MEDICAL IMAGE PROCESSING
ANALYSIS OF TEXTURE
CHAPTER 6
MEDICAL IMAGE PROCESSING
- Texture: spatial distribution function of pixel values,
one of the important characteristics of images
- Texture analysis is encountered in several areas
including recognition and classification
- Due to the existence of a wide variety of texture no
single method of analysis would be applicable to several
different situations
- Two categories: (quasi-) periodic and random
+ (quasi-) periodic: repetition of texture at regular or
(quasi-) periodic interval →textons
+ no texton can be identified → random
-2-
MEDICAL IMAGE PROCESSING
-3-
Texture on CT image: liver, kidney, spine, lung (left to right)
Directional texture on
mammograms Ordered texture on
image of endothelial
cells in the cornea
6.1. Statistical Analysis of TextureMEDICAL IMAGE PROCESSING
4
- Simple measures of texture may be derived based upon
the moments of the gray-level PDF or normalized
histogram of the given image
- Variance: inhomogeneity
-Skewness: asymmetry
- Kurtosis: uniformity
( )
( )
2
1
220.
L
f
l
m l p l

−
=
= = −
33/2
2
m
skewness m
=
42
2
m
kurtosis m
=
6.1. Statistical Analysis of TextureMEDICAL IMAGE PROCESSING
-5-
• Gray-Level Co-occurrence Matrix (GLCM or GCM)
- Basic to calculate most commonly used measures of
textures that proposed by Haralick
- Another name: spatial gray-level dependence matrix
(SGLD).
- Each element Pd,θ(l1,l2) represents the probability of
occurrence of the pair of gray levels (l1,l2) separated by a
given distance d at angle θ (normally kπ/4).
- There are L gray levels in the image → size of the
GLCM is LxL.