Scribd
Upload
22 views15 pages

t5 Presentation

The document discusses texture analysis for image retrieval. It defines texture, describes common texture features like statistical, frequency domain, and watershed features. An example application calculates feature vectors for each texture class in a database and performs retrieval by finding the closest vector to a query image using Euclidean distance. Challenges include speed of comparison in large databases and ensuring features are invariant to transformations.

Uploaded by

deepak.cosasvoc
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
22 views15 pages

t5 Presentation

The document discusses texture analysis for image retrieval. It defines texture, describes common texture features like statistical, frequency domain, and watershed features. An example application calculates feature vectors for each texture class in a database and performs retrieval by finding the closest vector to a query image using Euclidean distance. Challenges include speed of comparison in large databases and ensuring features are invariant to transformations.

Uploaded by

deepak.cosasvoc
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 15

Texture analysis

Team 5
Alexandra Bulgaru
Justyna Jastrzebska
Ulrich Leischner
Vjekoslav Levacic
Güray Tonguç
Contents
 Project goal
 Definition of texture
 Features used in texture analysis
 Example of application for texture based
image query
 Results
 Conclusions
Project goal
 Defining a set of features which would help in
identifying the textures in the image
 Examining the relation between features and
the textures
 Defining a simple set of features to identify
similar textures in texture database
 Possibility of using texture classification and
segmentation in later applications
Definition of a texture
 Texture is used to describe two dimensional arrays of
variation.
 The elements and rules of spacing or arrangement in texture
may be arbitrarily manipulated, provided a characteristic
repetitiveness remains.
Features used in texture analysis
Problem of feature selection depends on:
 Type of application (medical, aerial, etc.)
 Need of invariances (rotational, shifting, scaling, lightning,
etc.)

Examples of features we used:


 Statistical (for example derived from co-occurrence matrix
like entropy, contrast, correlation)
 High level (derived from the watershed algorithm)
 Frequency domain (energy bands)
Co – occurrence matrix
neighbour pixel value

0 1 2 3

Reference pixel value:


0 2 2 1 0
1 0 2 0 0
2 0 0 3 1
3 0 0 0 1

Original image Contrast- feature derived from


the Co- occurrence matrix
Calculation of the feature value
0.16 0.08 0.04 0
0 1 4 9 0 0.08 0.16 0
0.08 0.16 0 0
1 0 1 4 0.08 0 0 0
0.04 0 0.24 0.04
x =
4 1 0 1 0.16 0 0 0.04
0 0 0.04 0.08 0 0 0.04 0
9 4 1 0

Standardized symmetric matrix

Sum of all cells


=
0.586

Original image Contrast- feature derived form


the Co - occurrence matrix
Statistical features - Entropy

Entropy    pi  ln( pi ) pi 
number of pixels with intensity value i
i
total number of pixels

Original image Entropy


Watershed segmentation
 Average area of components
 Number of components in specific region
 Ratio between circumference and components number

Original image Watershed from Watershed form the median


original image filtered image (smoothing of noise
to avoid oversegmentation)
Watershed analysis – Average area
of components

Watershed form With a big filter size:


Original image
the median Better features inside but
filtered image borders are imprecise
Frequency domain feature

Low spectrum Low to high spectrum Periodicity of image


Example of application for texture
based image query
Concept of work
 We wanted to represent each texture as a feature vector
 Each texture Fn, where n is number of textures in the database,
will be noted as unique class

E f1 f2 … fn = F1
Which classifier to use?
 SVM
 Can be used if multiple classifiers are used but
there are problems with small number of training
vectors and large number of classes
 Our solution
 Definition of a measure – Euclidean distance
 Simple comparing the length between input feature
vector with those in database and taking the closest
Problems to address
 In large database of textures how to compare
the feature vectors fast
 Using the features which are invariant to
different transformations
 How to include more sophisticated measure
which will favor selecting the feature vector of
the texture in a database which resembles most
to the input image

You might also like