You are on page 1of 18

FEATURE

EXTRACTION
DR. J.N. SWATHI
ASSOCIATE PROFESSOR GRADE2
SCHOOL OF COMPUTER SCIENCE AND ENGINEERING
VELLORE INSTITUTE OF TECHNOLOGY, VELLORE
FEATURE
EXTRACTION
After an image has been segmented into regions of interest,
image representation and image description in a form suitable
for further processing of the image is very important.
Representing a region can be done in two ways. (1) in terms of
its external characteristics and (2) in terms of its internal
characteristics.
An external representation is chosen when the primary focus is
on shape analysis. An internal representation is preferred when
the primary focus is on regional properties like color and
texture.
FEATURE
EXTRACTION
The textural description is done by extracting features from the
image.
An image feature is a distinguishing primitive characteristic or
attribute of an image.
IMAGE TEXTURE
Color is an important cue not only in human vision but also in
digital image processing where its impact is still rising. Color is
measured globally according to the histogram ignoring local
neighboring pixels.
Image texture measurements can be used to segment an image
and classify its segments. Texture is characterized by the
relationship of the intensities of neighboring pixels ignoring
their color.
Texture plays an important role in many machine vision tasks
such as surface inspection, scene classification, surface
orientation and shape determination. Texture is characterized
by the spatial distribution of gray levels in a neighborhood.
TEXTURE

• Texture is a feature used to partition images into regions


of interest and to classify those regions.
• Texture provides information in the spatial arrangement
of colours or intensities in an image.
• Texture is characterized by the spatial distribution of
intensity levels in a neighbourhood.
Texture is a repeating pattern of local variations in image
intensity:
– Texture cannot be defined for a point.
TEXTURE
For example, an image has a 50% black and 50% white
distribution of pixels.

Three different images with the same intensity distribution, but


with different textures.
TEXTURE
Texture consists of texture primitives or texture elements,
sometimes called texels.
– Texture can be described as fine, coarse, grained, smooth, etc.
– Such features are found in the tone and structure of a texture.
– Tone is based on pixel intensity properties in the texel, whilst
structure represents the spatial relationship between texels.
If texels are small and tonal differences between texels are large
a fine texture results.
– If texels are large and consist of several pixels, a coarse
texture results.
TEXTURE ANALYSIS
There are two primary issues in texture analysis:
• texture classification
• texture segmentation
Texture segmentation is concerned with automatically
determining the boundaries between various texture
regions in an image.
Texture classification is concerned with identifying a
given textured region from a given set of texture
classes.
– Each of these regions has unique texture
characteristics.
– Statistical methods are extensively used.
e.g. GLCM, contrast, entropy,
homogeneity
TYPES OF TEXTURE FEATURES

 Texture analysis has played an important role in many


areas including medical imaging, remote sensing and
industrial inspection and image retrieval. The texture
analysis is diverse and differs from each other by the
method used for extracting textural features.
 Four categories of extracting textural features are:
 1) Statistical methods
 2) Structural methods
 3) Model based methods
 4) Transform-based methods.
STATISTICAL
TEXTURE ANALYSIS
Statistical: texture is a quantitative measure of the arrangement
of intensities in a region. This set of measurements is called a
feature vector.
Statistical texture analysis techniques describe texture of
regions in an image through higher order moments of their
grayscale histograms. The most commonly used method for
texture analysis is based on extracting various textural features
from a gray level co-occurrence matrix (GLCM).
The GLCM approach is based on the use of second-order
statistics of the grayscale image histograms.
Structural texture analysis techniques describe a texture as the
composition of well-defined texture elements such as regularly
spaced parallel lines.
Texture is a set of primitive texels in some regular or repeated
relationship.
Model based texture analysis techniques generate an empirical
model of each pixel in the image based on a weighted average of
the pixel intensities in its neighborhood. The estimated
parameters of the image models are used as textural feature
descriptors.
Transform based texture analysis techniques convert the image
into a new form using the spatial frequency properties of the pixel
intensity variations. The success of this type lies in the type of
transform used to extract textural characteristics from the image.
STATISTICAL
TEXTURAL FEATURES
One of the simplest approaches for describing texture is to use
the statistical moments of the gray level histogram of the image.
The various statistical textural features are based on gray level
histogram, gray level co-occurrence matrix.
HISTOGRAM BASED FEATURES
The histogram-based features are first order statistics that include mean, variance, skewness
and kurtosis.
Let z be a random variable denoting image gray levels and p(zi), i = 0,1,2,3,…….L-1, be the
corresponding histogram, where L is the number of distinct gray levels. The features are
calculated using the above-mentioned histogram.
(a) Mean
The mean gives the average gray level of each region and it is useful only as a rough idea of
intensity not really texture.
𝐿−1
𝑚= 𝑖=0 𝑧𝑖 𝑝(𝑧𝑖 )
(b) Variance
The variance gives the amount of gray level fluctuations from the mean gray level value.
𝐿−1
𝑖=0 (𝑧𝑖 −𝑚) 𝑝(𝑧𝑖 )
2
𝜇2 (𝑧) =
(c) Skewness
Skewness is a measure of the asymmetry of the gray levels around the sample mean. If
skewness is negative, the data are spread out more to the left of the mean than to the right. If
skewness is positive, the data are spread out more to the right.
𝐿−1
𝑖=0 (𝑧𝑖 −𝑚) 𝑝(𝑧𝑖 )
3
𝜇3 (𝑧) =
(d) Kurtosis
Kurtosis is a measure of how outlier-prone a distribution is. It describes the shape of the tail of
the histogram.
𝐿−1
𝑖=0 (𝑧𝑖 −𝑚) 𝑝(𝑧𝑖 )
4
𝜇4 (𝑧) =
CO-OCCURRENCE MATRIX
BASED FEATURES
The second-order gray level probability distribution of a texture
image can be calculated by considering the gray levels of pixels in
pairs at a time. A second-order probability is often called a GLC
probability.
The various features that can be calculated from the co-occurrence
matrices (C) are inertia (contrast), absolute value, inverse
difference, energy, and entropy.

(a) Contrast
Contrast is the element difference moment of order 2, which has a
relatively low value when the high values of C are near the main
diagonal.

𝐶𝑜𝑛𝑡𝑟𝑎𝑠𝑡 = (𝑖 − 𝑗)2 𝐶𝑖𝑗


𝑖 𝑗
(b) Energy
Energy value is highest when all values in the co-
occurrence matrix are all equal.

𝐸𝑛𝑒𝑟𝑔𝑦 = 𝐶𝑖𝑗2
𝑖 𝑗

(c) Entropy
Entropy of the image is the measure of randomness of the
image gray levels.

𝐸𝑛𝑡𝑟𝑜𝑝𝑦 = − 𝐶𝑖𝑗 𝑙𝑜𝑔2 𝐶𝑖𝑗


𝑖 𝑗
GLCM – GREY LEVEL CO-
OCCURRENCE MATRIX
DISCLAIMER
The material for the presentation has been compiled from
various sources such as prescribed text book by Rafael C.
Gonzalez and team, tutorials, lecture notes and other
resources available on internet. The information contained in
this lecture/ presentation is for educational purpose only.

17
Dr. J. N.Swathi, SCOPE, VIT
THANK YOU

18
Dr. J. N.Swathi, SCOPE, VIT

You might also like