Vipparthi and Nagar Human-centric Computing and Information Sciences 2014, 4:6

http://www.hcis-journal.com/content/4/1/6

RESEARCH Open Access

Color Directional Local Quinary Patterns for

Content Based Indexing and Retrieval
Santosh Kumar Vipparthi* and Shyam Krishna Nagar

* Correspondence:
Santu155@gmail.com Abstract
Department of Electrical
Engineering, Indian Institute of This paper presents a novel evaluationary approach to extract color-texture features
Technology BHU, Varanasi, India for image retrieval application namely Color Directional Local Quinary Pattern
(CDLQP). The proposed descriptor extracts the individual R, G and B channel wise
directional edge information between reference pixel and its surrounding neighborhoods
by computing its grey-level difference based on quinary value (2, 1, 0, 1, 2) instead of
binary and ternary value in 0, 45, 90, and 135 directions of an image which are not
present in literature (LBP, LTP, CS-LBP, LTrPs, DExPs, etc.). To evaluate the retrieval
performance of the proposed descriptor, two experiments have been conducted on
Core-5000 and MIT-Color databases respectively. The retrieval performances of the
proposed descriptor show a significant improvement as compared with standard local
binary pattern LBP, center-symmetric local binary pattern (CS-LBP), Directional binary
pattern (DBC) and other existing transform domain techniques in IR system.
Keywords: Content based image retrieval (CBIR); Multimedia retrieval; Local patterns;
local ternary patterns (LTP); Directional Binary Patterns (DBC)

Introduction
With the radical expansion of the digitization in the living world, it has become
imperative to find a method to browse and search images efficiently from immense
database. In general, three types of approaches for image retrieval are, text-based,
content-based and semantic based. In recent times, web-based search engines such as,
Google, Yahoo, etc., are being used extensively to search for images based on text
keyword searching. Here, any image needs to be indexed properly before retrieving by
text-based approach. Such an approach is highly tiresome and also unrealistic to
handle by human annotation. Hence, more efficient search mechanism called content
based image retrieval (CBIR) is required. Image retrieval has become a thrust area in
the field of medicine, amusement and science etc.. The search in content based
approach is made by analyzing the actual content of the image rather using metadata
such as, keywords, tags or descriptions associated with an image. Hence, system can
filter images based on their content would provide better indexing and return more
accurate results. The effectiveness of a CBIR approach is greatly depends on feature
extraction, which is its prominent step. The CBIR employs visual content of an image
such as color, texture, shape and faces etc., to index the image database. Hence these
features can be further classified as general (texture, color and shape) and domain
specific (fingerprints, human faces) features. In this paper, we mainly focused on low-
2014 Vipparthi and Nagar; licensee Springer. This is an Open Access article distributed under the terms of the Creative Commons
Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any
medium, provided the original work is properly credited.
Vipparthi and Nagar Human-centric Computing and Information Sciences 2014, 4:6 Page 2 of 13
http://www.hcis-journal.com/content/4/1/6

level features; the feature extraction method used in this paper is an effective way of
integrating low-level features into whole. Widespread literature survey on CBIR is
accessible in [1-4].
The concept of color is one of the significant feature in the field of content-based
image retrieval (CBIR), if it is maintained semantically intact and perceptually oriented
way. In addition, color structure in visual scenery changes in size, resolution and
orientation. Color histogram [5] based image retrieval is simple to implement and has
been well used and studied in CBIR system. However, the retrieval performance of
these descriptors is generally limited due to inadequacy in discrimination power mainly
on immense data. Therefore, several color descriptors have been proposed to exploit
special information, including compact color central moments and color coherence
vector etc. reported in the literature [6,7].
Texture is one of the most important characteristic of an image. Texture analysis
has been extensively used in CBIR systems due to its potential value. Texture
analysis and retrieval has gained wide attention in the field of medical, industrial,
document analysis and many more. Various algorithms have been proposed for
texture analysis, such as, automated binary texture feature [8], Wavelet and Gabor
Wavelet Correlogram [9,10], Rotated Wavelet and Rotated Complex Wavelet filters
[11-13], Multiscale Ridgelet Transform [14] etc.. In practice texture features can be
combined with color features to improve the retrieval accuracy. One of the most
commonly used method is to combining texture features with color features; these
include wavelets and color vocabulary trees [15] and Retrieval of translated, rotated and
scaled color textures [16] etc..
In addition to the texture features, the local image features extraction attracting
increasing attention in recent years. A visual content descriptor can either be local
or global. A local descriptor uses the visual features of regions or objects to
describe the image, where as the global descriptor uses the visual features of the
whole image. Several local descriptors have been described in the literature [17-29],
where the local binary pattern (LBP) [17] is the most popular local feature
descriptor.
The main contributions of the proposed descriptor are given as follows. (a) A new
color-texture descriptor is proposed, it extracts texture (DLQP) features from an
individual R, G and B color channels. (b) To reduce the feature vector length of the
proposed descriptor, the color-texture features were extracted from horizontal and
vertical directions only.
The organization of this paper is as follows, In Section Introduction, introduction is
presented. The local patterns with proposed descriptor are presented in Section Local
patterns with proposed Descriptor. Section Experimental results and discussions,
presents the retrieval performances of proposed descriptor and other state-of-the art
techniques on two bench mark datasets (Corel-5000 and MIT-Color). Based on the
above work Section Conclusions concludes this paper.

Local patterns with proposed Descriptor

The concept of LBP [17], LTP [21] and DBC [29] has been utilized to extract texture
features (DLQP) from individual color channels (R, G and B) to generate a new color-
texture feature called CDLQP.
Vipparthi and Nagar Human-centric Computing and Information Sciences 2014, 4:6 Page 3 of 13
http://www.hcis-journal.com/content/4/1/6

Local binary patterns (LBP)

The concept of LBP was derived from the general definition of texture in a local
neighborhood. This method was successful in terms of speed and discriminative
performance [17].
In a given 3 3 pixel pattern, the LBP value is calculated by comparing its center
pixel value with its neighborhoods as shown below:

X
N1
LBP N; R 2i  f 1 pi pc 1
i0

1 x0
f 1 x 2
0 x<0

where N stands for the number of neighbors, R is the radius of the neighborhood,
pc denotes the grey value of the centre pixel and pi is the grey value of its
neighbors.
The LBP encoding procedure from a given 3 3 pattern is illustrated in Figure 1.

Local ternary patterns (LTP)

The local ternary pattern (LTP) operator, introduced by Tan and Triggs [20] extends
LBP to 3-valued codes called LTP.
8 
< 1; x pc 
f 2 x; pc ; 0; jx pc j <  3
:
1; x pc xp p
i c

where, is user-specified threshold.

After computing local pattern LP (LBP or LTP) for each pixel (i, j), the whole image
is represented by building a histogram as follows:

X
N1 X
N2
  
H LP l f 2 LP i; j; l; l 0; 2P 1 4
i1 j1

Figure 1 The calculation of DBC pattern along 0 direction.

Vipparthi and Nagar Human-centric Computing and Information Sciences 2014, 4:6 Page 4 of 13
http://www.hcis-journal.com/content/4/1/6


1 xy
f 2 x; y 5
0 else

where the size of input image is N1 N2.

Directional binary code (DBC)

The directional binary code (DBC) was proposed by Baochang et al. [30]. DBC encodes
the directional edge information as follows.
In a given an image I, the first-order derivative is calculated, I'(gi) along 0, 45, 90
and 135 directions. The detailed calculation of DBC in red color channel along 0
direction is shown in Figure 1.
0     
I g i I g c I g i ; i 1; 2; 3; 4 6

The directional edges are obtained by,

   
^I DBC g f 4 I 0 g ; j 1 =45 0 ; 45 ; 90 and 135 7
c j

The DBC is defined ( = 0, 45, 90 and 135) as follows:

( )
      
DBC I g c j ^I DBC g c ; ^I DBC g 1 ; ^I DBC g 2 ; :::::::::^I DBC g 8
8

Proposed descriptor
Color Directional Local Quinary Pattern (CDLQP)
In this section, the procedure to generate a new color-texture feature (CDLQP)
descriptor is explained. Let Ii be the ith plain (color space) of the image (e.g., Red color
component from the RGB color space), where i = 1,2,3. The DLQP feature is
computed independently from each (R, G and B) color channels.
For a given image I, the first-order derivatives of 0, 45, 90 and 135 directions are
calculated using Eq. (6).
The directional edges were obtained by Eq. (9). The local quinary values were
obtained by Eq. (10).
DLQP    
^I i g c f 5 I 0 g j ; j 1 =45 0 ; 45 ; 90 and 135 9
8 
> 2; pc 2 
>
> 
>
< 1; 1 p c > 
2 
f 6 pc ; 2 ; 1 0; 1 > pc < 1  10
>
>
>
> 1; 2 < pc 1 
: 
2; pc 2

where, 2, 1 are the upper and lower threshold parameter respectively.

The Nth-order CDLQP is defined (= 0, 45, 90 and 135) as follows:
   n i   i i DLQP i  o
CDLQP i;N I g c j ^I DLQP g c ; ^I DLQP g 1 ; ^I g 2 ; :::::::::^I DLQP g 8
11

DLQP is a quinary (2, 1, 0, 1, 2) pattern, which is further converted into four

binary patterns such as, two upper patterns (UP) and two lower patterns (LP). The
Vipparthi and Nagar Human-centric Computing and Information Sciences 2014, 4:6 Page 5 of 13
http://www.hcis-journal.com/content/4/1/6

detailed representation of these four patterns is shown in Figure 2. Finally, the whole
image is represented by building a histogram supported by Eq. (12).

N1 X
X N2
 
H i CDLQPj l f 7 CDLQP j; k j ; l ; l0; 511 12
j1 k1

where

1; if j k
f 7 j; k
0; if j k

The size of the input image is N1 N2.

In this brief, to reduce the feature vector length color-texture features were extracted
from horizontal and vertical directions only.
The details of the proposed color-texture descriptor is given as follows. The steps for
extracting 0 degree information is shown in Figure 1. Figure 2 and Eq. (10) explain the
procedure to calculate the quinary pattern. The generated quinary pattern is further
coded into two upper (A & B) and two lower (C& D) binary patterns which are shown
in Figure 2. The two upper (A & B) patterns were obtained by retaining 2 by 1 and
replacing 0 for 2, 1, 1 and 0 for A pattern. Likewise, pattern B was obtained by
retaining 1 by 1 and replacing 0 for other values. A similar procedure was followed for
other two lower patterns.
From the Figure 2, -11, 3, 14, 8, 5, 5, 2, 4, 1 texture information are obtained
when first-order derivative applied in 0 direction. Further, the derivatives are coded in
to quinary pattern -2, 1, 2, 2, 2, 2, 1, 2, 1 using upper and lower thresholds
(1 = 2 & 2 = 1). Finally, the quinary pattern was converted into four binary patterns
(two UP and two LP). The entire operation was applied on individual color channels to
generate color-texture features.

Figure 2 DLQP operator calculation in 0 direction.

Vipparthi and Nagar Human-centric Computing and Information Sciences 2014, 4:6 Page 6 of 13
http://www.hcis-journal.com/content/4/1/6

Figure 3 Proposed image retrieval framework.

Proposed system framework for image retrieval

Figure 3 illustrates the proposed image retrieval system frame work and algorithm for
the same is given below.
Algorithm: The proposed algorithm involves following steps
Input: Image; Output: Retrieval Result

1. Separate RGB color components from an image.

2. Calculate the directional edge information on each color space.
3. Compute the local quinary value for each pixel.
4. Construct the CDLQP histogram for each pattern.
5. Construct the feature vector.
6. Compare the query image with images in the database using Eq. (16).
7. Retrieve the images based on the best matches.

Query matching
The retrieval performance of any descriptor not only depends on feature extraction
approach, but also on good similarity metrics. In this paper four types of similarity
distance measures are used as discussed below:
X  

Manhattan or L1 cityblock Distance: DQ; T i
 f i Q f j T  13

X  2 1=2
 
Euclidean or L2 Distance: DQ; T f Q f j T 
i i
14

 
 
T m ;i f Qm ;i 
XLg  f
Canberra Distance: Ds Qm ; T m   15
 
i1  f T m ;i f Q ;i
m

 
XLg  
 f T ;i f Q;i 
d 1 Distance: DQ; T   16
i1
1 f T ;i f Q;i 

where Q is query image, Lg is feature vector length, T is image in database; fI,i is ith
feature of image I in the database, fQ,i is ith feature of query image Q.
Vipparthi and Nagar Human-centric Computing and Information Sciences 2014, 4:6 Page 7 of 13
http://www.hcis-journal.com/content/4/1/6

Advantages of proposed methods

1. A new color-texture descriptor is proposed, it extracts texture (DLQP)

features from an individual R, G and B color channels.
2. To reduce the feature vector length of the proposed descriptor, the color-texture
features were extracted from horizontal and vertical directions only.
3. To verify the retrieval performances of CDLQP, two extensive
experiments have been conducted on Corel-5000 and MIT-Color
databases respectively.
4. The retrieval performances show a significant improvement nearly
10.78% in terms of ARP on Corel-5000 database and 9.12% improvement
on MIT-Color database in terms of ARR as compared with LBP.

Experimental results and discussions

In image retrieval, various datasets are used for several purposes; these includes Corel
dataset, MIT dataset and Brodtz texture dataset etc.. The Corel dataset is the most
popular and commonly used dataset to test the retrieval performance, MIT-Color
dataset used for texture and color feature analysis and Brodtz dataset used for texture
analysis. In this paper, to verify the retrieval performances of the proposed descriptor
Corel-5000 and MIT-Color datasets are used respectively.
In these experiments, each image in the database is used as the query image. The
retrieval performance of the proposed method is measured in terms of recall,
precision, average retrieval rate (ARR) and average retrieval precision (ARP) as
given in Eq. (17) - Eq. (21) [26]
The recall is defined for a query image Iq is given in Eq. (17).

1 X 
jDBj
     
R Iq; n f 3 I i ; f 3 I q  Rank I i ; I q nj 17
N G i1

where, NG is the number of relevant images in the database, n is the number of top
matches considered, f3(x) is the category of ' x ', Rank(Ii, Iq) returns the rank of image Ii
(for the query image Iq) among all images in the database (|DB|).

  
   1 f 3 I i f 3 I q
f 3 I i ; f 3 I q 18
0 Otherwise

Similarly, the precision is defined, as follows:

Table 1 The retrieval performances of the proposed method (PM) and other existing
methods on Corel-5000 database in terms of ARP and ARR
Database Performance (%) Methods
CS-LBP BLK-LBP LBP LTP DBC DLExP LTrPs PM
Corel-5000 ARP 32.96 45.75 43.62 49.05 50.52 48.72 48.79 54.40
ARR 13.99 20.29 19.22 21.40 22.19 21.05 21.86 23.13
Vipparthi and Nagar Human-centric Computing and Information Sciences 2014, 4:6 Page 8 of 13
http://www.hcis-journal.com/content/4/1/6

Figure 4 Comparison of proposed method with other existing methods in terms of: (a) precision,
(b) recall, (c) ARP and (d) ARR.

  1X jDBj
     
P Iq; n  f I i ; f I q Rank I i ; I q nj 19
3 3
n i1

The average retrieval rate (ARR) and average retrieval precision (ARP) are defined in
Eq. (20) and Eq. (21) respectively.
Table 2 The retrieval results of the proposed method on Corel-5000 database with
different distance measures in terms of ARP and ARR
Performance Distance measure
L1 Canberra L2 d1
ARP (%) 50.02 37.72 54.11 54.40
ARR (%) 20.54 14.86 22.98 23.13
Vipparthi and Nagar Human-centric Computing and Information Sciences 2014, 4:6 Page 9 of 13
http://www.hcis-journal.com/content/4/1/6

55

50
L1
45 L2
Canberra
40 d1

ARP (%) 35

30

25

20

15

10
0 10 20 30 40 50 60 70 80 90 100
No of Top Matching Images
Figure 5 The retrieval performance of the proposed method with different distance measures in
terms of ARP on Corel-5000 database.

1 X
j
jDBj
ARR RI i ; n 20
jDBj i1 n NG

1 X
jDBj
ARP P I i ; n 21
jDBj i1

where, |DB| is the total number of images in the database.

Where N1 is the number of relevant images (Number of images in a group), NC is a
number of groups and N2 is Total number of images to retrieve. The results obtained
are discussed in the following subsections.

Experiment on Corel-5000 database

To verify the performances of the proposed descriptor the Corel 5000 database [30] is
used. It comprises 5000 images of 50 different categories; each category has 100 images,
either in size 187 126 or 126 187. The Corel database is a collection of various
contents ranging from natural images, animals to outdoor sports. In this experiment
the retrieval performances of the proposed descriptor is calculated in terms of
precision, average retrieval precision (ARP), recall, average retrieval rate (ARR). Table 1
illustrates the retrieval performances of the proposed descriptor on Corel-5000
database in terms of ARP and ARR. Figure 4(a) and (b) shows category wise retrieval

Figure 6 Two query results of the proposed descriptor (CDLQP) (top left image is the query image)
on Corel-5000 database.
Vipparthi and Nagar Human-centric Computing and Information Sciences 2014, 4:6 Page 10 of 13
http://www.hcis-journal.com/content/4/1/6

Figure 7 Example images from MIT-Color database, one image from each category.

performance of the proposed method with other existing methods. Figure 4(c) and (d)
shows entire database retrieval performance on Corel-5000 dataset. Table 2 illustrates
retrieval performance of the proposed descriptor with different distance measures.
From Figure 4, it is observed that the proposed method shows less retrieval
performance on categories 1, 25, 26, 27 and 50 as compared to the other existing
methods. The reason behind this is, the categories 1, 25, 26, 27 and 50 contain the
distinct color information within the categories. However, the overall (average)
performance of the proposed method shows a significant improvement as
compared to the existing methods in terms of precision, recall, average and average
retrieval rate on Corel-5 K database. From Table 1 and Figure 4 it is evident that
the proposed method outperforms than other existing methods on Corel-5000
database. From the Table 2 and Figure 5 it is clear that d1 distance measure show
better retrieval rate than other existence distance measures. From this experiment
it is observed that the proposed descriptor shows 10.78% improvement as

100

95
Average Retrieval Rate (ARR %)

90

85
CS-LBP
80 BLK-LBP
LBP
75 LTP
DBC
DLExP
70
LTrPs
PM
65

0 7 14 21 28 35 42 49 56 63 70 77 84 91 98 105 112

No of Top MAtches Considered

Figure 8 Comparison of proposed method with other existing methods on MIT-Color database.
Vipparthi and Nagar Human-centric Computing and Information Sciences 2014, 4:6 Page 11 of 13
http://www.hcis-journal.com/content/4/1/6

98
96
94
92
90
88
86 L1
L2
ARR (%)

84
Canberra
82
d1
80
78
76
74
72
70
68
8 16 24 32 40 48 56 64 72 80 88 96 104 112
No of Top Matches Considered
Figure 9 The retrieval performance of the proposed method with other distance measures in terms
of ARR on MIT-Color database.

compared to LBP. The query result of the proposed method on Corel-5000

database is shown in Figure 6 (top left image is the query image). Finally, from
above discussion and observations it is clear that the proposed descriptor show a
significant improvement as compared to other existing methods in terms of their
evaluation measures on Corel-5000 database.

Experiment on MIT-Color database

In this experiment, we first demonstrate about MIT-Color dataset [31]. Further, we
describe the retrieval performances of the proposed descriptor. MIT-Color dataset
consists of 40 different textures with size 512 512. Further, these textures are divided
into sixteen 128 128 non-overlapping sub images, thus creating a database of 640
(40 16) images. The sample images of this database are shown in Figure 7. The
retrieval performance of the proposed descriptor and other state-of-the-art techniques
are shown in Figure 8. From the Figure 8 it is clear that the proposed descriptor has
shown significant improvement around 9.12% as compared to LBP in terms of ARR.
From Figure 9 it is clear that d1 distance measure show better retrieval rate than other
existence distance measures. From these observations it is concluded that the proposed

Figure 10 Two query results of the proposed descriptor (CDLQP) (top left image is the query image)
on MIT-Color database.
Vipparthi and Nagar Human-centric Computing and Information Sciences 2014, 4:6 Page 12 of 13
http://www.hcis-journal.com/content/4/1/6

descriptor yields better retrieval rate than other state-of-art techniques. Further, the
query result of the proposed descriptor on the MIT-Color database is shown in
Figure 10 (top left image is the query image).

Conclusions
A novel evaluationary color-texture descriptor namely Color Directional Local Quinary
Pattern (CDLQP) is proposed for image retrieval application. CDLQP extracts the
texture features from individual R, G and B color channels using directional edge infor-
mation in a neighborhood with gray-level differences between the pixels by a quinary
value instead of a binary and ternary one. The extensive and comparative experiment
has been conducted to evaluate our color-texture features for IR on two public natural
databases namely, Corel-5000 and MIT-Color dataset. Experimental results of the
proposed descriptor CDLQP show a significant improvement as compared to other
state-of-the art techniques in IR system.

Competing interests
All authors declare that they have no competing interests.

Authors contributions
All authors VSK and NSK work together in conception, implementation, write and apply the proposed methods in this
paper then they read and approved the final manuscript.

Authors informations
Santosh Kumar Vipparthi was born in 1985 in India. He received the B.E and M.Tech degrees in Electrical, Systems
Engineering from Andhra University, IIT-BHU, India in 2007 and 2010 respectively. Currently he is pursuing the Ph.D.
degree in the Department of Electrical Engineering at Indian Institute of Technology BHU, Varanasi, India. His major
interests are image retrieval and object tracking.
Shyam Krishna Nagar was born in 1955 in India. He received the Ph.D degree in Electrical engineering from Indian
Institute of Technology Roorkee, Roorkee, India, in 1991. He is currently working as Professor in Department of
Electrical Engineering, Indian Institute of Technology BHU, Varanasi, Uttar Pradesh, India. His fields of interest are
includes digital image processing, digital control, model order reduction and discrete event systems.

Received: 26 November 2013 Accepted: 19 March 2014

References
1. Rui Y, Huang TS (1999) Image retrieval: Current techniques, promising directions and open issues. J Vis Commun
Image Represent 10:3962
2. Smeulders AWM, Worring M, Santini S (2000) A. Gupta, and R. Jain, Content-based image retrieval at the end of
the early years. IEEE Trans Pattern Anal Mach Intell 22(12):13491380
3. Kokare M, Chatterji BN, Biswas PK (2002) A survey on current content based image retrieval methods. IETE J Res
48(3):261271
4. Ying L, Dengsheng Z, Guojun L, Wei-Ying M (2007) A survey of content-based image retrieval with high-level
semantics. Elsevier J. Pattern Recognition 40:262282
5. Swain MJ, Ballar DH (1991) Indexing via color histograms. Proc. In: 3rd Int. Conf. Computer Vision, Rochester Univ.,
NY., pp 1132
6. Pass G, Zabih R, Miller J (1997) Comparing images using color coherence vectors. In: Proc. 4th. ACM Multimedia
Conf, Boston, Massachusetts, US, pp 6573
7. Stricker M, Oreng M (1995) Similarity of color images. In: Proc. Storage and Retrieval for Image and Video
Databases, SPIE, pp 381392
8. Smith JR, Chang SF (1996) Automated binary texture feature sets for image retrieval. In: Proc. I
EEE Int. Conf. Acoustics, Speech and Signal Processing, Columbia Univ., New York, pp 22392242
9. Moghaddam HA, Khajoie TT, Rouhi AH, Saadatmand MT (2005) Wavelet Correlogram: A new approach for image
indexing and retrieval. Elsevier J. Pattern Recognition 38:25062518
10. Moghaddam HA, Saadatmand MT (2006) Gabor wavelet Correlogram Algorithm for Image Indexing and Retrieval.
In: 18th Int. Conf. Pattern Recognition, K.N. Toosi Univ. of Technol., Tehran, Iran., pp 925928
11. Subrahmanyam M, Maheshwari RP, Balasubramanian R (2011) A Correlogram Algorithm for Image Indexing and
Retrieval Using Wavelet and Rotated Wavelet Filters. International Journal of Signal and Imaging Systems
Engineering 4(1):2734
12. Kokare M, Biswas PK, Chatterji BN (2007) Texture image retrieval using rotated Wavelet Filters. Elsevier J Pattern
recognition letters 28:12401249
13. Kokare M, Biswas PK, Chatterji BN (2005) Texture Image Retrieval Using New Rotated Complex Wavelet Filters.
IEEE Trans. Systems, Man, and Cybernetics 33(6):11681178
Vipparthi and Nagar Human-centric Computing and Information Sciences 2014, 4:6 Page 13 of 13
http://www.hcis-journal.com/content/4/1/6

14. Gonde AB, Maheshwari RP, Balasubramanian R (2010) Multiscale Ridgelet Transform for Content Based Image
Retrieval. IEEE Int. Advance Computing Conf, Patial, India, pp 139144
15. Subrahmanyam M, Maheshwari RP, Balasubramanian R (2012) Expert system design using wavelet and color
vocabulary trees for image retrieval. International Journal of Expert systems with applications 39:51045114
16. Cheng-Hao Y, Shu-Yuan C (2003) Retrieval of translated, rotated and scaled color textures. Pattern Recognition
36:913929
17. Ojala T, Pietikainen M, Harwood D (1996) A comparative study of texture measures with classification based on
feature distributions. Pattern Recognition 29(1):5159
18. Ojala T, Pietikainen M, Maenpaa T (2002) Multiresolution gray-scale and rotation invariant texture classification with
local binary patterns. IEEE Transactions on Pattern Analysis and Machine Intelligence 24(7):971987
19. Guo Z, Zhang L, Zhang D (2010) Rotation invariant texture classification using LBP variance with global matching.
Pattern Recognition 43(3):706719
20. Zhenhua G, Zhang L, Zhang D (2010) A completed modeling of local binary pattern operator for texture
classification. IEEE Transactions on Image Processing 19(6):16571663
21. Tan X, Triggs B (2010) Enhanced local texture feature sets for face recognition under difficult lighting conditions.
IEEE Tans Image Proc 19(6):16351650
22. Subrahmanyam M, Maheshwari RP, Balasubramanian R (2012) Directional local extrema patterns: a new descriptor
for content based image retrieval. Int J Multimedia Information Retrieval 1(3):191203
23. Subrahmanyam M, Maheshwari RP, Balasubramanian R (2012) Local Tetra Patterns: A New Feature Descriptor for
Content Based Image Retrieval. IEEE Trans Image Process 21(5):28742886
24. Subrahmanyam M, Maheshwari RP, Balasubramanian R (2012) Local Maximum Edge Binary Patterns: A New
Descriptor for Image Retrieval and Object Tracking. Signal Processing 92:14671479
25. Subrahmanyam M, Maheshwari RP, Balasubramanian R (2012) Directional Binary Wavelet Patterns for Biomedical
Image Indexing and Retrieval. Journal of Medical Systems 36(5):28652879
26. Vipparthi S, Nagar SK (2013) Directional Local Ternary Patterns for Multimedia Image Indexing and Retrieval.
Int J Signal and Imaging Systems Engineering. Article in press
27. Takala V, Ahonen T, Pietikainen M (2005) Block-Based Methods for Image Retrieval Using Local Binary Patterns.
LNCS 3450:882891
28. Marko H, Pietikainen M, Cordelia S (2009) Description of interest regions with local binary patterns. Pattern
Recognition 42:425436
29. Baochang Z, Zhang L, Zhang D, Linlin S (2010) Directional binary code with application to PolyU near-infrared face
database. Pattern Recognition Letters 31:23372344
30. Corel-5000 Corel image database, [Online]. Available: http://www.ci.gxnu.edu.cn/cbir/Dataset.aspx
31. MIT Vision and Modeling Group, Vision Texture, [Online], Available: http://vismod.media.mit.edu/pub/

doi:10.1186/s13673-014-0006-x
Cite this article as: Vipparthi and Nagar: Color Directional Local Quinary Patterns for Content Based Indexing and
Retrieval. Human-centric Computing and Information Sciences 2014 4:6.

Submit your manuscript to a

journal and benet from:
7 Convenient online submission
7 Rigorous peer review
7 Immediate publication on acceptance
7 Open access: articles freely available online
7 High visibility within the eld
7 Retaining the copyright to your article

Submit your next manuscript at 7 springeropen.com