Texture classification based on overlapped texton co occurrence matrix (otcom) features

IJRET: International Journal of Research in Engineering and Technology eISSN: 2319-1163 | pISSN: 2321-7308
_______________________________________________________________________________________
Volume: 04 Issue: 01 | Jan-2015, Available @ http://www.ijret.org 280
TEXTURE CLASSIFICATION BASED ON OVERLAPPED TEXTON
CO-OCCURRENCE MATRIX (OTCoM) FEATURES
Patnala S. R. Chandra Murthy1
, U Ravi Babu2
, R Venkatalakshmi3
1
Assistant Professor, Department of CSE, University College of Engineering and Technology, ANU-Guntur
2
Professors, Department of CSE, Malla Reddy Engg, College (Autonomous) TS, INODA
3
Research Schalor, JNTUK. Kakinada, India
Abstract
Abstract: The pattern identification problems such as stone, rock categorization and wood recognition are used texture
classification technique due to its valuable usage in it. Generally, texture analysis can be done one of the two ways i.e. statistical
and structural approaches. More problems are occurred when working with statistical approaches in texture analysis for texture
categorization. One of the most popular statistical approaches is Gray Level Co-occurrence Matrices (GLCM) approach. This
approach is used to discriminating different textures in images. This approach gives better accuracy results but this takes high
computational cost. Usually, texture analysis method depends upon how the texture features are extracted from the image to
characterize image. Whenever a new texture feature is derived it is tested whether it is precisely classifies the textures or not.
Texture features are most important for precise and accurate texture classification and also important that the way in which they
are extracted and applied. The present paper derived a new co-occurrence matrix based on overlapped textons patterns. The
present paper generates overlapped texton patterns and generates co-occurrence matrices derived a new matrix called
Overlapped Texton Co-occurrence Matrices (OTCoM) for stone texture classification. The present paper integrates the
advantages of co-occurrence matrix and texton image by representing the attribute of co-occurrence. The co-occurrence features
extracted from the OTCoM provides complete texture information about a texture image. The proposed method is experimented
on Vistex, Brodatz textures, CUReT, Mayang, Paul Brooke, and Google color texture images. The experimental results indicate
the proposed method classification performance is superior to that of many existing methods.
Keywords: co-occurrence matrix, texton, Texture Classification
--------------------------------------------------------------------***----------------------------------------------------------------------
1. INTRODUCTION
Texture classification and segmentation is an important
research area from industrial to bio-medical images. The
classification problem is basically the problem of identifying
an observed textured sample as one of several possible
texture classes by a reliable but computationally attractive
texture classifier. This implies that the choice of the textural
features should be as compact as possible and yet as
discriminating as possible. In other words, the extraction of
texture features should efficiently embody information
about the textural characteristics of the image. The ultimate
goal of texture characterization systems is to recognize
different textures. To design an effective algorithm for
texture classification, it is essential to find a set of texture
features with good discriminating power. Previously a
number of different texture analysis methods have been
introduced namely statistical, structural, transform based
and model based methods [1, 2, 3] Normally textures are
studied through statistical and syntactical methods. The
statistical method measures the coarseness and the
directionality of textures in terms of averages on a window
of the image [4, 5, 6]. On the other hand syntactical method
describes the shape and distribution of the entities. The
statistical method has the main features which are to be
extracted that includes the autocorrelation function, Fourier
transform domain, Markov random field models, local linear
transforms, power spectra, difference gray level statistics,
co-occurrence matrices and from sum and different statistics
[7, 8, 9, 10, 11, 12, 13].
Initially, texture analysis was based on the first order or
second order statistics of textures. The co-occurrence matrix
features were first proposed by Haralick [6]. Weszka [14]
compared texture feature extraction schemes based on the
Fourier power spectrum, second order gray level statistics,
the co-occurrence statistics and gray level run length
statistics. The co occurrence features were found to be the
best of these features. This fact is demonstrated in a study
by Conners and Harlow [15]. In [16], Haralick features are
obtained from wavelet decomposed image yielding
improved classification rates.
S.S Sreeja mole [17] in this method classifies the textures on
a pixel basis, where each pixel is associated with textural
features extracted from co-occurrence matrices that differs
the pixel itself. Here the windows related with the adjacent
pixels are mostly overlapping resulting the pixels can be
obtained by updating values already found. The
classification rate in this method is 90%. Jing Yi Tou [18]
proposed a method. In this method two popular texture
analysis methods i.e. Gabor filters and the Grey Level Co-
occurrence Matrices (GLCM). By using this method
achieved a recognition rate of 88.52%. Guang-Hai Liu [19]
proposed another method. In this method uses the Textons
concept and the Grey-level Co-occurrence Matrices
(GLCM) techniques used for texture categorization. The
preset method uses the combination of the Grey-level Co-

_______________________________________________________________________________________
occurrence Matrices (GLCM) and textons are used for stone
texture classification. This method can achieve higher
classification rate compare to existing methods. The present
paper derived a new co-occurrence matrix based on
overlapped textons for texture classification. The new co-
occurrence matrix is called as Overlapped Texton Co-
occurrence Matrix (OTCoM)
This paper is organized as follows. In Section 2, OTCoM
and texture features are proposed. Section 3 discusses results
and discussions. Conclusions are given in Section 4.
2. GENERATION OF OVERLAPPED TEXTON
CO-OCCURRENCE MATRIX (OTCoM) AND
EXTRACTION OF FEATURES
Various algorithms are proposed by many researchers to
extract color, texture and other features. Color is the most
distinguishing important and dominant visual feature. That’s
why color histogram techniques remain popular in the
literature. The main drawback of this is, it lacks spatial
information. The proposed method consists of three steps
which are listed below. In the first step of the proposed
method is the color image is converted in to grey level
image by using any HSV color model. The following
section describes the RGB to HSV conversion procedure
2.1 RGB to HSV Color Model Conversion
Recent literature revel various color models in color image
processing. In order to extract facial image features from
color image information, the proposed method utilized the
HSV color space. In the RGB model, images are represented
by three components, one for each primary color – red,
green and blue. Hue is a color attribute and represents a
dominant color. Saturation is an expression of the relative
purity or the degree to which a pure color is diluted by white
light. HSV color space describes more accurately the
perceptual color relationship than RGB color space because
it is adopted with a non-linear transform. The present paper
has used HSV color space model conversion, because the
present study is aimed to classify the human age in to four
groups with a gap of 15 years.
HSV color space is created by Hue (H), saturation (S) and
value (V). Hue is the property of color such as red, green
and blue. Saturation is the intensity of a specific color.
Value is brightness of a specific color. However, HSV color
space separates the color into three categories i.e. hue,
saturation, and value. Separation means variations of color
observed individually.
The transformation equations for RGB to HSV color model
conversion is given below.
(1)
S = (2)
(3)
(4)
(5)
The range of color component Hue (H) is [0,255], the
component saturation (S) range is [0,1] and the Value (V)
range is [0,255]. In this work, the color component Hue (H)
is considered as color information for the classification of
facial images. Color is an important attribute for image
processing applications.
2.2 Overlapped Texton Matrix Detection
The texton patterns are defined as a set of blobs or growing
patterns sharing a common property on the image [21, 22].
Based on the texton theory, texture can be decomposed into
elementary units. Julesz’s texton theory mainly focuses on
analyzing regular textures, while the overlapped textons can
be considered as the extension of Julesz’s textons. Since
overlapped texton involve texture and shape (edge)
information, they can better present features for texture
classification.
The present paper utilized a 2×2 sub window texton pattern
as shown in Fig 1(a). In figure 1(a), the four pixel values of
a 2×2 sub window are denoted as PV1, PV2, PV3 and PV4. If
two pixels are highlighted in gray color of same value then
the grid will form a texton. The six texton types denoted as
TP1, TP2, TP3, TP4, TP5 and TP6 are shown in figure 1(b) to
1(g).
PV1 PV2
PV3 PV4
(a) (b) (c) (d)
(e) (f) (g)
Fig 1 Six special types of Textons: a) 2×2 grid b) TP1 c) TP2
d) TP3 e) TP4 f) TP5 and g) TP6.
Julesz’s texton theory mainly focuses on analyzing regular
textures, but in all those texture analysis use non overlapped
texton features for texture analysis. The disadvantage of non
overlapped texton pattern, texton matrix consists of more
number of zeros so that more information about the image
lost. Another major disadvantage is that non overlapped
textons does not consider the neighboring texton elements,
even though they form the texton pattern with neighboring
texton elements. Form Fig. 2 we observe that non
overlapped texton image consists more number of zeros. To
overcome the above disadvantages overlapped texton
concept applies that causes the texton image consists less
number of zeros and more information about the image is
available for precise texture classification.

_______________________________________________________________________________________
Fig 2. Non overlapped texton matrix example (a) original matrix (b) non overlapped matrix
There are many types of texton patterns in texture images. In this paper, we define six types of texton patterns and co-occurrence
matrix derived from the overlapped texton pattern image for texture analysis.
Fig.3: Illustration of the OTM process: (a) 2×2 grid (b) Original image (c) Overlapped Texton Matrix (OTM)
2.3 Co-occurrence Matrix and Features
Recently Texton Co-occurrence matrix (TCM) is proposed
in the literature [27] for higher retrieval rate. TCM is
defined to be the distribution of co-occurring texton with a
given offset over the texton index image. More over TCM is
a computationally expensive procedure. To overcome this,
the present paper considers Overlapped Texton Matrix
(OTM), which is directly obtained from the original image.
To extract precise texture features, the present study
computes co occurrence matrix for OTM. Due to co-
occurrence matrices are typically large and sparse they
are used to measure the texture image. GLCM is proposed
by Haralick et al back in 1973 [6]. It is widely used for
various texture analysis applications, such as texture
Analysis [24], rock texture classification, wood
classification and etc. GLCM is a popular statistical
technique for extracting textural features from different
types of images. In order to find the spatial relationships
effectively, the classification method is used and Grey-level
co-occurrence matrix (GLCM) is one of the most widely
used statistical texture measures. The idea of the method is
to consider the relative frequencies for which two
neighboring pixels are separated by a distance on the image.
Since the GLCM collects information about pixel pairs
instead of single pixels and which is called by a name as
second-order statistics.
The GLCM is generated by cumulating the total numbers of
grey pixel pairs from the images. Each GLCM will be
generated by defining a spatial distance d and an orientation,
which can be 0 degree, 45 degree, 90 degree or 135 degree
at a selected grey level G. The GLCM produced will be of
size G × G. When the GLCM is constructed, Cd(r,n)
represents the total pixel pair value where r represents the
reference pixel value and n represents the neighboring pixel
value according to the spatial distance and orientation
defined. Co occurrence matrix is generated from the OTM is
called Overlapped Texton Co-occurrence Matrix (OTCoM).
Based on this, OTCoM with different orientations 00
, 450
,
900
, and 1350
are formed as shown in Fig.4(a)-(e)
respectively. Textural features are extracted from the
OTCoM for classification process. There are a total of
fourteen features for GLCM [25]. The textural features used
in this method are energy, entropy, contrast, local
homogeneity, correlation, and inertia are shown in Eq (1) to
Eq (6) [5].
151 143 143 143 152 153 146 146 151 143 143 143 152 153 146 146
151 143 143 143 152 153 146 146 151 143 143 143 152 153 146 146
155 142 142 138 147 153 148 148 0 0 0 0 0 0 148 148
157 143 143 135 142 151 149 149 0 0 0 0 0 0 149 149
157 143 143 135 142 151 149 149 0 0 0 0 0 0 149 149
154 146 146 140 143 148 146 146 0 0 0 0 0 0 146 146
154 146 146 140 143 148 146 146 0 0 0 0 0 0 146 146
145 149 149 150 147 144 143 143 0 0 0 0 0 0 143 143
(a) (b)
151 143 143 143 152 153 146 146 151 143 143 143 152 153 146 146
151 143 143 143 152 153 146 146 151 143 143 143 152 153 146 146
V1 V2 155 142 142 138 147 153 148 148 0 142 142 0 0 153 148 148
V3 V4 157 143 143 135 142 151 149 149 157 143 143 135 142 151 149 149
157 143 143 135 142 151 149 149 157 143 143 135 142 151 149 149
154 146 146 140 143 148 146 146 154 146 146 140 143 148 146 146
154 146 146 140 143 148 146 146 154 146 146 140 143 148 146 146
145 149 149 150 147 144 143 143 0 149 149 0 0 0 143 143

_______________________________________________________________________________________
3 0 0 0 0 0 0 4 0 0 0 0 0 0 0 7
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
2 0 0 0 0 0 0 47 3 0 0 0 0 0 0 46
///////(a) (b)
0 0 0 0 0 0 0 7 0 0 0 0 0 0 0 5
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
2 0 0 0 0 0 0 40 3 0 0 0 0 0 0 41
(c) (d)
Fig. 4: a) Overlapped Texton matrix (b) (c), (d) and (e) represents the co occurrences on OTCoM of 00
, 450
, 900
and 1350
.
3. RESULTS AND DISCUSSIONS
The present paper carried out the experiments on two
Datasets. The Dataset-1 consists of various Mosaic, Granite,
Marble and Brick stone textures with resolution of 256×256
collected Vistex, Brodatz textures, Mayang, Google color
texture images and also from natural images taken by using
digital camera. Some of stone texture images in Dataset-1
are shown in the Fig. 5. The Dataset-2 consists of various
Mosaic, Granite, Marble and Brick stone textures with
resolution of 256×256 collected from CUReT, Paul Bourke,
and also from natural images taken by using digital camera.
Some of images in Dataset2 are shown in the Fig. 6.
Dataset-1 and Dataset-2 contains 80 and 96 original color
texture images respectively.
Every texture image is subdivided into 16 sub images of
non-overlapped image regions of size (64×64). This results
into a total of 2816 sub image regions. The classification is
done for all sub texture regions derived from each texture
image in Dataset-1 and Dataset-2. Feature set leads to
representation of the training and testing textures features.
The absolute difference of the feature vector values of the
query texture and database textures are also calculated.
Fig.5: Input texture group of 8 samples of Brick, Granite, Mosaic., Marble with size of 256×256

_______________________________________________________________________________________
Fig 6: Input texture group of 8 samples of Brick, Granite, Mosaic, Marble with size of 200×150
To classify the relevant textures, fixed threshold, K-NN classifier is used to measure the similarity between query texture and the
database textures. In case of fixed threshold, the threshold values are computed for different query textures. The best threshold
value is chosen as the threshold of that particular texture feature. The Euclidean distance between these FVs helps in classifying
the texture into correct group.
The results from two datasets are obtained in Table 1 & 2 which shows the average classification rates of the proposed OTCoM
method.
Table 1a: Database-1: (%) mean classification rate of brick and marble stone textures
Sno Texture Name Classification Rate Texture Name Classification Rate
1 Brick1 96.75 marble1 95.45
2 Brick2 92.81 marble2 97.47
3 Brick3 90.34 marble3 95.12
4 Brick4 96.28 marble4 96.58
5 Brick5 97.47 marble5 91.78
6 Brick6 96.9 marble6 87.57
7 Brick7 90.92 marble7 93.65
8 Brick8 92.71 marble8 93.78
9 Brick9 91.29 marble9 97.42
10 Brick10 96.62 marble10 87.53
11 Brick11 94.74 marble11 95.86
12 Brick12 93.17 marble12 91.79
13 Brick13 91.71 marble13 95.89
14 Brick14 92.76 marble14 95.67
15 Brick15 91.76 marble15 91.17
16 Brick16 93.37 marble16 96.37
17 Brick17 91.76 marble17 90.22
18 Brick18 91.79 marble18 96.23
19 Brick19 91.71 marble19 95.53
20 Brick20 91.76 marble20 91.75
Average 93.331 Average 93.8415
Table 1b: Database-1: (%) mean classification rate of mosaic and granite stone textures
1 granite1 91.51 mosiac1 93.87
10 granite10 96.27 mosiac10 94.71
11 granite11 91.72 mosiac11 98.24

_______________________________________________________________________________________
12 granite12 99.74 mosiac12 96.91
13 granite13 97.62 mosiac13 97.52
14 granite14 91.75 mosiac14 92.74
15 granite15 96.93 mosiac15 97.63
16 granite16 97.19 mosiac16 97.48
17 granite17 98.23 mosiac17 91.91
18 granite18 96.83 mosiac18 97.24
19 granite19 91.85 mosiac19 91.56
20 granite20 91.78 mosiac20 97.64
Table 2a: Database-2: (%) mean classification rate of brick and marble stone textures
1 Brick1 92.90 marble1 93.00
2 Brick2 98.13 marble2 91.7
3 Brick3 93.63 marble3 91.65
4 Brick4 95.37 marble4 94.17
5 Brick5 93 marble5 94.13
6 Brick6 93.57 marble6 89.53
7 Brick7 96.13 marble7 97.53
8 Brick8 94.77 marble8 88.37
9 Brick9 93.07 marble9 95.37
10 Brick10 93.97 marble10 94.6
11 Brick11 93.75 marble11 87.5
12 Brick12 97.13 marble12 98.03
13 Brick13 91.25 marble13 88.37
14 Brick14 88.37 marble14 99.83
15 Brick15 88.37 marble15 96.43
16 Brick16 88.37 marble16 97.43
17 Brick17 88.37 marble17 88.37
18 Brick18 94.37 marble18 92.70
19 Brick19 97.67 marble19 88.37
20 Brick20 89.5 marble20 89.6
21 Brick21 93.50 marble21 91.75
22 Brick22 91.90 marble22 95.85
23 Brick23 95.20 marble23 92.03
24 Brick24 91.13 marble24 95.70
Table 2a: Database-2: (%) mean classification rate of granite and marble stone textures
1 granite1 mosiac1 94.3
10 granite10 93.07 mosiac10 90.20
11 granite11 94.55 mosiac11 88.37
12 granite12 93.07 mosiac12 85.45
13 granite13 95.33 mosiac13 95.23
14 granite14 94.83 mosiac14 93.37
15 granite15 90.73 mosiac15 95.23

_______________________________________________________________________________________
18 granite18 90.60 mosiac18 94.43
20 granite20 92.17 mosiac20 96.65
21 granite21 98.83 mosiac21 97.33
22 granite22 91.70 mosiac22 98.13
23 granite23 92.43 mosiac23 97.43
24 granite24 96.20 mosiac24 97.33
Mean classification rates for the proposed OTCoM and the other existing methods for classification stone textures using K-NN
classifier is shown in Table 3 which clearly indicates that the proposed OTCoM outperforms the other existing methods. Fig.7
shows the comparison chart of the existing methods which are specified in table 3 and proposed OTCoM method.
Table 3: Mean classification rates for the two different texture image datasets using k-NN classifier
Image Dataset Pixel Based image Classification GLCM and Gabor Filters Proposed Method FBTCoM
Brodatz 90.34 88.53 96.93
VisTex 89.12 88.15 97.19
Mayang 91.23 87.08 98.23
CUReT 89.12 84.61 96.83
Paul Bourke 91.23 87.01 94.55
Fig 7: Classification accuracy comparison of K-NN classifier obtained in Brodatz Vistex, Paul Bourke, CUReT, and Mayang
dataset for Pixel Based image Classification, GLCM and Gabor Filters and proposed method.
4. CONCLUSION
Paragraph The present paper derived a new co-occurrence
matrix called as Overlapped Texton Co-occurrence Matrix
(OTCoM) for rotation invariant texture classification. Julesz
[21] proposed texton which represents the patterns of texture
which is useful in texture analysis. The disadvantage of
TCM is that, the computationally expensive. To overcome
this problem, the present paper considered overlapped
Texton Matrix (OTM), which is directly obtained from a
original image and to extract a precise texture features.
The features on color textures are extracted by means of
GLCM [17] statistical method though the concept of
overlapping window for neighboring pixels. So that, it is the
computationally expensive. The present paper statistical
method though the concept of non overlapping window for
neighboring pixels. The experimental results clearly indicate
the efficacy of the proposed OTCoM over the various
existing methods.
REFERENCES
[ 1 ] “Measuring Texture Classification Algorithms” by
G. Smith and I. Burns, , Pattern Recognition Letters,
1997, vol. 18, pp. 1495-1501
[ 2 ] “Statistical and Structural Approaches to Texture,”
by Haralick, R.M., Proceedings of the IEEE, 67, pp.
786-804, 1979
[ 3 ] ‘A theoretical comparison of texture algorithms’ by
Conners, r.w., and harlow, C.A.:, IEEE
Transactions., May 1980, PAMI-2, pp. 20.5222
[ 4 ] Texture Classification By Using Advanced Local
Binary Patterns And Spatial Distribution Of
Dominant Patterns”, by Shu Liao and Albert C. S.
Chung “in 2007.pp.1221-1224
[ 5 ] “A comparative study of texture measures for terrain
classification” by Weszka, j., dyer, c., and rosenfeld,
A in IEEE Transactions. April 1976, SMC- 6, (4),
pp. 269-285
75
80
85
90
95
100
Brodatz VisTex Mayang CUReT Paul
Bourke
Pixel Based image
Classification
GLCM and Gabor Filters
Proposed Method
FBTCoM

_______________________________________________________________________________________
[ 6 ] Haralick, r.m., shanmugam, k., and dinstein,
“Textural features for image Classification’, IEEE
Transactions, November 1973, SMCJ, (6), pp. 610-
621
[ 7 ] Unser, M.: ‘Sum and difference histograms for
texture classification’, IEEE Trans., January 1986,
PAMI-8, (I), pp. 118-125
[ 8 ] Davis, l.s., johns, s.a., and aggarwal, J.K.: “Texture
analysis using generalized co-occurrence matrices’,
IEEE Trans.,July 1979, PAMI-1, (3). pp. 251-259
[ 9 ] Alparone, l., argenti, f., and benelli, g.: ‘Fast
calculation of co-occurrence matrix parameters for
image segmentation’,Electron. Lett.,January
1990,26, (I), pp. 23-24
[ 10 ] A. Laine and J. Fan, “Texture classification by
wavelet packet signatures”, IEEE Transactions. on
PAMI, 15(11), 1993, pp. 1186-1190.
[ 11 ] “Multichannel Texture Analysis Using Localized
Spatial Filters” by A. Bovik , M. Clark , W. S.
Geisler, IEEE Transactions on Pattern Analysis and
Machine Intelligence, 12 (1), 1990, pp. 55-73.
[ 12 ] A. K. Jain and F. Farrokhnia, “Unsupervised texture
segmentation using Gabor filters”, in Pattern
Recognition, 24(12), 1991, pp. 1167-1186.
[ 13 ] M. Unser, M. Eden, “Multiresolution feature
extraction and selection for texture segmentation”,
IEEE Transactions on Pattern Analysis and Machine
Intelligence, 11, 1989, pp. 717-728
[ 14 ] J. S. Weszka, C. R. Dyer, A. Rosenfeld, “A
comparative study of texture measures for terrain
classification”, IEEE Transactions in System Man
Cybernat,6(4), 1976, pp. 269-285.
[ 15 ] R. W. Conners and C. A. Harlow. “A Theoretical
Comparison Texture Algorithms”, IEEE
Transactions on Pattern Analysis and Machine
Intelligence, 2, 1980, pp. 204-222.
[ 16 ] “Texture classification using wavelet transform” by
S.Arivazhagan, L. Ganesan, , Pattern Recognition
Letters, 24, 2003, pp. 1513-1521.
[ 17 ] “Texture classification using Wavelet Packet
Decomposition”, by P. S.Hiremath, and S.
Shivashankar, ICGSTs GVIP Journal, 6(2), 2006,
pp. 77-80.
[ 18 ] “ Pixel Based Classification On Color Images In
Statistical Texture Analysis” by S.S Sreeja Mole,
Dr.L.Ganesan, IJCSET -2010, vol 1 no 2 pages 41-
46.
[ 19 ] “Gabor Filters and Grey-level Co-occurrence
Matrices in Texture Classification” by Jing Yi Tou,
Yong Haur Tay and Phooi Yee Lau Pattern
Recognition 34, 2001, pp. 727-739.
[ 20 ] Julesz B., ―Textons, The Elements of Texture
Perception, and their Interac-tions,” Nature, vol.290
(5802): pp.91-97, 1981.
[ 21 ] Julesz B., ―Texton gradients: the texton theory
revisited,” Biological Cybernet-ics, vol.54 pp.245–
251, 1986.
[ 22 ] Guang-Hai Liu, Zuo-Yong Li, Lei Zhang, Yong Xu,
"Image retrieval based on micro-structure
descriptor," Pattern Recognition, vol. 44, pp:2123-
2133, 2011.
[ 23 ] M. Tuceryan, and A. K. Jain, “Texture Analysis,
The Handbook of Pattern Recognition and
Computer Vision, Ed. 2”, World Scientific
Publishing Co., 1998.
[ 24 ] “Design of an Intelligent Wood Recognition System
for the Classification of Tropical Wood Species”,
by Y. L. Lew, Master of Engineering (Electrical)
thesis, Faculty of Electrical Engineering, Universiti
Teknologi Malaysia, Malaysia, 2005
[ 25 ] M. Petrou, and P. G. Sevilla, “Image Processing:
Dealing with Texture”, Wiley, 2006
[ 26 ] N. Jhanwar, S. Chaudhuri, G. Seetharaman, B.
Zavidovique “Content based image retrieval using
motif cooccurrence matrix”, Image and Vision
Computing 22 (2004) 1211–1220
[ 27 ] Guang-Hai Liu∗, Jing-Yu Yang “Image retrieval
based on the texton co-occurrence matrix”, Pattern
Recognition 41 (2008) 3521 -- 3527
BIOGRAPHIES
Dr. Patnala S. R. Chandra Murty
completed B.Tech in the year of 2005
from JNTu hydrabad and M,Tech from
ANU, Guntur in the year of 2008. He
received Ph D degree in Computer
Science and Engineering from JNTUK,
Kakinada in year 2013. He published 8
papers in National and International
Journals and 7 papers in National and International
Conferences. Mail Id chandra_psr@rediffmail.com
Dr. U Ravi Babu completed M,Tech from
RVD university in the year of 2005. He
received Ph D degree in Computer Science
and Engineering from ANU, Guntur in
year 2013. He published 10 papers in
National and International Journals and 9
papers in National and International
Conferences. He working as professor in the depart of CSE
at Malla Reddy Engendering College (Autonomous). He is
life member of ISCA, IRS, IAENG and CSTA. Mail id:
uppu.ravibabu@gmail.com
R. Venkata Lakshmi received B.Tech
degree in Computer Science &
Engineering from Acharya Nagarjuna
University. She received her M.Tech
degree in Computer Science &
Engineering from Andhra University.
Presently she is doing her Ph.D in JNTU kainada. Her Area
of interest is Image Processing.

Texture classification based on overlapped texton co occurrence matrix (otcom) features

More Related Content

What's hot (16)

Viewers also liked (9)

Similar to Texture classification based on overlapped texton co occurrence matrix (otcom) features (20)

More from eSAT Journals (20)

Recently uploaded (20)

Texture classification based on overlapped texton co occurrence matrix (otcom) features