Information Technology Reference
In-Depth Information
Figure 1. Content-based image classification and retrieval architecture
Definition 1
(Descriptor).
Using a co-occurrence matrix, different properties of the pixel distribution
can be obtained by applying mathematical operations to the matrix values. These operations are called
descriptors. Each descriptor is related to a particular visual feature about the texture.
Haralick et al. proposed a set of 14 descriptors (Haralick, 1979) derived from the co-occurrence
matrix. In this paper, five features (descriptors) were selected for further study: maximum probability
(
MP
), contrast (
Cont
), inverse different moment (
IM
), angular second moment (
AM
), and entropy (
En-
tro
). Where:
•
Maximum probability (
MP
) is defined as follows:
∑
,
(8)
MP
=
max
P i j
( , )
i j
,
•
Contrast feature is a measure of the image contrast or the number of local variations is present in
an image. It takes the following form:
∑
2
,
(9)
Cont
=
(
i j P i j
−
) ( , )
i j
,
Search WWH ::
Custom Search