You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

Mapping without the sun
Zhang, Jixian

Lu Dehao
Huazhong University Science and Technology, Control Science & Engineering, Wuhan, China
KEY WORDS: foreign fibers, multi-wavelength imaging system, optimal light energy, optimal wavelength, image fusion
The objective of this research was to develop a multiwavelength imaging system (MIS) for detecting foreign fibers in the spectral
region from 405 nm to 940 nm. After determined an optimal wavelength for detecting a certain kind of foreign fiber, the
multiwavelength images were captured using a multispectral CCD camera at different optimal wavelength source of illumination. An
image fusion algorithm based on wavelet analysis was created to acquire complete information on foreign fibers. Imaging Results
showed that the combination wavelengths of 405 nm and 850 nm were the most appropriate for detection of a wide range of foreign
fibers in cotton.
The contamination of foreign fibers with polypropylene, hair,
plastic film, which looks virtually white in their appearance as
cotton fiber is a major problem in the textile industry [1.2]. As
the cotton goes through physical processing, the foreign fibers
are fibrillated finer and finer and thus become more difficult
to detect and remove. Currently, although the method of
manual sorting foreign fibers is widely used, it is
time-consuming, labor-intensive and is not as reliable. Then,
developing an automated foreign fiber detection system is
urgent for providing better quality and reducing potential
economic losses in textile field. Recently, various approaches
based on visible light image analysis have been tried for
detecting these foreign fibers. For example, attempts were
made by P. Tantaswad and Tae Jin Kang using color image
analysis method to detect impurities in cotton [3.4]. However,
the use of color scanner and more sophisticated image
analysis technique was failed to detect the colorless foreign
fibers and accordingly it was not successful in assessing the
grades of cotton [5.6]. Research on fiber type identification by
J.S.Church and Wu RongHong has been carried out using
NIR absorption characteristic analysis [7.8]. Although this
method is effective to test the single type of fiber sample, it is
ineffective to separate the foreign fiber from the background
of cotton. Ajay Pai proposed the utilization of x-ray
microtomography system to detect cotton trash [9], but it is
hardly accepted because of the high-cost measurement system,
and therefore, sale promotion of this system is not successful
in most Chinese cotton factories.
In this study we have investigated how multispectral imaging
can be used for detection of foreign fibers from cotton. Since
foreign fibers present different reflectance values in
comparison to those from cotton fibers in different
wavelength, we determined the optimal wavelengths for
detection of each type of foreign fiber using spectral imaging.
Furthermore, the optimal imaging parameters such as
wavelength and light energy of the LED array source of
illumination was quantitatively analysis for achieve obvious
image features of foreign fibers. The results indicate that the
image features of foreign fibers differ sufficiently from those
of cotton fibers in the multispectral imaging system, and the
method shows promising results.
In general, the foreign fibers were divided into two categories,
the colored foreign fibers and colorless foreign fibers.
Colorless foreign fibers cannot be distinguished visually from
cotton, but are of a different type of substance from cotton. A
frequently occurring foreign fiber is, for example, white hair
look virtually white in their nature form. However, the
chemical composition between cotton and white hair are
different, so it is possible to distinguish the two kinds of fibers
from one another at specific wavelength by their different
specific absorption of this band light source. Then it is
reasonable to hypothesis that absorption characteristic may be
used as an effective discriminable feature between these
For detection of these foreign fibers, the light range can be in
the UV (200-400 nm), VIS (400-700 nm), or NIR
(700-2500nm). When radiation from the lighting system
illuminates the fibers, it is transmitted through, reflected, or
absorbed. These phenomena are referred to as optical
properties. Here, the optical properties are integrated
functions of the wavelength of the incident light and chemical
and physical composition of the fibers. In a reflectance
imaging system, the spectral reflectance characteristic of
fibers was transformed to image feature by CCD imaging. By
selecting the appropriate wavelength and light energy of the
spectrum, it is possible to separate foreign fibers from cotton
on the basis of their distinct image feature in a spectral
imaging system.
In a reflectance imaging system, the measured values of
image feature
are a strong function of wavelength
and light energy
parameters are cons
Rbr dach specific ty
opt anithé opt
contrast opl bei
in the following eqi
g 0j
Thus, we can de
including the optin'
for any type of dete
foreign fibers are m
At different optii
different optimal
detected foreign fit
foreign fibers is ^
get the optimal ima;
Then the fused infc
considered, which
range of foreign fil
fibers were enhance
the wavelet-based i
I =F
opt V
where > I op!
and Kavelet 1S
Thus, a multiwavi
implemented by th
each acquired imag
into the single imaj
can greatly enhance
from the backgroun
Since the image fe;
wavelength and lig]
wavelength and o
foreign fibers. Thai
optimal operating
system to detect the:
The light source us(
was LED array, w
visible and NIR rej
wavelength at 405m
and 940nm were a\
sufficient sensitive t