International Journal of Engineering and Science Invention (IJESI) is an international journal intended for professionals and researchers in all fields of computer science and electronics. IJESI publishes research articles and reviews within the whole field Engineering Science and Technology, new teaching methods, assessment, validation and the impact of new technologies and it will continue to provide information on the latest trends and developments in this ever-expanding subject. The publications of papers are selected through double peer reviewed to ensure originality, relevance, and readability. The articles published in our journal can be accessed online

1. International Journal of Engineering Science Invention
ISSN (Online): 2319 – 6734, ISSN (Print): 2319 – 6726
www.ijesi.org Volume 2 Issue 10ǁ October 2013 ǁ PP.71-74
Generate Attribute-Enhanced Sparse Codewords To Retrieve
Image From Large Image Database
1,
R.Sureshkumar, Me., Ap/Cse , 2,N.Arthi, Me Cse
1,2,
Sembodai Rukmani Varatharajan Engineering College, Sembodai
ABSTRACT:True multimedia is the combination of different elements (whether medium, modality, technology,
algorithm, or application) that provides a fuller experience of the effect of that combination. In this paper we
can retrieve image from the large scale image database. In content based image retrieval we can only retrieve
the image using low level features. But in this paper we use attribute-enhanced sparse coding and attribute
embedded inverted indexing to retrieve the image using high level features (hair style, gender, race).These
attributes are automatically detected by the above two algorithms.
INDEX TERMS: content based image retrieval, attribute enhanced sparse coding, and attribute embedded
inverted indexing, high level features.
I.
INTRODUCTION
Now a days the popularity of social networks like face book, twitter are mostly used by the people.
Many of them use human face images to their profile. Maximum of the user use the celebrities image. Due to
that more than two persons have the similar image. And also we can maintain large scale database for the image
storage. In Photo search by face positions and facial attributes on touch devices (1) the images attribute is
represented in the outline. They use common attribute outline image to all the images. For example use kid cap
for the retrieval of children’s image. And also have the common attribute for female and male images.In
retrieval-based face annotation (RBFA) (2) technique we can two challenges. The first challenge is how to
efficiently retrieve a short list of most similar facial images from a large facial image database, which typically
relies on an effective content-based facial image retrieval solution.The second challenge is how to effectively
exploit the short list of candidate facial images and their weak label information for the face name annotation
task, which is critical. The associated labels of web facial images are often noisy and incomplete due to the
nature of web images Our main goal of system is to retrieve the image from large scale content based image
retrieval. In the existing system we use the content based image retrieval (CBIR). In that we cannot
automatically detect the human attributes. Give an image as input to retrieve the image to find the similar image
from the large scale image database.Image retrieval using query based image retrieval (QBIR) technique to
display the similar images. Before store the image in the database we can give an index number to images. By
using the index no the features are to be extracted and stored in the features database.
It is an important
technique in many applications like automatic face annotation, crime investigation etc. Traditional image
retrieval using low level features to retrieve the image (3) (4) (5). (eg. expression, posing). But it makes some
lack of semantic meaning. In LBP (6) the images are scanned into various effects. Image can be splited into
various types then the binary value assigned to that image to analysis the image character to get the perfect
image from the large database. But it has some disadvantage. Our proposed system overcomes this by using the
high level attributes of the human.
www.ijesi.org
71 | Page

2. Generate Attribute-Enhanced Sparse Codewords…
Fig. 1. It shows that image retrieved from the large scale database. A single image given as input to that
system and then extract the features from the image. Match the features with the features database and
display the retrieved images.
Human attributes are high level semantic description about the person. Figure 2 shows the
segmentation image in attributes. This similar to the bag of words concept (BoW) (7). The recent work shows
the adequate quality of the human attributes. This will be used in many applications like face verification (8),
face identification (9), keyword based image retrieval (10) etc.
Fig.2. In this the image segmented in to many attributes (background, pants, hair, skin,shirt). The segmented
image shown in the right side.
II.
PROPOSED SYSTEM
In this paper we can automatically detect the human attributes using two orthogonal methods named
attribute-enhanced sparse coding and attribute embedded inverted indexing. Attribute-enhanced sparse coding
exploits the global structure of feature space and uses several important human attributes combined with lowlevel features to construct semantic codeword in the offline stage. On the other hand, attribute-embedded
inverted indexing locally considers human attributes of the designated query image in a binary signature and
provides efficient retrieval in the online stage. By incorporating these two methods, we build a large-scale
content-based face image retrieval system by taking advantages of both low level (appearance) features and
high-level (facial) semantics. By using this method we can improve the image retrieval up to 43.55%. We
combine the low level features and high level human attributes to construct the sparse coding.Using the
automatically detected human attributes we can achieve excellent performance in keyword based image
retrieval. The local binary pattern is used to segment the image into many parts. But it use textual descriptions
the segmented image are assigned as 1s and 0s.Finally we can retrieve the list of image from the large scale
image database. Viola – Jones face detection method used to fix the landmark on the face. We can obtain the
175 grid points including many facial high level human attributes. The attribute enhanced sparse coding method
detects the high level attributes to retrieve the image.
www.ijesi.org
72 | Page

3. Generate Attribute-Enhanced Sparse Codewords…
Attribute enhanced sparse coding (ASC):
It describes the automatic detection of human attribute from the image and also creates the different
sparse coding. These collections of sparse coding represent the original image.
Attribute embedded inverted indexing:
It collects the sparse code words from the attribute enhanced sparse coding and check the code words
with the online feature database and retrieve the related images similar the query image.
III.
RELATED WORK
Whenever we give an image as input the image formatted without the background images only get the
facial position of the image. And split the image into many grid points. From each point we can obtain the
landmarks of the face expression. By using the landmarks the patches of the image is get through the attribute
enhanced sparse code words.These code words are created in the offline storage. And then the sparse code
words are compared by the online code words and give the rank list of images similar to that given query image.
Finally we can get large amount of similar human face images. If we store the images in the database it
automatically generate the sparse codeword and then store he image. If the same image retrieved from the
database means it compare the features database with the original sparse codeword. All the codewords of the
image should be stored in the large scale image database. The human attributes like race, hair color, face
position, eye length, simile level are have the individual sparse code words the attribute embedded inverted
indexing method compare each and every code word with the individual sparse code word. Every related code
word images are retrieved and displayed to the user.The related images are displayed separately on the window.
They are displayed by dark outlined images. The images which are not displayed in the dark outline are not
related to the query image. Fig.1. Clearly describes the work of attribute enhanced sparse codeword method.
Architecture diagram:
Fig 3 image retrieval from video file
In this diagram the image retrieval of video source. The source is converted into video frames. Then
the point tracker assign the pointer to every video frame and then given to the model detector finally get the high
level description of image. And then the description stored in the database.Fig 3 shows the neat dataflow
diagram of the proposed system. The images are retrieved from the large image database. The images are
converted into local patches and the features are turn into sparse coding. The embedded inverted indexing
indicates the attribute embedded inverted indexing. It should be performed in the online storage.
www.ijesi.org
73 | Page

4. Generate Attribute-Enhanced Sparse Codewords…
Dataflow diagram:
Fig 4 dataflow in generation of attribute enhanced sparse codewords
IV.
CONCLUSION
In our proposed system we can get the high performance on the image retrieval in large scale image database. In
the existing system we cannot use the human attributes only use the low level features of the human images. But
in the proposed system we use the high level attribute. It increases the effectiveness of the image.
Attribute enhanced sparse codewords retrieve less number of images due to that we can get only the related
images. From that we can obtain the main image from the large image database.
V.
ACKNOWLEDGEMENT
I would like to thank Mr.SureshKumar for give excellent insights and suggestions.
REFERENCE
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
Y.-H. Lei, Y.-Y. Chen, L. Iida, B.-C. hen, H.-H. Su, and W. H. Hsu,“Photo search by face positions and facial attributes on touch
devices,” ACM Multimedia, 2011.
D. Wang, S. C. Hoi, Y. He, and J. Zhu, “Retrieval-based face annotation by weak label regularized local coordinate coding,” ACM
Multimedia, 2011.
Z. Wu, Q. Ke, J. Sun, and H.-Y. Shum, “Scalable face image retrieval with identity-based quantization and multi-reference reranking,” IEEE Conference on Computer Vision and Pattern Recognition, 2010.
B.-C. Chen, Y.-H. Kuo, Y.-Y. Chen, K.-Y. Chu, and W. Hsu, “Semisupervised face image retrieval using sparse coding with
identity constraint,” ACM Multimedia, 2011.
D. Wang, S. C. Hoi, Y. He, and J. Zhu, “Retrieval-based face annotation by weak label regularized local coordinate coding,” ACM
Multimedia, 2011.
T. Ahonen, A. Hadid, and M. Pietikainen, “Face recognition with local binary patterns,” European Conference on Computer Vision,
2004.
L. Wu, S. C. H. Hoi, and N. Yu, “Semantics-preserving bag-of-words models and applications,” Journal of IEEE Transactions on
image processing, 2010.
N. Kumar, A. C. Berg, P. N. Belhumeur, and S. K. Nayar, “Describable visual attributes for face verification and image search,” in
IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), Special Issue on Real-World Face Recognition, Oct
2011.
W. Scheirer, N. Kumar, K. Ricanek, T. E. Boult, and P. N. Belhumeur, “Fusing with context: a bayesian approach to combining
descriptive attributes,” International Joint Conference on Biometrics, 2011.
B. Siddiquie, R. S. Feris, and L. S. Davis, “Image ranking and retrieval based on multi-attribute queries,” IEEE Conference on
Computer Vision and Pattern Recognition, 2011.
www.ijesi.org
74 | Page