
combination of color in it. Color becomes one of
the intuitive feature of the image and histograms
are used to describe the colors. A color histogram
shows the proportion of pixels of each color
within an image. This calculated color histogram
is stored in the database. A color histogram is the
distribution of colors in an image.
1.3.2 Texture retrieval: Visual patterns with
properties of homogeneity. Contrast, density and
uniformity makes up texture of an image. Classes
of texture descriptors include statistical model
based and transform based.

1.3.3 Shape retrieval: Shape is a well-defined feature
which mostly does not depends on the size of the
object. Shape of the object is taken into
consideration for similarity matching between
images.
2. LITERATURE REVIEW
We analyzed various research papers that described the
new trends of CBIR techniques by some authors in various
domain of computer vision and its applications
2.1 A. Khokhar” Content-based Image Retrieval: Feature
Extraction Techniques and Applications” In this paper
they have discussed the difficulties faced by researchers
when text based image retrieval systems are used.
The first problem discussed is the size of the database
manual annotation of such large database is a very tedious
process the second problem is that description of images
could be highly subjective that can generate different text
labels for the same particular image. These problems have
led to the development of systems which retrieve images
and related information based on the automatically
derived features which covers features such as shape,
color and texture and such system is called Content Based
Image Retrieval.
2.2 Zobeir Raisi, Farahnaz Mohanna, Mehdi Rezaei
” A Journal on Content Based Information Retrieval on
Tourism Applications ” In this paper they have discussed
the techniques to retrieve images using all the three
features shape , color, texture one by one. From this paper
we have picked the study material for shape feature as our
research paper is based on shape feature computation.
Edge histogram descriptor (EHD) is bought in use. EHD
represents the distribution of five types of edges in any
local sub area of an image. The whole image is divided in
16 sub images and dominant edge is found out in each in
sub image block. The five types of edges are horizontal,
vertical, 45 degree diagonal, 135 degree diagonal and non-
directional. If the maximum value among five edge
magnitudes exceeds a particular threshold value then that
sub image block is said to have that particular edge as its
dominant edge, otherwise that block has no edge.
2.3 Ramesh K Lingadalli , N.Ramesh , ”Content
Based Image Retrieval using Color, Shape and Texture”
This paper describes feature extraction techniques for all
the three features. Color is the basic visual attribute for
humans and computer vision so color is perhaps the most
expensive of all the features. The scalable color descriptor
(SCD) is defined in the hue-saturation-value (HSV) color
space with fixed color space quantization, and uses a novel
Haar transform encoding. For texture feature extraction
Grey-Level co-occurrence matrix (GLCM) is used widely
and its proven that results obtained using GLCM are better
than any other texture discrimination algorithms. GLCM is
a combination of how often various different combinations
of pixel brightness values also called as grey levels occur in
a particular image. But this is found that this matrix is
sensitive to rotation.
3. METHODOLOGY:
For this research Content Based Information Retrieval
(CBIR) is taken as methodology. CBIR approach has the
capability of retrieving the most similar images to a given
query image from a database of images. It is the process
of searching for digital images in large database.
Content based means search will analyze the actual
content of the image.
3.1 Generic Process
Step1: Input the query image
Step2: Querying the database with the help of an image
Step3: The features of the image are extracted and
matched with the features of the images stored in the
database.
Step4: Most similar image with the query image is
retrieved as result.
In a typical CBIR system features related to visual
content such as shape, color, texture are initially
extracted from the query image and then they are
matched with the target image stored in the database.
In this paper, we have taken Shape as the feature
extraction basis i.e images will be matched basis of their
shape. The distribution of edges is a good texture
signature that is useful for image to image matching.
For that we are using Edge Histogram Descriptor (EHD)
to capture the spatial distribution of edges.
Our idea is to take the query image of a monument as
input and then retrieve the information of that particular
monument as result.
CBIR PROCESS
Fig 1 : CBIR PROCESS

3.2 Edge Histogram Descriptor (EHD)
The MPEG-7 Edge Histogram Descriptor (EHD) is an
efficient visual feature descriptor to extract the contents
from images.
It is also used as a Shape descriptor as long as the edge
field contains the true object boundaries.
The EHD method represents the distribution of 5 types of
edges in each local area called as sub-image which are
vertical, horizontal, 45-degree diagonal, 135-degree
diagonal and non edge.
The methodology followed is given below:-
Step 1: First of all, the query image is divided into 16 non-
overlapping sub-blocks called as sub-images.
Step 2 : Each sub-block image serves basic region to
Step 2: To characterize the EHD, each sub-image serves a
basic region to generate an edge histogram which consists
of 5 bins with vertical, horizontal,45-degree-diagonal,135-
degree diagonal and non directional edge types.
Fig 3 : Types of Edges
Step 3: The histogram for each sub block image represents
the relative frequency of occurrence types of edge in the
corresponding sub-image.
Step 4 : To extract edge features, each sub-image is further
divided into four sub-blocks.
Fig 4 : Sub-blocks and their labeling
Step 5: The luminance values of the four sub-blocks are
used for the edge-detection. The mean values of the four
sub-blocks are convolved with the filter coefficients to
obtain the edge magnitude.
Step 6: For the kth (k=0,1,2,3) sub-block of the (i,j)th image-
block, the average gray level is calculated.
Step 7: The filter coefficients for vertical, horizontal, 45-
diagonal, 135-diagonal, non-directional edges are shown
as fv(k), fh(k), fd-45, fd-135, fnd(k) respectively.
Step 8:Now the respective edge magnitudes mv(i.j),
mv(i,j),mh(i,j),m(d-45)(k),m(d-135)(i,j) and mnd (i,j) for
the (i,j)th image block can be obtained as follows :-
Step 9: If the maximum value among 5 edge magnitude
obtained from (1) to (5) is greater than the threshold, then
The image-block is considered to have the corresponding
edge (6).Otherwise the image block has no edge.
Fig 2 : Original image partitioned into sub-images
Fig 5 : Filters for Edge Detection

4. EXPERIMENT/RESULT:
For a particular monument one image was taken and its
edge magnitude values were compared with the most
similarly appearing images, it gave the following edge
magnitudes mentioned in the table.
The table consists of two columns the first column
consists of 16 region wise edge magnitudes of the first
image and the second column consists of 16 region wise
edge magnitudes of a similar image( as of first image).
Each row of the table shows the edge magnitude of the
respective region number in both the images. If the two
values are numerically close to each other in a particular
row it depicts that respective region number has a similar
kind of edge. If more respective set of regions have a
similar edge present in them, it concludes that both the
images have similar shape of objects within it.
1.HUMAYUN TOMB
Image: i3 Image: i4
Image: i1 Image: i2
TABLE I : Edge Magnitude Value (i1 ,i2)
TABLE II : Edge Magnitude Value (i3,i4)
5. CONCLUSION:
We can conclude that if two images have similar
shape (close-set of edge magnitudes) of objects
in it, they are more likely to be the image of the
same object. This solves our agenda of similar
image matching for monuments. Thus shape
can be considered as a prominent image feature
for similarity matching.
2. STATUE OF LIBERTY

6. REFERENCES
[1] A. khokher and R. Talwar, “Content-based Image
Retrieval: Feature Extraction Techniques and
Applications”, International Conference on Recent
Advances and Future Trends in Information
Technology Proceedings, Published in International
Journal of Computer Applications® (IJCA), Volume
6, Issue IRAF-IT 2012.
[2] R. K. Lingadalli, N. Ramesh, ”Content Based
Image Retrieval using Color, Shape and Texture”,
International Advanced Research Journal in Science,
Engineering and Technology, Volume 2, Issue 6,
June 2015.
[3]A. N. Bhute1, B.B. Meshram, “Journal on Content
Based on Image Indexing and Retrieval”,
International Journal
Of Graphics and Image Processing, Volume 3, Issue
4, November 2013.
5.1 Future Scope
1. There is a rapid growth in the use of digital
images on the internet so there is a need to
manage them effectively
2. Specifically when the query is through image
only. It finds immense scope in criminal
identification if we only have images with us.
3. Google has introduced an image recognition
mobile app known as Google lens. It is designed
to bring up the relevant information using
visual analysis
5.2 Limitations
We do not yet have a universally acceptable
means of characterizing the human vision more
specifically in the context of image
understanding.
Hence it is not surprising to see continuing
efforts.
5.3 Applications of CBIR
 Medical image retrieval
 Forensic Investigation
 Face recognition
 Finger-Print Recognition
[4] Z. Raisi , F. Mohanna, M. Rezaei, “A Journal on
Content Based Information Retrieval on Tourism
Application”, International Journal of
Information and Communication Technology
Research,, Volume 4, No. 1, December 2011.
[5] G. A. P, Dr. J. Prem Kumar, “A Journal based
on Study of Content Based Image Retrieval
Techniques”, International journal of Advanced
Research in Computer Science and Software
Engineering, Volume 6, Issue 2, February 2016 .
[6] S. A. Taksande, Prof. A. V. Deorankar, “A
journal on Image Based Information Retrieval”,
IRJET, Volume 4, Issue 4, April 2017.
[7] V. Ramya, “a Journal based on Content Based
Image Retrieval System using Clustering with
Combined Patterns ” , International Journal of
Scientific Research in Computer Science
Engineering and Information Technology ©
2018 IJSRCSEIT, Volume 3, Issue 1, ISSN:2456-
3307

IRJET- Image based Information Retrieval

More Related Content

What's hot (20)

Similar to IRJET- Image based Information Retrieval (20)

More from IRJET Journal (20)

Recently uploaded (20)

IRJET- Image based Information Retrieval