Вы находитесь на странице: 1из 19

Hindawi Publishing Corporation

EURASIP Journal on Advances in Signal Processing


Volume 2008, Article ID 783898, 12 pages
doi:10.1155/2008/783898

Research Article
Fabric Defect Detection Using Modified Local Binary Patterns

F. Tajeripour,1 E. Kabir,1 and A. Sheikhi2


1 Department of Electrical Engineering, Tarbiat Modarres University, P.O. Box 14115-111, Tehran, Iran
2 Department of Electrical and Electronics Engineering, Shiraz University, P.O. Box 71348-51154, Shiraz, Iran

Correspondence should be addressed to Farshad Tajeripour, tajeri@modares.ac.ir

Received 24 December 2006; Revised 22 May 2007; Accepted 4 October 2007

Recommended by Liang-Gee Chen

Local binary patterns (LBPs) are one of the features which have been used for texture classification. In this paper, a method based
on using these features is proposed for fabric defect detection. In the training stage, at first step, LBP operator is applied to an
image of defect free fabric, pixel by pixel, and the reference feature vector is computed. Then this image is divided into windows
and LBP operator is applied to each of these windows. Based on comparison with the reference feature vector, a suitable threshold
for defect free windows is found. In the detection stage, a test image is divided into windows and using the threshold, defective
windows can be detected. The proposed method is multiresolution and gray scale invariant and can be used for defect detection
in patterned and unpatterned fabrics. Because of its simplicity, online implementation is possible as well.

Copyright 2008 F. Tajeripour et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

1. INTRODUCTION detect specific types of the defects. Therefore, detection speed


and the range of the detectable defects are two main issues in
Defect detection is an important problem in fabric quality the field of automatic fabric inspection.
control process. Cost reduction in production and inspection Many attempts have been made in the past three decades
process is also an important objective for textile manufac- to solve these problems. These attempts have been based on
turers. At present the quality inspection process is manually three dierent approaches: statistical, spectral, and model
performed by experts. Typical fabrics are 13 m wide and are based. In statistical approach, gray-level texture features ex-
driven with speeds ranging from 20 to 200 m/min. Experts tracted from cooccurrence matrix [12], mean and standard
cannot detect more than 60% of the overall defects if the fab- deviations of subblocks [13], autocorrelation of subimages
ric moves faster than 30 m/min or wider than 2 m [1]. Like [14], and Karhunen-Loeve (KL) transform [15] have been
other inspection processes, it has depended on workers ex- used for the detection of fabric defects. Bodnarova et al.
perience until now. The development of a flexible, ecient, [16] made use of normalized cross-correlation functions for
reliable, and integrated real-time vision system for industrial detecting defects in fabrics. There exist many model-based
application is an essential issue in quality control process for techniques for fabric defect detection. For example, Cohen et
textile manufacturers. In particular, if there is a defect, it re- al. [17] used a Markov random field (MRF) model for defect
duces the price of the fabric by 45%65%. To increase the inspection of fabrics. Chen and Jain [18] used a structural
overall quality, the homogeneity of fabric, and reliability, an approach to detect defects in textured images. Atalay [19] has
automated visual inspection system is needed for better pro- implemented an MRF-based method on TMS320C40 paral-
ductivity. Therefore, automation of visual inspection tasks lel processing system for real-time defect inspection of fab-
can increase the eciency of production lines and improve rics. Methods that use low-order MRF are not capable of de-
quality of the products as well. The previous works in the tecting all kinds of the defects in fabric texture. In order to
field of automatic defect detection are mainly done on paper detect all kinds of the defects, the order of the model should
[2], steel roll [3], wood [4, 5], carpet [6], and textile [711]. be increased. This yields an increase in computational com-
Most of the automatic fabric inspection systems are oine plexity of the algorithm. There also exist many spectral ap-
and have detection speed up to 100 m/min. An important proaches for fabric defect detection. For example, Kumar
point regarding these systems is that each of them can only and Pang [20] proposed a method for defect detection using
2 EURASIP Journal on Advances in Signal Processing

Unpatterned plain fabric Large repetitive unit


(flower) P = 4, R = 1 P = 8, R = 1 P = 12, R = 1.5
Fabric
(a) (b) (c)

Figure 2: Circularly symmetric neighborhoods for dierent P and


R [24].
Patterned fabric
Dot patterned fabric
classification. This method is based on recognizing that cer-
tain LBP features are fundamental properties of local image
Repetitive pattern
texture and their occurrence histograms are proven to be a
very powerful texture feature [24]. LBP is a highly discrimi-
Figure 1: Classification of fabrics [23].
native texture operator. It records the occurrence of various
patterns in neighborhood of each pixel in P-dimensional his-
tograms. Therefore, this method is used for detecting textural
Gabor filters which needs a large amount of computations. defects in fabric. The proposed method is simple, multires-
They also developed a method for defect detection using only olutional, and invariant to gray scale. Experimental results
imaginary part of Gabor filters. Chan and Pang [21] oered show that a wide range of the defects can be detected through
a method for defect detection in textile fabrics using Fourier this method. This method is applicable for defect detection
analysis. Since Fourier bases are of infinite length, the con- in both unpatterned and patterned fabrics which have a re-
tribution from each of the spectral components is dicult peated and periodic texture.
to quantify. Therefore, Fourier analysis is not suitable for This paper is organized as follows: in Section 2, local bi-
detecting local defects. Kumar and Pang [22] developed a nary patterns are described in its basic and modified versions.
method for defect detection in textile fabrics using optimized In Section 3, the proposed method for defect detection is pre-
filters. sented. Section 4 is devoted to implementation and experi-
It should be noted that most of the researches about fab- mental results and the conclusion is provided in Section 5.
ric defect detection are made on unpatterned fabrics and
only a few works for defect detection in patterned fabrics
have been reported. For example, Ngan et al. [23] used a 2. USING LBP IN TEXTURE CLASSIFICATION
wavelet-based method for defect detection in patterned fab-
rics. A patterned fabric is defined with repetitive patterned One of the methods used in texture classification is LBPs
units in its design. Under the class of patterned fabric, there [24]. In this method, a neighborhood of the image is consid-
are many categories. Patterned fabrics that are used in this ered and the gray value of the pixel in the center is compared
research are Jacquard patterned fabrics. In these types of fab- with the gray values of the other pixels in the neighborhood.
rics a flower or a graphical logo may appear on the fabric. Usually the neighborhood is in circular form and the gray
The repetitive unit can range from the simplest charter box, values of the neighbors which do not fall exactly in the center
dots, to the most complicated multiple flower, animals, or of pixels are estimated by interpolation. Figure 2 illustrates
other designed patterns. Besides there are a lot of subcate- circularly symmetric neighbor sets for various radii, R, and
gories under patterned fabric. Figure 1 illustrates a classifica- dierent numbers of neighbors, P.
tion of fabrics [23]. In basic form of this method, LBP operator in one neigh-
The researches in this area can be divided into two dif- borhood of the image is defined as follows [24]:
ferent categories. In the first category, all attempts are con-
ducted to increase the range of the defects to be detected, 1


P
  1, x 0,
while they need a large amount of computations. References LBPP,R = s gi gc 2i , s(x) = (1)
i=0 0, x < 0.
[2022] belong to this category. In the second one, increasing
detection speed is the aim, while a restricted range of defects
can be detected. References [1219] belong to the second cat- In (1), the gray value of the central pixel is gc and the gray
egory. value of the ith neighbor is gi . According to this definition, it
In this paper, a simple and straightforward method for is seen that the output of the operator is a P bit binary num-
detecting irregularities in fabric texture is proposed, which ber with 2P distinct values. On the other hand, the value of
can detect a wide range of the defects. In this method, local the output is quite dependent on the labeling of the neigh-
binary patterns (LBPs) are used. It should be noted that LBP bors. When the image is rotated, since the neighborhood is
is used for texture classification [24] but in this paper for the considered in circular form, neighbors will correspondingly
first time it is used for detecting textural defects in fabric. LBP move along the perimeter of the circle. In order to make the
is theoretically very simple, yet ecient approach for texture algorithm invariant to rotation and assign a unique value to
F. Tajeripour et al. 3

each neighborhood, the output of LBP operator is rotated 6 12 10 1 1 1


and the minimum value is selected: Thresholding
1
7 5 2 0
    1 1 1
LBPriP,R = min ROR LBPP,R , i | i = 0, 1, . . . , P 1 . (2) 5 9 5

In (2), a P-bit number is rotated i times and the minimum (a)


value between resulting numbers for i between 0 to P 1 is
selected. In (2), ROR is the abbreviation of rotate right. 1 1 1 0 0
10 12 13 7 7
In modified version of LBP [24], at first a uniformity 5 5 6 7 10 Thresholding
0 1
measure, U, is defined as the number of spatial transitions 1 1
8 8 8 9 9 1 1
between 1 s and 0 s in the pattern. Then patterns that have 10 11 13 15 10 1 1 1 1 1
uniformity measure less than UT are defined as uniform pat- 9 9 9 8 9
terns. The modified LBP is defined as follows: (b)
P 1

   Figure 3: LBPs for two square neighborhoods: (a) wm = 3, P = 8
s g p gc if U UT ,
LBPriu
P,R
T
=
p=0 (3) and (b) wm = 5, P = 16.


P+1 otherwise.

Equation (3) shows that modified LBP assigns labels from 0 encountering these labels, as a key feature for our proposed
to P to uniform neighborhoods and label P + 1 to nonuni- defect detection method. Experimental results show that if
form ones. After applying this operator to the image, the the probability of encountering label P + 1 which is assigned
probability of encountering a specific label can be approxi- to nonuniform patterns is small (less than 1%) these features
mated by the ratio of the number of neighbors which have can classify the texture correctly, otherwise dierent patterns
that label to the number of all neighbors. Therefore, at the in the texture take the same label (P + 1) and cannot be clas-
end of this process P+2 probabilities will be computed. These sified.
probabilities can be used as powerful features for texture clas- Our simulation results show that if in the definition of
sification. For classification task, the log-likelihood ratio is LBP operator the value of UT is selected equal to P/4, only
used. The sample under test belongs to class K if the com- a negligible portion of the patterns in the texture takes label
puted probabilities minimize the following ratio: P + 1.
In the subsequent sections, the proposed method for de-

P+1
Si fect detection in unpatterned and patterned fabrics will be
L(S, K) = Si log . (4) presented. Output of the proposed algorithm is a binary im-
i=0
M iK
age which is called defect pattern. Black pixels in the defect
In (4), MiK is the probability of encountering label i in the pattern represent nondefective areas of the fabric and white
patterns of class K, and Si is the probability of encountering pixels represent defective areas. The size of the defect pattern
label i found from the sample under test. According to (3), is the same as the size of the input image.
it is obvious that any monotonic change in gray values does
not change the pattern and this method is invariant to gray 3.1. Defect detection in unpatterned fabrics
scale changes.
For detecting defects in unpatterned fabrics, in training stage,
3. THE PROPOSED ALGORITHM FOR LBP operator is applied to the whole image of a defect free
TEXTILE DEFECT DETECTION fabric and reference feature vector, M, is computed. Each el-
ement of this vector is the probability of encountering la-
In this section, a new method for defect detection in fab- bels 0 to P + 1 in the defect free image. If the number of
rics, based on a modified version of LBP, is presented. In points in LBP (number of pixels in the neighborhood) op-
basic and modified versions of LBP, selecting neighborhood erator is P, the reference feature vector which is called M will
in circular form is to make the algorithm invariant to rota- be P + 2 dimensional. After computing M, the image of the
tion. Since during inspection process, rotation of fabric rolls defect free fabric is divided into nonoverlapping windows of
can be avoided, selecting circular neighborhood is not neces- size Wd Wd which are called detection windows. Then LBP
sary. On the other hand, computing brightness using inter- operator is applied to each of these windows. The window
polation in circular neighborhood takes a lot of time. There- for applying LBP operator is called LBP mask. Each element
fore, in our proposed method, a square neighborhood is con- of the feature vector computed for a window is the proba-
sidered. In this case, the notation of LBPs is renamed from bility of encountering a specific label in that window. In or-
LBPP,R to LBPP,wm , where the size of the window for apply- der to estimate these probabilities accurately, the number of
ing LBP operator is wm wm pixel. Figure 3 illustrates square operators applied to each detection window should be large
neighborhoods and how to apply the LBP operator. Using enough (as a rule of thumb, at least 100 operators in each
LBP operator, a label from 0 to P + 1 is assigned to each window). If the size of the detection window is Wd Wd
neighborhood of the image. These labels reflect the relation and the size of LBP mask is wm wm (Wd  wm ), then the
between a pixel and its neighbors. We use the probability of number of operators in each window will be (Wd wm + 1)2 .
4 EURASIP Journal on Advances in Signal Processing

Therefore, if the minimum number of operators applied in windows. In this case, if the size of LBP mask is wm wm
each window is 100, then Wd 9 + wm . pixel, overlapping step of wm 1 between detection win-
It should be noted that the features extracted by LBP op- dows is sucient to take into consideration the interaction
erator can describe fabric texture correctly if the textures ap- between all pixels in dierent detection windows (interaction
peared in detection windows are similar to fabric texture. between pixels in the last column and the last row of a win-
Therefore, the size of the detection window (Wd Wd ) cre- dow with pixels in the first column and the first row of the
ated on the image should be greater than the size of the repet- adjacent window). Using these types of windows, the thresh-
itive unit of fabric texture. However, in unpatterned fabrics old is computed as in (5) and (6). The remaining stages are as
the only condition for window size is Wd 9 + wm . As the for unpatterned fabrics. It should be noted that this method
size of the detection window increases, the capability of the is a multiresolution and the results of selecting dierent win-
algorithm in detecting small defects and the resolution of the dows can be combined easily as follows:
defect pattern decrease. By applying LBP operator to each of
these windows, vector Sk which is P + 2 dimensional is com- 
N
 
puted. The log-likelihood ratio for each of these windows will LNK = LK SnK , M n , (7)
n=1
be computed as follows:
where N is the number of windows (neighborhood) selected
  
P+1
Sik
Lk Sk , M = Sik log , k = 1, 2, . . . , N. (5) for applying LBP operator and K is the index of the test-
i=0
Mi ing window. The overall block diagram of the proposed al-
gorithm is shown in Figure 4.
In (5), N is the number of detection windows. Since the min- It should be mentioned that, in the training stage, if the
imization of log-likelihood ratio shows the similarity to a number of defect free sample is Ndf , then the reference fea-
specific class, the maximum value between these ratios will ture vector can be estimated more accurately using
be used as a threshold for defect-free windows as follows:
Ndf
  i=1Mi
T = Max Lk , k = 1, 2, . . . , N. (6) M= , (8)
Ndf
After computing reference feature vector M and the
threshold T, in the detection stage, the test image is divided where Mi is the reference feature vector computed for sam-
into the detection windows and log-likelihood ratio is com- ple i. After computing the reference feature vector, M, the
puted for each of these windows. If log-likelihood ratio ex- threshold is computed using
ceeds the threshold, the relevant window belongs to the de-  
fective areas of the fabric. In order to increase the detection T = max Ti , (9)
i
capability of the algorithm, a large area of the detection win-
dow should be occupied by the defect. Therefore, in the de- where Ti is the threshold computed for sample i using (5)
tection stage, image is divided into overlapping windows. Ac- and (6) and computed M.
cording to the simulation results, if the overlapping step of
the detection windows is Wd /2, the proposed algorithm has 4. IMPLEMENTATION AND RESULTS
appropriate detection power. Increasing overlapping step will
decrease detection speed. In this research, two types of unpatterned fabrics, twill and
plain, are used. Six dierent types of the defects like dou-
3.2. Defect detection in patterned fabrics ble yarn, missing yarn, broken fabric, weft crack, float, and
knots in these types of fabrics are considered. These defects
For detecting defects in patterned fabrics which have a peri- are shown in Figure 5. All these fabrics have yarns of dierent
odic texture, in training stage, as it was mentioned in the pre- thicknesses, and their warp and weft direction thicknesses are
vious sections, at first LBP operator is applied to the whole also dierent. Detection results for unpatterned fabrics are
image of a defect free sample then the image is divided into also shown in Figure 5.
the detection windows and LBP operator will be applied to In patterned fabrics, six dierent types of defects that
each of these windows. Since the fabrics are patterned if the usually appear in fabric texture are considered. These defects
size of the window is less than the size of the repetitive unit are shown in Figure 6. Detection results for patterned fabrics
in the fabric texture, then the texture in the detection win- are also shown in Figure 6. Images are 8-bit gray scale images
dows will be dierent. So, the size of the detection window of size 256 256. These images are also used in [23, 26]. The
should be much greater than the repetitive unit of the fab- resolution of the camera used for these images is 200 dpi.
ric texture. Increasing the size of the detection window will For detecting defects, dierent sizes of detection windows
increase computational complexity and decreases the resolu- were tested and the size of 16 16 yielded the best results.
tion of the defect pattern. In order to solve this problem, the In patterned fabrics the size of the repetitive unit is nearly
size of the detection window will be chosen a little greater 16 16. As explained in Section 3.1, in order to estimate the
than the size of the repetitive unit in the fabric texture and in probability of encountering dierent labels (0 to P + 1) ac-
order to take into account the interaction between all pixels curately, the size of the detection windows should be large
with their neighbors, the image is divided into overlapping enough. It is also dependent to P. Increasing the size of the
F. Tajeripour et al. 5

Applying LBP Calculating reference


mask on image feature vector, M
Defect free image

Dividing image Computing log likelihood Finding the


into detection ratio, L for each detection
windows window threshold, T

(a) Training stage

M T

Dividing image Computing log likelihood If L > T, label window as


Test sample into detection ratio, L for each detection
defective else non defective
windows window using M

(b) Testing stage

Figure 4: Overall block diagram of the proposed algorithm.

(a) (b) (c) (d) (e)

Figure 5: (a) Images of defective unpatterned fabrics from top to bottom: double yarn, missing yarn, broken fabric, weft crack, float, and
knot. Detection results in the form of defect pattern using (b) LBP8,3 , (c) LBP16,5 , (d) LBP24,7 , and (e) LBP8,3+16,5 .

detection windows decreases the capability of the algorithm mated accurately. Since the maximum size of LBP mask in
in detecting small size defects. Experimental results show that the proposed algorithm is 7 7, the size of 16 16 for detec-
if there are at least 100 operators in each detection window, tion windows is sucient for estimating these probabilities
the probability of encountering dierent labels can be esti- accurately.
6 EURASIP Journal on Advances in Signal Processing

(a) (b) (c) (d) (e)

Figure 6: (a) Images of defective patterned fabrics from top to bottom: dirty yarn, hole, thick bar, broken end, netting multiple, knot.
Detection results in the form of defect pattern using (b) LBP8,3 , (c) LBP16,5 , (d) LBP24,7 , and (e) LBP8,3+16,5 .

Table 1: Detection rate (%) for dierent defect types of patterned fabrics: (a) dirty yarn, (b) oil stain, (c) broken end, (d) netting multiple,
(e) hole, and (f) knots.

Defect types
Operator P, w Number of features
(a) (b) (c) (d) (e) (f)
8,3 10 97 95 95 92 96 95
16,5 18 96 94 94 N.D1 95 N.D
24,7 26 97 95 94 N.D 95 N.D
LBPP,w 8, 3 + 16, 5 10 + 18 98 98 98 95 96 95
8, 3 + 24, 7 10 + 26 98 96 96 94 95 97
16, 5 + 24, 7 18 + 26 97 95 94 N.D 95 N.D
8, 3 + 16, 5 + 24, 7 10 + 18 + 26 98 98 98 96 97 97
1 Not detected, when the number of detected defective windows is less than half of the number of true defective windows.

One of the ways to measure the performance of defect de- where Ncc is the number of defect free windows which are de-
tection algorithms is calculating the detection rate [25] which tected as nondefective (true negative). Ndd is the number of
is defined as follows: defective windows which are detected as defective (true pos-
itive). Ntotal is the total number of detection windows that
is created on the image. Table 1 illustrates the detection rate
Ncc + Ndd for dierent types of the defects using dierent LBP masks.
DR = 100 , (10)
Ntotal For computing the detection rate, defect pattern generated
F. Tajeripour et al. 7

Table 2: Detection rate (%) for dierent defect types of unpatterned fabrics: (a) Double Yarn, (b) Missing Yarn, (c) Broken Fabric, (d) Weft
Crack, (e) Float, and (f) Knots.

Defect types
Operator P, w Number of features
(a) (b) (c) (d) (e) (f)
8,3 10 97 85 95 92 94 95
16,5 18 98 90 93 90 94 93
24,7 26 95 N.D 93 90 93 90
LBPP,w 8, 3 + 16, 5 10 + 18 98 98 98 95 97 97
8, 3 + 24, 7 10 + 26 98 96 96 94 95 95
16, 5 + 24, 7 18 + 26 97 N.D 94 92 95 94
8, 3 + 16, 5 + 24, 7 10 + 18 + 26 99 98 98 96 97 97

(a) (b) (c) (d)

Figure 7: (a) Images of defective fabrics, (b) defect pattern generated by the method of [26], (c) defect pattern generated by proposed
algorithm (overlapping step 8 pixels), and (d) defect pattern generated by proposed algorithm (overlapping step 15 pixels).

by the algorithm is divided into 16 16 windows and a win- type of the defects there exist five dierent images in our data
dow that contains at least one white pixel is considered as set. The detection rate listed in Table 1 is the average of the
defective. Since images are of size 256 256, Ntotal is 256. In detection rates for all of the defective images in the database.
our image dataset of patterned fabrics, there exist sixty im- In Table 2 detection rates for dierent types of the defects in
ages of dot patterned Jacquard fabrics. Half of these images unpatterned fabrics are listed.
are defect free. Training stage is done using only five sam- As shown in Figures 5 and 6, a 3 3 mask for apply-
ples of defect free images. In testing stage all of the defect free ing LBP operator, LBP8,3 , detects almost all kinds of the de-
samples are classified correctly. On the other hand, for each fects, but the defect pattern found for patterned fabrics is not
8 EURASIP Journal on Advances in Signal Processing

Table 3: Detection rate (%) of proposed algorithm and algorithm of [26] tested on dot patterned fabrics: (a) dirty yarn, (b) hole, (c) oil
stain, (d) knot, (e) broken end, and (f) netting multiple.

Defect type
Algorithm Average detection rate
(a) (b) (c) (d) (e) (f)
Reference [26] 96 97 94 95 95 94 95.1
Proposed1 98 95 97 94 97 93 95.8
Proposed2 98 96 98 94 98 95 96.5
1
Overlapping step 8 pixels.
2
Overlapping step 15 pixels.

continuous and exact in unpatterned fabrics. Therefore, the


features of 3 3 mask are combined with the features of 5 5
and 7 7 masks using (7). On the other hand, using 5 5
and 7 7 (LBP16,5 and LBP24,7 ) masks for detection as illus-
trated in Table 1 do not detect defects like netting multiple
and knots. Table 1 shows that combination of features of 3 3
Broken end
and 5 5 masks (LBP8,3+16,5 ) has an appropriate detection
rate (more than 95%).

4.1. Performance comparison

In the subsequent subsections, the performance of the pro-


posed algorithm is compared with the performance of simi- Holes
lar algorithms. For performance comparison two criteria are
used:
(1) defect detection accuracy,
(2) computational complexity and speed.
For comparing the detection accuracy of the algorithms, Netting multiple
the defect patterns generated by dierent algorithms are
compared visually. The detection rate is also used to compare
the detection accuracy.
In order to compare the computational complexity, the
number of operations required for processing a test sample
is considered.
Thick bar

4.1.1. Patterned fabrics

In order to make a comparison between the proposed


method and the existing methods for defect detection in pat-
terned fabrics, the results of the proposed method are com-
pared with the results of methods in [26, 27]. These methods Thin bar
are the two newest methods for detecting defects in patterned (a) (b) (c)
fabrics. In the training stage of the method of [26], a win-
dow of a defect free sample is selected. The size of this win- Figure 8: Detection results for star patterned fabrics: (a) defective
samples, (b) detection results of the proposed algorithm, and (c)
dow should be greater than the size of the repeated part of
detection results of the method of [27].
fabric texture. This window will be moved on a defect free
sample image pixel by pixel. At each point, dierence be-
tween gray values of pixels in the window and gray values
of pixels in the underlying window on the image is com- stain, and broken end, the proposed method generates more
puted. The average of absolute value of dierences is com- accurate defect pattern. So, the proposed method and the
puted for each point. By defining a suitable threshold, de- method in [26] have almost the same accuracy. The method
fective points can be detected. Images of patterned fabrics of [26] requires a large amount of computations and it is
in both methods are the same. Comparison of the results not suitable for online implementation. However, the pro-
in Figure 7 reveals that in some specific types of the defects posed method due to its simplicity can be used for online
like knot and hole, the method of [26] yields more accurate defect detection. It should be noted that the resolution of
defect pattern; and in some other types like dirty yarn, oil the defect pattern generated by our proposed algorithm can
F. Tajeripour et al. 9

Table 4: Detection rate (%) of proposed algorithm and algorithm of [27] tested on star patterned fabrics: (a) thin bar (dirty yarn), (b) hole,
(c) thick bar (oil stain), (d) broken end, and (e) netting multiple.

Defect type
Algorithm Average detection rate
(a) (b) (c) (d) (e)
Reference [27] 95 95 94 93 96 94.6
Proposed1 97 97 96 98 96 96.8
Proposed2 98 97 99 98 95 97.4
1
Overlapping step 12 pixels.
2
Overlapping step 24 pixels.

Table 5: Detection rate (%) of proposed algorithm and algorithm of [27] tested on box patterned fabrics: (a) thin bar (dirty yarn), (b) hole,
(c) thick bar (oil stain), (d) broken end, and (e) netting multiple.

Defect type
Algorithm Average detection rate
(a) (b) (c) (d) (e)
Reference [27] 95 97 94 94 94 94.8
Proposed1 97 96 98 98 95 96.8
Proposed2 98 96 98 98 98 97.6
1
Overlapping step 12 pixels.
2
Overlapping step 24 pixels.

be increased by increasing the overlapping of the detection windows will be 961. In each detection window, for ap-
windows. Moving windows pixel by pixel on the image will plying LBP8,3 , 1568 comparisons, 10 multiplications, and
generate a defect pattern like those in [26]. It is shown in 10 divisions are required. For applying LBP16,5 , 2304 com-
the fourth column of Figure 7. However, increasing the over- parisons, 18 multiplications, and 18 divisions are required.
lapping between detection windows will increase the com- Therefore, to process a test sample by the proposed method
plexity of the algorithm. Table 3 illustrates the detection rate (LBP16,5+8,3 ), 3 720 992 comparisons, 26 098 multiplications,
computed for our proposed algorithm and algorithm of [26]. and 26 098 divisions are required.
The method of [27] uses Bollinger bands for detecting defects It should be noted that for computing log-likelihood
in patterned fabrics. Bollinger bands consist of three bands: ratio, the log() operation can be omitted and approxi-
upper, middle, and lower. In this method, patterned fabrics mated log-likelihood ratio can be used. In approximated log-
can be considered as comprising many rows (columns), with likelihood ratio, log(x) is approximated as x 1 so approxi-
a pattern designed on each row (column). The principle of mated log-likelihood ratio (L) can be computed by
this method is that the patterned rows (columns) will gen-
erate periodic upper and lower bands. Any defect region in 
P+1 
P+1
Si Si
patterned fabric means that there would be a break of pe- L(S, K) = Si log Si 1
i=0
MiK i=0
MiK
riodicity in the pattern. For better evaluation performance
of the proposed algorithm, it has been tested on two other 
P+1
S2i 
P+1 
P+1
S2i

types of patterned fabrics named as star patterned and box = Si = 1 (11)
i=0
MiK i=0 i=0
MiK
patterned. Detection results of the proposed algorithm and
method of [27] are compared in Figures 8 and 9. The size of 
P+1
S2i
 K) =
L(S, .
a repetitive unit for both fabrics is 25 25. Therefore, the size i=0
MiK
of 25 25 is selected for detection windows. The overlapping
step is 12 pixels. Tables 4 and 5 illustrate the detection rate Simulation results show that using approximated log-
of the proposed algorithm and algorithm of [27] tested on likelihood ratio does not change the detection rate of the pro-
star patterned and box patterned fabrics. In the method of pose algorithm.
[26] if the size of repetitive unit of fabric texture is 16 16 In the method of [27] if the size of repetitive unit of fab-
and the size of test sample is 256 256 pixel, moving the de- ric texture is 25 25 in each row (column) of the test sample
tection windows on test sample pixel by pixel yields 58 081 of size 256 256, 232 segments of length 25 pixels can be
dierent positions. For each position of detection window in considered. In each segment, 48 additions, 25 subtractions,
this method, 256 subtractions, 255 additions, 256 compar- 25 multiplications, 2 divisions, and one square root are re-
isons, and one division are required. Therefore, processing a quired. Table 6 illustrates the number of operations which
test sample of size 256 256 pixel by this method requires are required for processing a test sample of size 256 256,
14 868 736 subtractions, 14 810 655 additions, 14 868 736 using proposed method and methods of [26, 27].
comparisons, and 58 081 divisions. As shown the computational complexity of the proposed
In the proposed method if overlapping step between algorithm can be reduced by reducing the overlapping step
detection windows is 8 pixels, total number of detection between detection windows and reducing the resolution of
10 EURASIP Journal on Advances in Signal Processing

Table 6: Number of dierent operations which are required to process a test sample using dierent algorithms.

Algorithm Size of repetitive unit of fabric texture Addition subtraction Multiplication comparison division Square root
Proposed1 16 16 23064 0 26098 3720992 26098 0
Proposed2 25 25 8424 0 9072 3657312 9072 0
[26] 16 16 14810655 14868736 0 14868736 58081 0
[26] 25 25 33586176 33640000 0 33640000 53824 0
[27] 16 16 3701760 1974272 1974272 0 123392 123392
[27] 25 25 5701632 2969600 2969600 0 118784 118784
1
Overlapping step of 8 pixels between detection windows.
2
Overlapping step of 12 pixels between detection windows.

Broken end

Holes

Netting multiple

Thick bar

Thin bar
(a) (b) (c)
(a) (b) (c)
Figure 10: (a) Defective fabric images, (b) detection results using
Figure 9: Detection results for box patterned fabrics: (a) defective
our proposed algorithm, and (c) detection results using Gabor fil-
samples, (b) detection results of the proposed algorithm, and (c)
ters.
detection results of the method of [27].

the defect pattern, which cannot be done in similar algo- posed algorithm are compared with those of the defect de-
rithms. tection using Gabor filters [28]. Defect detection using Ga-
bor filters is one of the most accurate methods [29]. In this
4.1.2. Unpatterned fabrics method [28], an image of a defect free sample is passed
through a bank of Gabor filters. Transfer function of each
For evaluating performance of the algorithm in detecting de- filter is obtained by changing scale and orientation of the Ga-
fects of unpatterned fabrics, the detection results of the pro- bor functions. In the method selected for comparison, there
F. Tajeripour et al. 11

exist 16 filters in the filter bank, which are obtained by chang- ACKNOWLEDGMENTS
ing the orientation and scale of Gabor functions to four ori-
entations and four scales. Output of each filter is thresholded The authors would like to thank Henry Y. T. Ngan and Indus-
and combined with the output of the other filters using an trial Automation Research Laboratory, Department of Elec-
image fusion technique. So, for each filter in the filter bank, a trical and Electronic Engineering, The University of Hong
set of suitable parameters are computed. Passing test sample Kong, for providing the database of patterned fabrics. They
through these filters will generate a defect pattern that detects also would like to thank S. Arivazhagan for providing some
and localizes the irregularities in fabric texture. Figure 10 il- samples of unpatterned fabrics.
lustrates the detection results of the proposed algorithm and
the detection results using Gabor filters. Some of the images
of unpatterned fabrics were scanned from album of standard REFERENCES
defects in textile industry [30] and the others were collected
[1] C.-S. Cho, B.-M. Chung, and M.-J. Park, Development of
from textile mills around TamilNadu state of India. These
real-time vision-based fabric inspection system, IEEE Trans-
images are gray-scale images with 8 bits long and with the actions on Industrial Electronics, vol. 52, no. 4, pp. 10731079,
size of 256 256. If we use a procedure for computing detec- 2005.
tion rate like what was used for patterned fabrics, the aver- [2] J. W. Roberts, S. D. Rose, G. A. Jullian, et al., PC based real-
age detection rate for proposed algorithm will be 97.1% and time defect imaging system for high speed web inspection, in
for Gabor filters will be 98%. The methods that have been Machine Vision Applications in Industrial Inspection, vol. 1907
based on Gabor filters use the idea of multichannel filtering of Proceedings of SPIE, pp. 164176, San Jose, Calif, USA,
which takes a lot of time. Therefore, these methods have a February 1993.
little chance for online implementation. [3] J. Laitinen, Image quality in automated visual web inspec-
tion, in Machine Vision Applications in Industrial Inspection
V, vol. 3029 of Proceedings of SPIE, pp. 7889, San Jose, Calif,
5. CONCLUSION USA, February 1997.
[4] R. W. Conners, C. W. McMillin, K. Lin, and R. E. Vasquez-
In this paper, a new method for detecting textural defects Espinosa, Identifying and locating surface defects in wood:
in fabrics based on modified local binary patterns has been part of an automated lumber processing system, IEEE Trans-
actions on Pattern Analysis and Machine Intelligence, vol. 5,
presented. The proposed method has an acceptable detec-
no. 6, pp. 573583, 1983.
tion rate (more than 95%) in all kinds of the defects which
[5] E. Young, Use of linescan cameras and a DSP processing sys-
is more than that of human experts in both unpatterned and
tem for high-speed wood inspection, in Machine Vision Ap-
patterned fabrics. Furthermore, detecting all kinds of the de- plications, Architectures, and Systems Integration IV, vol. 2597
fects needs considering dierent windows which can be easily of Proceedings of SPIE, pp. 259264, Philadelphia, Pa, USA,
done using the multiresolution property of this method. October 1995.
This method is a useful tool for textile industries to in- [6] L. Siew, R. M. Hodgson, and E. J. Wood, Texture measures for
spect, identify, and localize local defects in textile products. carpet wear assessment, IEEE Transactions on Pattern Analysis
The proposed method is simple, gray-scale invariant, and and Machine Intelligence, vol. 10, no. 1, pp. 92105, 1988.
multiresolution. According to the performance comparison [7] S. H. Sheen, H. T. Chien, W. P. Lawrence, and A. C. Raptis,
and implementation results, it is evident that the proposed Ultrasonic imaging system for in-process fabric defect detec-
method can detect and localize defects in patterned fabrics, tion, U.S. Patent no. 5665 907, September 1997.
more accurately than similar algorithms. The computational [8] L. Dorrity and G. Vachtsevanos, In-process fabric defect de-
complexity of the proposed algorithm is less than similar al- tection and identification, in Proceedings of Mechatronics,
gorithms. Therefore, it can be used in online defect detection. Skovde, Sweden, September 1998.
In spite of low-resolution defect pattern, the defect pat- [9] C.-H. Chan and G. K. H. Pang, Fabric defect detection by
tern generated by the proposed method is suitable for a wide Fourier analysis, IEEE Transactions on Industry Applications,
range of the defects. It should be mentioned that, practically, vol. 36, no. 5, pp. 12671276, 2000.
the detection rate and detection speed are of higher privilege [10] H. Sari-Sarraf and J. S. Goddard Jr., Vision system for on-
than the resolution of the defect pattern generated by the al- loom fabric inspection, IEEE Transactions on Industry Appli-
cations, vol. 35, no. 6, pp. 12521259, 1999.
gorithm.
[11] A. Kumar and G. Pang, Fabric defect segmentation using
The proposed method is implemented on three types
multichannel blob detectors, Optical Engineering, vol. 39,
of patterned fabrics: dot patterned, box patterned, star pat- no. 12, pp. 31763190, 2000.
terned and two types of unpatterned fabrics: twill and plain. [12] I.-S. Tsai, C.-H. Lin, and J.-J. Lin, Applying an artificial neu-
Six dierent types of patterned fabric defects and six dierent ral network to pattern recognition in fabric defects, Textile
types of unpatterned fabrics are considered in this research. Research Journal, vol. 65, no. 3, pp. 123130, 1995.
Implementation results show that the proposed method is [13] X. F. Zhang and R. R. Bresee, Fabric defect detection and
capable of detecting all kinds of the defects in both types of classification using image analysis, Textile Research Journal,
the fabrics. It should be noted that this method can be used vol. 65, no. 1, pp. 19, 1995.
for defect identification too. It can also be used for defect de- [14] E. J. Wood, Applying fourier and associated transforms to
tection in products which have regular and periodic textures pattern characterization in textiles, Textile Research Journal,
such as timber or plastic. vol. 60, no. 4, pp. 212220, 1990.
12 EURASIP Journal on Advances in Signal Processing

[15] M. Unser and F. Ade, Feature extraction and decision proce-


dure for automated inspection of textured materials, Pattern
Recognition Letters, vol. 2, no. 2, pp. 181191, 1984.
[16] A. Bodnarova, M. Bennamoun, and K. K. Kubik, Defect de-
tection in textile materials based on aspects of the HVS, in
Proceedings of IEEE International Conference on Systems, Man
and Cybernetics (SMC 98), vol. 5, pp. 44234428, San Diego,
Calif, USA, October 1998.
[17] F. S. Cohen, Z. Fan, and S. Attali, Automated inspection of
textile fabrics using textural models, IEEE Transactions on
Pattern Analysis and Machine Intelligence, vol. 13, no. 8, pp.
803808, 1991.
[18] J. Chen and A. K. Jain, A structural approach to identify de-
fects in textured images, in Proceedings of IEEE International
Conference on Systems, Man, and Cybernetics (SMC 88), vol. 1,
pp. 2932, Beijing, China, August 1988.
[19] A. Atalay, Automated defect inspection of textile fabrics using
machine vision techniques, M.S. thesis, Bogazici University,
Istanbul, Turkey, 1995.
[20] A. Kumar and G. K. H. Pang, Defect detection in textured
materials using Gabor filters, IEEE Transactions on Industry
Applications, vol. 38, no. 2, pp. 425440, 2002.
[21] C.-H. Chan and G. K. H. Pang, Fabric defect detection by
Fourier analysis, IEEE Transactions on Industry Applications,
vol. 36, no. 5, pp. 12671276, 2000.
[22] A. Kumar and G. K. H. Pang, Defect detection in textured
materials using optimized filters, IEEE Transactions on Sys-
tems, Man, and Cybernetics, Part B, vol. 32, no. 5, pp. 553570,
2002.
[23] H. Y. T. Ngan, G. K. H. Pang, S. P. Yung, and M. K. Ng,
Defect detection on patterned jacquard fabric, in Proceed-
ings of the 32nd Applied Imagery Pattern Recognition Work-
shop (AIPR 03), pp. 163168, Washington, DC, USA, October
2003.
[24] T. Ojala, M. Pietikainen, and T. Maenpaa, Multiresolution
gray-scale and rotation invariant texture classification with lo-
cal binary patterns, IEEE Transactions on Pattern Analysis and
Machine Intelligence, vol. 24, no. 7, pp. 971987, 2002.
[25] A. Bodnarova, M. Bennamoun, and K. K. Kubik, Suitability
analysis of techniques for flaw detection in textiles using tex-
ture analysis, Pattern Analysis and Applications, vol. 3, no. 3,
pp. 254266, 2000.
[26] H. Y. T. Ngan, G. K. H. Pang, S. P. Yung, and M. K. Ng,
Wavelet based methods on patterned fabric defect detection,
Pattern Recognition, vol. 38, no. 4, pp. 559576, 2005.
[27] H. Y. T. Ngan and G. K. H. Pang, Novel method for patterned
fabric inspection using Bollinger bands, Optical Engineering,
vol. 45, no. 8, Article ID 087202, 15 pages, 2006.
[28] S. Arivazhagan, L. Ganesan, and S. Bama, Fault segmentation
in fabric images using Gabor wavelet transform, Machine Vi-
sion and Applications, vol. 16, no. 6, pp. 356363, 2006.
[29] A. Latif-Amet, A. Ertuzun, and A. Ercil, Ecient method for
texture defect detection: sub-band domain co-occurrence ma-
trices, Image and Vision Computing, vol. 18, no. 6-7, pp. 543
553, 2000.
[30] Graniteville Company, Manual of Standard Fabric Defects in
the Textile Industry, Graniteville, SC, USA, 1975.
EURASIP Journal on Bioinformatics and Systems Biology

Special Issue on
Applications of Signal Processing Techniques to
Bioinformatics, Genomics, and Proteomics

Call for Papers


The recent development of high-throughput molecular ge- Guest Editors
netics technologies has brought a major impact to bioin-
Erchin Serpedin, Department of Electrical and Computer
formatics, genomics, and proteomics. Classical signal pro-
Engineering, Texas A&M University, College Station, TX
cessing techniques have found powerful applications in ex-
77843-3128, USA; serpedin@ece.tamu.edu
tracting and modeling the information provided by genomic
and proteomic data. This special issue calls for contributions Javier Garcia-Frias, Department of Electrical and
to modeling and processing of data arising in bioinformat- Computer Engineering, University of Delaware, Newark, DE
ics, genomics, and proteomics using signal processing tech- 19716-7501, USA; jgarcia@ee.udel.edu
niques. Submissions are expected to address theoretical de-
Yufei Huang, Department of Electrical and Computer
velopments, computational aspects, or specific applications.
Engineering, University of Texas at San-Antonio, TX
However, all successful submissions are required to be tech-
78249-2057, USA; yhuang@utsa.edu
nically solid and provide a good integration of theory with
practical data. Ulisses Braga Neto, Department of Electrical and
Suitable topics for this special issue include but are not Computer Engineering, Texas A&M University, College
limited to: Station, TX 77843-3128, USA; ulisses@ece.tamu.edu
Time-frequency representations
Spectral analysis
Estimation and detection
Stochastic modeling of gene regulatory networks
Signal processing for microarray analysis
Denoising of genomic data
Data compression
Pattern recognition
Signal processing methods in sequence analysis
Signal processing for proteomics
Authors should follow the EURASIP Journal on Bioinfor-
matics and Systems Biology manuscript format described
at the journal site http://www.hindawi.com/journals/bsb/.
Prospective authors should submit an electronic copy of
their complete manuscript through the EURASIP Journal on
Bioinformatics and Systems Biologys Manuscript Tracking
System at http://mts.hindawi.com/, according to the follow-
ing timetable:

Manuscript Due June 1, 2008


First Round of Reviews September 1, 2008
Publication Date December 1, 2008

Hindawi Publishing Corporation


http://www.hindawi.com
EURASIP Journal on Wireless Communications and Networking

Special Issue on
Fairness in Radio Resource Management for
Wireless Networks

Call for Papers


Radio resource management (RRM) techniques (such as ad- RRM framework and QoS architecture
mission control, scheduling, sub-carrier allocation, channel Complexity and scalability issues
assignment, power allocation, and rate control) are essential Experimental and implementation results and issues
for maximizing the resource utilization and providing qual- Fairness in multiple-antenna transmission/reception
ity of service (QoS) in wireless networks. systems
In many cases, the performance metrics (e.g., overall Fairness in end-to-end QoS provisioning
throughput) can be optimized if opportunistic algorithms
are employed. However, opportunistic RRM techniques al- Authors should follow the EURASIP Journal on Wire-
ways favor advantaged users who have good channel condi- less Communications and Networking manuscript for-
tions and/or low interference levels. The problem becomes mat described at the journal site http://www.hindawi.com/
even worse when the wireless terminals have low mobility journals/wcn/. Prospective authors should submit an elec-
since the channel conditions become slowly varying (or even tronic copy of their complete manuscript through the jour-
static), which might lead to long-term unfairness. The prob- nal Manuscript Tracking System at http://mts.hindawi.com/,
lem of fair resource allocation is more challenging in multi- according to the following timetable:
hop wireless networks (e.g., mesh and multihop cellular net-
works). Manuscript Due July 1, 2008
The performance fairness can be introduced as one of
the QoS requirements (e.g., as a condition on the minimum First Round of Reviews October 1, 2008
throughput per user). Fair RRM schemes might penalize ad- Publication Date January 1, 2009
vantaged users; hence, there should be a tradeo between the
overall system performance and the fairness requirements.
We are soliciting high-quality unpublished research pa- Guest Editors
pers addressing the problem of fairness of RRM techniques Mohamed Hossam Ahmed, Faculty of Engineering and
in wireless communication systems. Topics include (but are Applied Science, Memorial University of Newfoundland,
not limited to): St. Johns, NF, Canada A1B 3V6; mhahmed@engr.mun.ca
Fairness of scheduling schemes in wireless networks Alagan Anpalagan, Department of Electrical and
Tradeo between maximizing the overall throughput Computer Engineering, Ryerson University, Toronto, ON,
and achieving throughput fairness Canada M5G 2C5; alagan@ee.ryerson.ca
RRM fairness: problem definition and solution tech-
niques Kwang-Cheng Chen, Department of Electrical
Engineering, National Taiwan University, Taipei, Taiwan;
Fairness performance in emerging wireless systems
chenkc@cc.ee.ntu.edu.tw
(WiMAX, ad hoc networks, mesh networks, etc.)
Cross-layer RRM design with fairness Zhu Han, Department of Electrical and Computer
Short-term and long-term fairness requirements Engineering, College of Engineering, Boise State University
Adaptive RRM to support fairness Boise, Idaho, USA; zhuhan@boisestate.edu
Fairness in cooperative wireless communications Ekram Hossain, Department of Electrical and Computer
Issues and approaches for achieving fairness in multi- Engineering, University of Manitoba, Winnipeg, MB,
hop wireless networks Canada R3T 2N2; ekram@ee.umanitoba.ca

Hindawi Publishing Corporation


http://www.hindawi.com
EURASIP JOURNAL ON ADVANCES IN SIGNAL PROCESSING

Special Issue on
Cross-Layer Design for the Physical, MAC, and Link
Layer in Wireless Systems
Call for Papers
Cross-layer design is a common theme for emerging wire- PHY-aware scheduling and resource allocation
less networks, such as cellular systems and ad hoc networks, Analysis of fundamental tradeos at the PHY/MAC
though the objectives and requirements can be very dier- interface: rate-delay, energy-delay
ent. A cross-layer approach is particularly important when
PHY/MAC/link aspects of network coding in wireless
designing protocols at the physical (PHY) and medium ac-
networks
cess control (MAC)/link layer for several reasons. The past
decade gave birth to innovative techniques such as oppor- PHY/MAC issues in wireless sensor networks and dis-
tunistic communication, rate adaptation, and cooperative tributed source co
coding, which all require close interaction between layers. Cross-layer optimization of current standards
For example, optimizing the rate-delay tradeo in a mul- (WiMAX, W-CDMA, etc.)
tiuser wireless system requires knowledge of PHY parame- Interaction of MAC/PHY design with the circuitry
ters such as channel state information (CSI) as well as delay and the battery management
information from the MAC. As another example, the notion
of link in wireless relay systems is generalized towards a col- Authors should follow the EURASIP Journal on Ad-
lection of several physical channels related to more than two vances in Signal Processing manuscript format described
nodes. The potential for novel solutions through cross-layer at http://www.hindawi.com/journals/asp/. Prospective au-
design is further seen in the fact that it motivates collabora- thors should submit an electronic copy of their com-
tion among researchers of dierent specialties, such as anten- plete manuscript through the EURASIP Journal on Ad-
nas and propagation, signal processing, error-control coding, vances in Signal Processing Manuscript Tracking System
scheduling, and networking (MAC) protocols. at http://mts.hindawi.com/, according to the following
This special issue is devoted to research eorts that use cross- timetable:
layer design to jointly optimize the lowest two layers of the
wireless communication systems. Such cross-layer solutions
Manuscript Due February 1, 2008
will have decisive impact in future wireless networks.
Contributions on the following and related topics are so- First Round of Reviews May 1, 2008
licited:
Publication Date August 1, 2008
Random access protocols with multipacket reception
(MPR), multiantenna MPR protocols
MAC/ARQ protocols for multiantenna and wireless
relay systems GUEST EDITORS:
MAC-level consideration of time-varying channels:
Petar Popovski, Center for TeleInFrastructure, Aalborg
channel estimation, channel state information (CSI),
University, Niels Jernes Vej 12, A5-209, 9220 Aalborg, Den-
and link adaptation
mark; petarp@kom.aau.dk
MAC-level interference mitigation via management
of PHY-level multiuser networks: for example, inter- Mary Ann Ingram, School of Electrical and Computer Engi-
fering multipleaccess (MA) channels neering, Georgia Institute of Technology, 777 Atlantic Drive
MAC/PHY protocols and analysis of cognitive radio NW, Atlanta, GA 30332-0250, USA;
networks maryann.ingram@ee.gatech.edu
MAC protocols for PHY-restricted devices/circuitry Christian B. Peel, ArrayComm LLC, San Jose, CA 95131-
(e.g., RFID) 1014, USA; chris.peel@ieee.org
Shinsuke Hara, Graduate School of Engineering, Osaka
City University, Osaka-shi 558-8585, Japan;
hara@info.eng.osaka-cu.ac.jp
Stavros Toumpis, Department of Electrical and Computer
Engineering, University of Cyprus, 1678 Nicosia, Cyprus;
toumpis@ucy.ac.cy

Hindawi Publishing Corporation


http://www.hindawi.com
EURASIP Journal on Image and Video Processing

Special Issue on
3D Image and Video Processing

Call for Papers


3D image and 3D video processing techniques have received Object recognition and tracking
increasing interest in the last years due to the availability of 3D video/communications
high-end capturing, processing, and rendering devices. Costs Streaming of 3D data
for digital cameras have dropped significantly and they allow Augmented/mixed reality
the capturing of high-resolution images and videos, and thus Image-based rendering
even multi-view acquisition becomes interesting in commer- Free viewpoint video
cial applications. At the same time, processing power and
storage have reached the level that allows the real-time pro- Authors should follow the EURASIP Journal on Im-
cessing and analysis of 3D image information. The synthe- age and Video Processing manuscript format described
sized 3D information can be interactively visualized on high- at http://www.hindawi.com/journals/ivp/. Prospective au-
end graphics boards that become available even on small mo- thors should submit an electronic copy of their com-
bile devices. 3D displays can additionally help to increase the plete manuscripts through the EURASIP Journal on Im-
immersiveness of a 3D framework. All these developments age and Video Processing Manuscript Tracking System
move the interest in 3D image and video processing meth- at http://mts.hindawi.com/, according to the following
ods from a more academic point of view to commercially at- timetable:
tractive systems and enable many new applications such as
3DTV, augmented reality, intelligent human-machine inter- Manuscript Due March 1, 2008
faces, interactive 3D video environments, and much more.
Although the technical requirements are more and more First Round of Reviews June 1, 2008
fulfilled for a commercial success, there is still a need of so- Publication Date September 1, 2008
phisticated algorithms to handle 3D image information. The
amount of data is extremely high, requiring ecient tech-
niques for coding, transmission, and processing. Similarly, Guest Editors
the estimation of 3D geometry and other meta data from
Peter Eisert, Fraunhofer Institute for Telecommunications,
multiple views, the augmentation of real and synthetic scene
Heinrich Hertz Institute, Berlin, Germany;
content, and the estimation of 3D object or face/body motion
eisert@hhi.fraunhofer.de
have been addressed mainly for restricted scenarios but often
lack robustness in a general environment. This leaves many Marc Pollefeys, Department of Computer Science,
interesting research areas in order to enhance and optimize University of North Carolina at Chapel Hill, USA, Institute
3D imaging and to enable fully new applications. Therefore, for Computational Science, ETH Zurich, Switzerland;
this special issue targets at bringing together leading experts marc@cs.unc.edu
in the field of 3D image and video processing to discuss and
Stefano Tubaro, Dipartimento di Elettronica e
investigate this interesting topic.
Informazione, Politecnico di Milano, Milano, Italy;
Specifically, this special issue will gather high-quality, orig-
stefano.tubaro@polimi.it
inal contributions on all aspects of the analysis, interaction,
streaming, and synthesis of 3D image and video data. Topics
of interest include (but are not limited to):
3D image and video analysis and synthesis
Face analysis and animation
3D reconstruction

Hindawi Publishing Corporation


http://www.hindawi.com
EURASIP Journal on Image and Video Processing

Special Issue on
Distributed Video Coding

Call for Papers


Distributed source coding (DSC) is a new paradigm based on manuscripts through the journal Manuscript Tracking Sys-
two information theory theorems: Slepian-Wolf and Wyner- tem at http://mts.hindawi.com/, according to the following
Ziv. Basically, the Slepian-Wolf theorem states that, in the timetable:
lossless case, the optimal rate achieved when performing
joint encoding and decoding of two or more correlated Manuscript Due May 1, 2008
sources can theoretically be reached by doing separate encod-
ing and joint decoding. The Wyner-Ziv theorem extends this First Round of Reviews August 1, 2008
result to lossy coding. Based on this paradigm, a new video Publication Date November 1, 2008
coding model is defined, referred to as distributed video cod-
ing (DVC), which relies on a new statistical framework, in-
stead of the deterministic approach of conventional coding Guest Editors
techniques such as MPEG standards. Frederic Dufaux, Ecole Polytechnique Fdrale de
DVC oers a number of potential advantages. It first al- Lausanne, Lausanne, Switzerland; frederic.dufaux@epfl.ch
lows for a flexible partitioning of the complexity between the
encoder and decoder. Furthermore, due to its intrinsic joint Wen Gao, School of Electronic Engineering and Computer
source-channel coding framework, DVC is robust to channel Science, Peking University, Beijing, China;
errors. Because it does no longer rely on a prediction loop, wgao@pku.edu.cn
DVC provides codec independent scalability. Finally, DVC is Stefano Tubaro, Dipartimento di Elettronica e
well suited for multiview coding by exploiting correlation be- Informazione, Politecnico di Milano, Milano, Italy;
tween views without requiring communications between the stefano.tubaro@polimi.it
cameras.
High-quality original papers are solicited for this special Anthony Vetro, Mitsubishi Electric Research Laboratories,
issue. Topics of interest include (but are not limited to): Cambridge, MA, USA; avetro@merl.com
Architecture of DVC codec
Coding eciency improvement
Side information generation
Channel statistical modeling and channel coding
Joint source-channel coding
DVC for error resilience
DVC-based scalable coding
Multiview DVC
Complexity analysis and reduction
DSC principles applied to other applications such as
encryption, authentication, biometrics, device foren-
sics, query, and retrieval
Authors should follow the EURASIP Journal on Im-
age and Video Processing manuscript format described
at http://www.hindawi.com/journals/ivp/. Prospective au-
thors should submit an electronic copy of their complete

Hindawi Publishing Corporation


http://www.hindawi.com
Research Letters in
Signal Processing
Why publish in this journal?
Research Letters in
Research Letters in Signal Processing is devoted to very fast publication of short, high-
quality manuscripts in the broad field of signal processing. Manuscripts should not exceed Signal Processing
Hindawi Publishing Corporation
4 pages in their final published form. Average time from submission to publication shall http://www.hindawi.com Volume 2007

be around 60 days.

Editorial Board
Why publish in this journal? Tyseer Aboulnasr, Canada
Wide Dissemination T. Adali, USA
All articles published in the journal are freely available online with no subscription or registration S. S. Agaian, USA
barriers. Every interested reader can download, print, read, and cite your article. Tamal Bose, USA
Jonathon Chambers, UK
Quick Publication Liang-Gee Chen, Taiwan
The journal employs an online Manuscript Tracking System which helps streamline and speed P. Dan Dan Cristea, Romania
the peer review so all manuscripts receive fast and rigorous peer review. Accepted articles appear Karen Egiazarian, Finland
online as soon as they are accepted, and shortly after, the final published version is released online Fary Z. Ghassemlooy, UK
following a thorough in-house production process. Ling Guan, Canada

Research Letters in
M. Haardt, Germany
Professional Publishing Services Peter Handel, Sweden
The journal provides professional copyediting, typesetting, graphics, editing, and reference

Signal Processing
Alfred Hanssen, Norway
validation to all accepted manuscripts. Andreas Jakobsson, Sweden
Jiri Jan, Czech Republic
Keeping Your Copyright
Soren Holdt Jensen, Denmark
Authors retain the copyright of their manuscript, which is published using the Creative Commons Stefan Kaiser, Germany
Attribution License, which permits unrestricted use of all published material provided that it is
C. Chung Ko, Singapore
properly cited.
S. Maw Kuo, USA
Extensive Indexing Miguel Angel Lagunas, Spain
Hindawi Publishing Corporation James Lam, Hong Kong
Articles published in this journal will be indexed in several major indexing databases to ensure
thehttp://www.hindawi.com
maximum possible visibility of each published article. Mark Liao, Taiwan Volume 2007
Stephen Marshall, UK
Stephen McLaughlin, UK
Submit your Manuscript Now... Antonio Napolitano, Italy
In order to submit your manuscript, please visit the journals website that can be found at C. Richard, France
http://www.hindawi.com/journals/rlsp/ and click on the Manuscript Submission link in the M. Rupp, Austria
navigational bar. William Allan Sandham, UK
Ravi Sankar, USA
Should you need help or have any questions, please drop an email to the journals editorial office
John J. Shynk, USA
at rlsp@hindawi.com
A. Spanias, USA
ISSN: 1687-6911; e-ISSN: 1687-692X; doi:10.1155/RLSP
Yannis Stylianou, Greece
Jarmo Henrik Takala, Finland
S. Theodoridis, Greece
Hindawi Publishing Corporation Luc Vandendorpe, Belgium
410 Park Avenue, 15th Floor, #287 pmb, New York, NY 10022, USA
Jar-Ferr Kevin Yang, Taiwan