Classification and Retrieval of Images from Databases Using Rough Set Theory - Distributed Artificial Intelligence, Agent Technology, and Collaborative Applications - page 184

degree of dependency

The degree of dependency ( , )

P Q of a set P of attributes with respect to a set Q of class labeling is

defined as:

POS Q ( )

(7)

P

( , ) P Q =

U

where S denotes the cardinality of set S .
The degree of dependency provides a measure of how important P is in mapping the dataset examples
into Q. If ( , )

P Q = 0, then classification Q is independent of the attributes in P, hence the decision
attributes are of no use to this classification. If ( , )

P Q = 1, then Q is completely dependent on P, hence

the attributes are indispensable. Values 0

P < < denote partial dependency, which shows that only
some of the attributes in P may be useful, or that the dataset was flawed to begin with. In addition, the
complement of ( , )

( , ) 1

P Q gives a measure of the contradictions in the selected subset of the dataset.

a r ule- Based syste M for iMage classifica tion

Figure 1 shows a typical architecture of a content-based image retrieval system. It contains two main
subsystems. The first one is concerned with the data insertion that is responsible for extracting appro-
priate features from images and storing them in the image database. This process is usually performed
offline. The second subsystem is concerned with the query processing, which is organized as follows:
the interface allows a user to specify a query by means of a query pattern and to visualize the retrieved
similar images. The query-processing module extracts a feature and rule vector from a query pattern and
applies a metric distance function to evaluate the similarity between the query image and the database
images. Next, the module ranks the database images in a decreasing order of similarity to the query
image and forwards the most similar images to the interface module.

Texture and Color f eature extraction
Texture is one of the most important defining characteristics of an image. Texture is characterized by
the spatial distribution of gray levels in a neighborhood (Kundu & Chen, 1992; Mari, Bogdan, Moncef
& Ari, 2002). In order to capture the spatial dependence of gray-level values that contribute to the
perception of texture, a two-dimensional dependence texture analysis matrix is discussed for texture
consideration. In the literature, different kinds of textural features have been proposed, such as multi-
channel filtering features, fractal-based features, and co-occurrence features (Haralick, 1979; Li, Gray
& Olshen, 2000; Zhang, H. Gong, Y. Low, C.Y. & Smoliar S.W , 1995). For our classification purposes,
the co-occurrence features are selected as the basic texture feature detectors due to their good perfor-
mance in many pattern recognition applications, including medical image processing, remote sensing,
and content-based retrieval. In the following paragraph, we describe the co-occurrence matrices and
the features we computed from them.
A co-occurrence matrix is the two-dimensional matrix of joint probabilities , ( , )

d P i j between pairs

of pixels, separated by a distance d in a given direction r .