01-06-2012, 01:23 PM
Object-Based Image Indexing and Retrieval in DCT Domain using Clustering Techniques
Object-Based Image Indexing and Retrieval in DCT Domain using Clustering Techniques.pdf (Size: 336.45 KB / Downloads: 64)
INTRODUCTION
In recent years, there has been an intense activity in
development image retrieval technique based on image
content. Various systems have been introduced for contentbased
image retrieval (CBIR). CBIR systems operate in two
phases: indexing and searching. In the indexing phase, each
image of the database is represented using a set of image
attribute, such as color[1][2], shape[3][4], texture[5] and
layout[6]. Extracted features are stored in a visual feature
database. In the searching phase, when a user makes a
query, a feature vector for the query is computed. Using a
similarity criterion, this vector is compared to the vectors in
the feature database. The images most similar to the query
are returned to the user.
JPEG COMPRESSEDOF STILL IMAGES
In this section, we describe the minimal subset of JPEG
compression standard, known as the baseline JPEG that is
based on DCT. To apply DCT, each pixel in the image is
level shifted by 128 by subtracting 128 from each value.
Then, the image is divided into fixed size blocks and a DCT
is applied to each block, yielding DCT coefficients for the
block[10][14]. These coefficients are quantized using
weighting functions optimized for the human eye. The
resulting coefficients are encoded using Haffman variable
word-length algorithm to remove redundancies.
FEATURE EXTRACTION AND IMAGE INDEXING
in the proposed method, each block of 8*8 pixels in DCT
domain is divided into 10 sub-bands(Fig.1).
As we known, in the JPEG compression of color images,
the YCbCr color space is used more than other color spaces.
For compression of color images, in this space each subimage(
Y, Cb and Cr) is coded separately. In the proposed
approach for each color block of size 8*8 a 12-D feature
vector is extracted.
IMAGE RETRIEVAL USING OBJECT FEATURES
As mentioned above, representative features extracted
from images are stored in feature database and used for
object-based image retrieval. With current computer
technology, it is impossible to exactly extract objects in the
image and index the object feature. In this paper, we
consider five major image’s content as representative object
features, which described by color and textural information
extracted from DCT domain.
AUTOMATIC IMAGE CLASSIFICATION USING OBJECT FEATURES
For large database with over tens of thousands of image,
effective indexing becomes an important issue in CBIR.
This problem has not been solved very successfully in
current image database systems. A successful categorization
of images will greatly enhance the performance of CBIR
systems by filtering out images from irrelevant classes
during matching [15].
CONCLUSION
In this paper, a novel and effective image indexing
technique is presented that extracts features directly from
DCT domain. Our proposed approach is an object-based
image indexing. For each color image block of size 8*8 in
DCT domain a feature vector is extracted. Then, feature
vectors of all blocks of an image using the k-means
algorithm is clustered into groups. Each cluster represents a
special object of the image. Then we select some clusters
that have largest members after clustering. The centroids of
the selected clusters are taken as image feature vectors and
indexed into the database.