Embed Size (px)
DESCRIPTION
newspaper
Citation preview
Detection of Contaminants in Cotton by using YDbDr Color Space
Gurjinder Singh Sahdra
Department of Computer Sci. and Tech.
111LPU University, Punjab, India
Abstract
Contaminants in cotton plays important role for its
poor quality. The digital image processing teqniques
based on computer vesion provides a good way to
eliminate such contaminants from cotton. This paper
presents a robust algorithm for detection of
contaminants in cotton. By converting the RGB
images to YDbDr color space. YDbDr is composed of
three components Y, Db and Dr. Y is the luminance,
Db and Dr are the chrominance componants. By
using YDbDr its helpful to distinguish luma and
chroma information from RGB image. Iterative
threshold segmentation is applied to binarize the
image Y, Db, Dr. Performing fusion technique to the
three binary images can fuse Luminance and
chrominance image information together. For
machine vision system. YCbCr and HSI color space
has been implemented previously. In which the
performance parameters like speed and accuracy of
detection and also the detection of small
contaminants or insects has to be improved. It is
proved by experimental results that the YDbDr is
effective and credible.
Keywords: cotton contaminants; detection;
YCbCr; HSI; YPbPr; YDbDr
1. Introduction Contamination has vital role in deciding the quality
of cotton apart from essential properties such as
length, strength, fineness. Contamination of raw
cotton can take place at every step i.e. from the farm
picking to the ginning stage. Contamination, even if
it is a single foreign fibre, can lead to the
downgrading of yarn, fabric or garments or even the
total rejection of an entire batch and can cause
irreparable harm to the relationship between growers,
Kamaljot Singh Kailey
Department of Computer Sci. and Tech.
LPU University, Punjab, India [email protected]
ginners, merchants, spinner and textile and clothing
mills [1]. An International Textile Manufacturers
Federation (ITMF) reported that claims due to
contamination amounted to between 1.4 – 3.2% of
total sales of 100% cotton and cotton blended yarns
[2]. A fairly large number of cotton fibers recognition
researches are based on RGB color space and YCbCr
color space. However, color information is not well
presented and extracted in RGB color space due to its
limitations and some deficiencies.
Reference [1] proposed Identification of
Cotton Contaminants Using Neighborhood Gradient
Based on YCbCr Color Space. Reference [2]
proposed HSI color space to detect Foreign Fibers
And Cotton Contaminants. In this paper we separate
luminance and chrominance information in YDbDr
color space. And design a cotton contaminants and
foreign fibers detection algorithm. Thresholding
iteration method is used to image binarizing process.
2. Design of the Proposed System 2.1 Diagram of Contaminants Detection
System. The diagram of cotton contaminants detection
system is shown in fig. 1, which consists of two main
parts. First part is cotton image collection part and
the second part is the contaminants detection part. In
first part we collect images of raw cotton layers on
the time of ginning and images are converted into
digital signal through A/D conversion and sent to
image processing system. Where needed features are
extracted according to Luminance and chrominance
by applying YDbDr color space and after the needed
processing of the images of cotton the resulted image
will be produced which will collect the information
about contaminants or non contaminants. Finally
after processing images will be printed in printer.
Gurjinder Singh Sahdra et al ,Int.J.Computer Technology & Applications,Vol 3 (3), 1118-1124
IJCTA | MAY-JUNE 2012 Available [email protected]
1118
ISSN:2229-6093
Figure 1: Diagram of cotton contaminants detection
system
2.2 Algorithm for Detection of Cotton
Contaminants. Our proposed method firstly transforms the RGB
color space images into YDbDr color space to obtain
luminance component Y and chrominance
components Db and Dr. after getting three different
images Y, Db and Dr. The iterative threshold
algorithm is used for image binarization of three
images Y, Db and Dr. after binarization the three
binary images are fused to get the final fused image.
The final fused image is obtained as the resulted
image of the contaminants detection algorithm.
Figure 2: Flowchart of cotton contaminants detection
algorithm.
3. Analysis of Cotton Contaminants After analyzing plenty of cotton impurity images,
we find that the gray value of cotton fibers is white or
mainly concentrates on a range near to it under
natural lighting conditions. Due to the different
illumination intensity, cotton fiber properties and
level, the position of cotton fiber
pixel have some difference in color space. On the
whole, the pixels in color space meet some certain
distributions under certain light. Common foreign
fiber contaminants in color space have the following
several distributions.[1]
Color is darker near to black, and the gray
value is low, such as black hair and particles
from cottonseed, etc.
Color is much lighter and easy to reflect
with high brightness, such as plastic fiber
and mulch, etc.
Color is not pure, but deviate into a kind of
color, such as colored cotton thread and
polypropylene fiber, etc.
Color is near to cotton fiber, such as wool
and white hair, etc.
4. Selection of Color Space There are different types of color spaces exit. All
the color spaces are for different applications.
Selecting the appropriate color space is the primary
stage for color image processing. Proper color space
can not only save calculation, but also avoid missing
useful information as far as possible.[2]
4.1 RGB Color Space RGB color space is the most fundamental and
commonly used color space of image processing.
Color information initially collected by image
acquisition devices is RGB value, which is also
finally used by color display devices. RGB model
uses three basic components values of R, B and G to
represent color. In this system, any color calculated is
all within the RGB colorized cube. However, RGB
color space has great shortcomings, the main one of
which is that it is not intuitionistic, so it is hard for us
to know color’s cognitive attributes expressed by a
value from its RGB value. Then, RGB color space is
one of the most uneven color spaces, as the visual
difference between two colors cannot be expressed as
the distance between two color points. In addition,
the correlation between RGB is much high, and RGB
space is sensitive to noise in low intensity area.[1]
4.2 YCbCr Color Space YCbCr, Y′CbCr, or Y Pb/Cb Pr/Cr, also written as
YCBCR or Y′CBCR, is a family of color spaces used as
a part of the color image pipeline in video and digital
photography systems. Y′ is the luma component and
CB and CR are the blue-difference and red
Gurjinder Singh Sahdra et al ,Int.J.Computer Technology & Applications,Vol 3 (3), 1118-1124
IJCTA | MAY-JUNE 2012 Available [email protected]
1119
ISSN:2229-6093
difference chroma components. Y′ (with prime) is
distinguished from Y which is luminance, meaning
that light intensity is non-linearly encoded
using gamma correction. Y′CbCr is not an absolute
color space; rather, it is a way of encoding
RGB information. The actual color displayed
depends on the actual RGB primaries used to display
the signal. Therefore a value expressed as Y′CbCr is
predictable only if standard RGB primary
chromaticities are used.[1]
4.3 HSI Color Space It is more often to think about a color in terms of
hue and saturation than in term of additive or
subtractive components. hue is the attribute of the
visual sensation to one of the perceived colors; red,
yellow, green and blue or combination of two of
them. Intensity it is the total amount of light passing
through a particular area. Saturation represents the
purity of color. HSL and HSV are the two most
common cylindrical-coordinate representations of
points in an RGB color model, which rearrange the
geometry of RGB in an attempt to be more
perceptually relevant than the Cartesian
representation.[2] 4.4 YPbPr Color Space YPBPR
is a color space used in video electronics, in
particular in reference to component video cables.
YPBPR is the analog version of the YCBCR color
space the two are numerically equivalent, but
YPBPR is designed for use in analog systems whereas
YCBCR is intended for digital video. YPBPR cables
are also commonly referred to as Yipper cables.
YPBPR is commonly called "component video", but
this is imprecise, as there are many other types of
component video, mostly RGB which syncs either on
green or one or two separate signals.
4.5 YDbDr Color Space
In this paper for our proposed algorithm we using
YDbDr color space. YDbDr is composed of three
components Y, Db and Dr. Y is the luminance, Db
and Dr are the chrominance components. The three
component created from an original RGB (Red,
Green, Blue) source. The weighted values of RG and
B are added together to produce a single Y signal,
representing the overall brightness, or luminance, of
that spot. The Db signal is then created by subtracting
the Y from the blue signal of the original RGB, and
then scaling, and Dr by subtracting the Y from the
red, and then scaling by a different factor.
These formulae approximate the conversion between
the RGB color space and YDbDr.
R,G,B,Y ∈ [0,1]
Db, Dr ∈ [-1.333,1.333]
RGB to YDbDr :
Y = + 0.299R + 0.587G + 0.114B
Db = - 0.450R – 0.883G + 1.333B
Dr = - 1.333R + 1.116G + 0.217B
RGB image YDbDr Image
Figure 3: RGB to YDbDr Conversion
These are the two images the first image is RGB
image of cotton and the second image is the
conversion of RGB to YDbDr color space.
Y component Db component Dr component
Figure 4: Separation of layers Y,Db and Dr
These are the three images of three different
components of YDbDr color space. As Y component
Db component and Dr component.
5. Experiments and Results 5.1 Experiments in YCbCr Color Space
RGB image YCbCr image
Figure 5: RGB to YCbCr conversion
Fig. 5 shows the conversion of RGB to YCbCr color
space.
.
Gurjinder Singh Sahdra et al ,Int.J.Computer Technology & Applications,Vol 3 (3), 1118-1124
IJCTA | MAY-JUNE 2012 Available [email protected]
1120
ISSN:2229-6093
Y component Cb component Cr component
Figure 6: separation of Y, Cb and Cr components
Fig. 6 shows the separation of Y, Cb and Cr
components of YCbCr color space.
Y binary Cb binary Cr binary
Figure 7: binarization of three components
Fig. 7 shows the binary images of YCbCr color
space. These images are further used for fusion to
obtain the final result of the detection algorithm.
Figure 8: final fused image in YCbCr color space
Figure 8 shows the final resulted fused image of
YCbCr color space. It detect all three contaminants
but it just shows the outlines of the big leaf
contaminant the inner part of the contaminant is not
detected by YCbCr color space. This is the biggest
drawback of the YCbCr color space. And also the
clarity of the final fused image is not as per the
requirement. From our more experiments we
observed that the YCbCr color space is not well
suited for the detection of colors like brown, yellow
and many other light colors. YCbCr is only suited for
dark colors. This is another drawback of YCbCR
color space.
5.2 Experiments in HSI Color Space
RGB image HSI image
Figure 9: RGB to HSI conversion
Fig. 9 shows the conversion of RGB to HSI color
space.
H component S component I component
Figure 10: separation of H, S, I components
Fig. 10 shows the separation of three layers of HSI
color space
H binary S binary I binary
Figure 11: binarization of H, S, I
Figure 12: final fused image in HSI color space
Fig. 12 shows the final fused image of HSI color
space. It failed to detect all contaminants as we can
saw it misses two of the upper side contaminants. So
Gurjinder Singh Sahdra et al ,Int.J.Computer Technology & Applications,Vol 3 (3), 1118-1124
IJCTA | MAY-JUNE 2012 Available [email protected]
1121
ISSN:2229-6093
it is proved that HSI color space is only able to detect
large contaminants but its not suitable for detection
of small contaminants in cotton.
5.3 Experiments in YPbPr Color Space
RGB image YPbPr image
Figure 13: RGB to YPbPr Conversion
Fig. 13 shows the conversion of RGB to YPbPr color
space.
Y component Pb component Pr component
Figure 14 separation of Y, Pb, Pr Components
Fig. 14 shows the separation of three layers of YPbPr
color space.
Y binary Pb binary Pr binary
Figure 15: binarization Y, Pb, Pr
Fig. 15 shows the binary images of YPbPr.
Figure 16: final fused image in YPbPr color space
Fig. 16 shows the final fused image of YPbPr color
space. It detect all three contaminants from the cotton
image. And also with more accuracy than YCbCr
color space and HSI color space.
5.4 Experiments in YDbDr Color Space
RGB image YDbDr image
Figure 17: shows the conversion of RGB to YDbDr
color space
Y component Db component Dr component
Figure 18: separation of YDbDr components.
Fig. 18 shows three layers of YDbDr color space
separately as Y, Db and Dr.
Y binary Db binary Dr binary
Figure 19: binarization of YDbDr
Fig. 19 shows the binary images of Y, Db, Dr
Figure 20: final fused image in YDbDr color space
Gurjinder Singh Sahdra et al ,Int.J.Computer Technology & Applications,Vol 3 (3), 1118-1124
IJCTA | MAY-JUNE 2012 Available [email protected]
1122
ISSN:2229-6093
Fig. 20 shows the final fused and resulted image of
YDbDr color space. It detect all the contaminants
from the cotton image. With more accuracy than
YCbCr color space and HSI color space. And with
more clarity than YPbPr color space.
6. Comparison We compare these four color spaces on the bases of
four parameters like time taken, variance, standard
deviation, and mean. The comparison results are as. Figure 21 clearly shows the difference between these
four color spaces as we can saw that YDbDr color
space takes less time then the other three color spaces
HSI takes more time among all other color spaces.
Figure 21: Time taken by all color spaces
Figure 22: mean difference between color spaces
Fig. 21 shows the mean difference among all the
color spaces. Figure shows that mean is almost equal.
YCbCr and HSI are little bit on the higher side.
While YPbPr and YDbDr are both little bit less than
YCbCr and HSI color space.
Figure 23: variance difference between all color
spaces
Fig. 21 shows the variance of all the color spaces. As
we can saw that the variance difference of YDbDr
color space is on the higher side compared to other
three color spaces. YCbCr and HSI both color spaces
are on the lower side. As we compared them with
YPbPr and YDbDr color spaces. YPbPr is also one
half as compared with YDbDr color space.
Figure 23: standard deviation difference between all
color spaces.
Gurjinder Singh Sahdra et al ,Int.J.Computer Technology & Applications,Vol 3 (3), 1118-1124
IJCTA | MAY-JUNE 2012 Available [email protected]
1123
ISSN:2229-6093
Fig. 23 shows the standard deviation difference
between the four color spaces. As we can saw that
standard deviation of YDbDr is on the higher side
then the other three color spaces.
7. Conclusion The paper presents the implementation and
comparative analysis of the YDbDr, YPbPr, YCbCr
and HSI color spaces for the detection of
contaminants from the cotton. One of the main
objective of this paper is to detect small contaminants
or insects from the cotton with more clarity which
was not possible in YCbCr and HSI color space.
Graph shows the comparison between these four
color spaces. on the basis of time taken, mean,
variance, and standard deviation. Which shows that
YDbDr is better than other three color spaces.
Figures also shows that the small contaminants and
insects are more accurately detected in YDbDr color
space. Compared to Previously implemented color
spaces YCbCr and HSI. Various experiments has
been carried out on different images of cotton having
different contaminants like grass, bark insects, fibres
of different materials and colors like red, green,
black, yellow etc. the performance of this algorithm
in YDbDr color space is also better than other
previously implemented algorithms.
8. Future Scope In future the implementation of this algorithm will
be purposed with any neural network technique for
classification. For automatic detection of
contaminants in real time.
References
[1] Chengliang Zhang, Xianying Feng, Lei Li, Yaqing Song (2010) “Identification of Cotton Contaminants Using
Neighborhood Gradient Based on YCbCr Color Space,”
2nd International Conference on Signal Processing Systems
(ICSPS), V(3), pp. V3-733 - V3-738 [2] Pooja Mehta; Naresh Kumar. “Detection Of Foreign
Fibers And Cotton Contaminants By using Intensity And
Hue Properties”, International Journal of Advances in
Electronics Engineering Publication year 2011. Page(s): 861 – 871
[3] Liu Shaojun, Wang Ku, "Detection of Damaged
Cottonseeds Using Machine Vision," ifita, vol. 2, pp.503-
506, 2009 International Forum on Information Technology and Applications, 2009
[4] Wenzhu Yang, Daoliang Lia, Xinhua Wei, Yuguo
Kang, Futang Li, “An automated visual inspection system
for foreign fiber detection in lint,” Computers and Electronics in Agriculture Volume 74 Issue 2, November,
2010 Pages 274-279
[5] Gaili Gao, Daoliang Li*, Shouhai Shang, Shulin Hou, “Accurate Time Control of Eliminating the Cotton Foreign
Fiber,” World Automation Congress (WAC), 2010 pages
235 – 240
[6] Rongfang Yuan; Beihai Zhou; Hui Qi; Huimin Yin;
“Bioinformatics and Biomedical Engineering,” 2009. ICBBE2009. 3rd International conference on Digital Object
Identifier: 10.1109/ICBBE.2009.5163239 Publication Year
2009, Page(s) 1-5
[7] Li Jingjing; Jiang Xiuming; Yang Gongyuan; “Electronic Measurement & Instruments (ICEMI),” 2011
10th International Conference on Volume: 2 Digital Object
Identifier: 10.1109/ICEMI.2011.6037764 Publication Year:
2011 , Page(s): 55 – 59 [8] Yupeng Zhang; Smith, P.W.; “Applications of
Computer Vision,” 2002. (WACV 2002). Proceedings.
Sixth IEEE Workshop on Digital Object Identifier:
10.1109/ACV.2002.1182156 Publication Year: 2002 Page(s): 51 – 56
[9] Jia Dong Yao; Ding Tianhuai; “Infrared and Millimeter
Waves,” 2004 and 12th International Conference on
Terahertz Electronics, 2004. Conference Digest of the 2004 Joint 29th International Conference on Digital Object
Identifier: 10.1109/ICIMW.2004.1422310 Publication
Year: 2004 , Page(s): 751 – 752
[10] A. Pai, H. Sari-Sarraf, E.F. Hequet, “Recognition of cotton contamination via X-ray microtomographic image
analysis,” IEEE Transactions on Industry Applications,
vol. 1, 2004, pp.77-85
[11] Tianhuai Ding, Xin Qu, “A real-time detection method for foreign fibers in cotton,” Chinese patent,
CN101403703, April,08,2009.
[12] Xianying Feng, Chengliang Zhang, Bingsheng Yang
and Lei Li,“Foreign fiber recognition and detection algorithm based on RGB color space,” Journal of
Shandong University(Engineering Science), Vol39, pp.
68-72, 2009.
[13] Caixia Guo, Jianchang Yuan, Shoufeng Jin, “Recognition algorithm of the foreign fibers in cotton by
BP neural networks,” Journal of Xi'an University of
Engineering Science and Technology, Vol. 20, pp. 542-
544, 2006. [14] Dongyao Jia and Ding Tianhuai, "Detecting foreign
fibres in cotton using a multi-spectral technique,"
Tsinghua Science and Technology, Vol. 45, pp. 193-196,
2005. [15] X in and Huai “ A Fast Feature Extraction Algorithm
for Detection of Foreign Fibre in Lint Cotton within a
Complex Background,” vol. 36, No 6 june, 2010
[16] Boshra D.Farah, “Measurements of trash contents and
grades in cotton using digital image analysis”. Proceedings
of ICSP’96 1545-1548
[17] Siddaiah, Murali, “Identification of trash types in
ginned cotton using neuro fuzzy techniques,” IEEE International Conference on Fuzzy Systems, v 2, 1999,
738-743
Gurjinder Singh Sahdra et al ,Int.J.Computer Technology & Applications,Vol 3 (3), 1118-1124
IJCTA | MAY-JUNE 2012 Available [email protected]
1124
ISSN:2229-6093
