Optics and Photonics Journal, 2013, 3, 229-232
http://dx.doi.org/10.4236/opj.2013.33037 Published Online July 2013 (http://www.scirp.org/journal/opj)
Multispectral Imaging for Authenticity Identification and
Quality Evaluation of Flos carthami*
Cuiying Hu1, Qingxia Meng1#, Ji Ma2, Qichang Pang3, Jing Zhao4
1Department of Physics, Jinan University, Guangzhou, China
2College of Traditional Chinese Medicine, Southern Medical University, Guangzhou, China
3Department of Optoelec tronic Engineering, Jinan University, Guangzhou, China
4Department of Applied Physics, South China Agricultural University, Guangzhou, China
Received May 4, 2013; revised June 11, 2013; revised June 19, 2013
Copyright © 2013 Cuiying Hu 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.
ABSTRACT
The identification and quality evaluation of Flos carthami were studied using tunable liquid spectral imaging instrument,
to discuss the application range and advantages of spectral imaging technology in Chinese medicine identification and
quality control field. The Flos carthami was indentified by extracting the normalized characteristic spectral curves of
Flos carthami, Crocus sativus and Dendranthema morifo lium, which were standard samp les supplied by National Insti-
tute for Drug Control. The qualities o f Flos carthamies collecting from different pharmacies were evaluated by extract-
ing their normalized characteristic spectral curves. The imaging spectrum testing system was designed independently.
The spectral resolution was 5 nm, and the spectral range was from 400 nm to 680 nm. The results showed that the nor-
malized characteristic spectral curve of Flos carthami was significantly different from those of Crocus sativus’ and
Dendranthema morifolium’s, and the fluorescence intensity of Flos carthami from different commercial sources were
different. Spectral imaging technology could be used to identify and evaluate Flos carthami, and operation method was
rapid, convenient and non-destructive.
Keywords: Flos carthami; Spectral Imaging; Rapid Identification; Quality Evaluation
1. Introduction
Flos carthami is the dried flower of Cartham u s tinctoriu s
L. It is commonly used in traditional Chinese medicine. It
contains a variety of ingredients such as flavonoids com-
pounds, phenolic acids, fatty acids, volatile oils, polyace-
tylene, adenosine, and so on. The main effects of Flos
carthami are promoting blood circulation, dilation of
blood vessels, improving microcirculation, eliminating
free radicals, anti-inflammatory and other functions. The
quality of Flos carthami is closely related with its effi-
cacy. High-performance liquid chromatography (HPLC)
can identify Flos carthami and evaluate its quality by
detecting the content of the main component. However,
HPLC methods can not be real-time detection, and it is
time-consuming.
Multispectral imaging is an emerging technology that
integrates conventional imaging and spectroscopy to attain
both spatial and spectral information from a sample.
Multispectral imaging was originally developed for re-
mote sensing applications [1] but has since found appli-
cation in diverse fields such as environment, telemetry,
agriculture and other fields [2-6]. A series of exploratory
studies have been conducted about different kinds of
Chinese he rb al medi cines by our rese arch group [7-11] .
In this paper, The Flos carthami was indentified, and
the qualities of Flos carthamies collecting from different
pharmacies were evaluated using the multispectral imag-
ing technolog y. The result shows th at multis pectral imag-
ing technology provides an objective, time-saving, real-
time detection, non-destructive and simple method for
the identification and quality evaluation of Flos carthami .
*This work was supported by the National Natural Science Foundation
of China under Contract 60908038, Science and Technology Planning
Project of Guangdong Province, China under Contract 2012B0403
02002 and Agricultural Science and Technology Project of Guangzhou
China under Contract GZCQC1002FG080 1 5.
#Corresponding author.
2. Materials and Methods
2.1. Samples Preparation
The standard sample(SS)s of Flos carthami (FC), Crocus
C
opyright © 2013 SciRes. OPJ

C. Y. HU ET AL.
230
sativus (CS) and Dendranthema morifolium (DM) were
supplied by Guangzhou Institute for Drug Control
(GZIDC) on December 8, 2010 and April 27, 2011, re-
spectively. The other Flos carthami samples collected
from Guangzhou Tong Ren Tang pharmacy(GZTRTP),
Guangzhou Bao Jian Tang pharmacy(GZBJTP), Guang-
zhou Er Tian Tang pharmacy(GZETTP), Guangzhou Bao
Zhi Lin pharmacy(GZBZLP), Guangzhou Oriental phar-
macy(GZOP), respectively. The sample details are listed
in Table 1.
2.2. Liquid Crystal Multispectral Imaging
System
The testing instrument was self-designed multispectral
spectrum measurement system [12]. It is composed of the
light source, a light source filter, a liquid crystal tunable
filter (LCTF), the controller of LCTF, lens, CMOS sen-
sor, data acquisition card and data processing software.
The ray path of the testing system is shown in Figure 1.
Table 1. Information of samples.
Sample name Source Collecting time
SS of FC
(batch number: GDIDC 120907-200609) 2010.12.08
SS of CS
(batch number: GZIDC 1009-200101) 2010.12.08
SS of DM
(batch number: GZIDC 120995-200704) 2011.04.27
FC sample 1 GZTRTP 2010.12.12
FC sample 2 GZBJTP 2010.12.12
FC sample 3 GZETTP 2010.12.12
FC sample 4 GZBZLP 2010.12.12
FC sample 5 GZOP 2010.12.12
Figure 1. The ray path of the system.
The centre wavelength of the light source is 254 nm.
The LCTF is an important component of the system. It is
a splitter component based on electrically controlled
birefringence of the liquid crystal. It is used to divide the
light coming from the samples in two dimensions. The
working wavelength of LCTF is from 400 nm to 1100
nm, which is controlled by the controller of LCTF. The
wavebands from 400 nm to 680 nm are chosen in our
experiment.
The samples partially reflect the light coming fro m the
light source after the interaction of light and the sample.
The light passes through LCTF carrying the information
of the samples. The LCTF divides the light and focus it
on the image sensor (CMOS). The images are captured
by the image acquisition card and saved on the host
computer, as JPG format. The two-dimensional spectrum
data can be processed and the results are displayed on the
monitor of the computer.
2.3. Methods
The excitation light sources are two mercury lamps with
the center wavelength of 254 nm in our research. The
bandwidth of each lamp is 30 nm, and its optical power
is 6 w. Single channel, continuous spectrum scan was
used in the detection process. The spectral scan range
was from 400 nm to 680 nm, controlled by the controller
of LCTF. The two adjacent frames interval is 5 nm. The
spectral resolution is up to 0.5 nm. The CMOS imaging
camera was adjusted so that the focal plane coincided
with the surface of the test samples at 55 0 nm waveband.
The CMOS camera was set to continuous mode with the
exposure time of 1000 ms, which was synchronized with
spectral scanning time, and then the fluorescence images
of the test sample were acquired at the whole wavebands.
The captured images were stored in the computer with
jpg format. A spectral cube of the test sample, formed by
57 frame spectral images, can be obtained in one test.
The detected sample was placed on the substrate with-
out any pre-treatment, and the two-dimension images of
it at a number of narrow wavebands can be obtained.
Removed noise in the images with a bandpass filter. Se-
lected the same area in every image to calculate the ave-
rage light intensity of the corresponding pixel and to
normalize them according to Equation (1), then the cha-
racteristic spectral curve of the sample was obtained.
Figure 2 shows the normalized spectral curve of Flos
carthami

 
1
max
,
N
in iii
n
i iiNormalizedi
II
II
NI





 (1)
where, N is the number of the pixel,

ii
I

is average
light intensity of the ith image, , 1, 2,, 57imaxi
I
is
the biggest light intensity among the 57 frame images.

C. Y. HU ET AL. 231
Figure 2. Normalized characteristic spectral curve of Flos
carthami.
2.4. Methodological Study
1) Stability test
The same sample is tested five times under the same
condition according to part C in the section II, and the
interval is 24-hour between two times. Extract the char-
acteristic spectral curves from the 5 times and compare
them. The similarity of the 5 curves peak shape (meas-
ured by its covariance) is greater than 0.95, and the posi-
tions of characteristic peak remain unchanged. The un-
certainty of characteristic peaks fluctuations in light in-
tensity is less than of the measurements. It shows
that the samples have good stability in the detection.
%86.1
2) Precision test
Repeat measuring the same sample five times accord-
ing to part C in the section II under the same condition.
Extract the characteristic spectral curves from the 5 times
and compare them. The similarity of the 5 curve peak
shape is greater than 0.95, and the positions of characte-
ristic peak remain unchanged. The uncertainty of charac-
teristic peaks fluctuations in light intensity is less than
of the measurements. It shows that the imaging
system has good precision.
1.86%
3) Reproducibility test
The same sample is divided into 5 parts. Every part is
tested using the same system under the same condition
according to part C in the Section II. Five characteristics
spectra curves can be obtained and compared. The simi-
larity of the 5 curve peak shape is greater than 0.95, and
the positions of characteristic peak remain unchanged.
The uncertainty of characteristic peaks fluctuations in
light intensity is less than of the measurements.
It shows that this method has good reproducibility.
1.86%
3. Result and Discussion
3.1. Characteristic Spectral Curves of FC, CS
and DM
The standard samples of FC, CS and DM were detected
according to Sections 2.3 and 2.4. Figure 3 shows their
normalized characteristic spectral curves. It indicates that
the normalized characteristic spectral curves are signi-
ficant differences each other. The general trends and
characteristics peaks are completely different. The cha-
racteristic spectral curve of FC has partial peak structure,
and the peak position is at 610 nm with peak height of
0.73. The characteristic spectral curve of DM is close to
the normal distribution in the 450 - 650 nm wavebands,
and the peak position is at 530 nm with peak height of
0.64. The fluorescence intensity of CS increased slowly
in the detection wavebands, there was no peak. FC and
DM are both composites, FC and CS in effect has si-
milarities. However, the characteristic spectral curves of
the three traditional Chinese medicines are significantly
different. The result indicates that spectral imaging me-
thod can distinguish between different types of flower
herbs.
3.2. Quality Evalu at io n of the Flos carthami
Samples
The FC samples listed in Table 1 were tested according
to Sections 2.3 and 2.4. Figure 4 shows their normalized
characteristic spectral curves. As can be seen from Fig-
ure 4, the general trends and peak positions of the char-
acteristic spectral curves are exactly the same, therefore,
they are all genuine Flos carthami. The changes in fluo-
rescence intensity reflect freshness and quality differ-
ences between the FC samples. These results are consis-
tent with the results of physical and chemical identifica-
tion under double-blind conditions. The purity levels of
No.1-No.4 samples are similar; however, No.5 sample is
impurity. In addition, from the fluorescence intensity of
view, the intensity of No.1 sample is highest, No.5 sam-
ple’s is lowest, and No.2 to No.4 samples’ is close. These
are also consistent with the results of experience identi-
fication, that is, No.1 sample is the freshest one, No.5
Figure 3. Normalized characteristic spectral curves of FC,
CS and DM.

C. Y. HU ET AL.
232
Figure 4. Normalized characteristic spectral curves of FC,
samples.
sample is the least fresh one, and the freshness of No.2 to
No.4 sample is in the middle.
4. Conclusion
The multispectral imaging method is developed to iden-
tify authenticity and evaluate quality of Flos carthami in
this paper. Compared to chemical methods, the multis-
pectral imaging method has more advantages: rapidity,
simplicity, safety, low operational costs and samples be-
ing tested directly using the system without any pre-
treatment. The measuring process is time-saving and the
results are steady, precis e and repeatab le. Th e result shows
that the multispectral imaging method is an effective,
nondestructiv e technique, and can be us ed to iden tify and
evaluate the traditional Chinese herbal medicine pow-
ders.
REFERENCES
[1] A. F. H. Goetz, G. Vane, T. E. Solomon and B. N. Rock,
“Imaging Spectrometry for Earth Remote Sensing,” Sci-
ence, Vol. 228, No. 4704, 1985, pp. 1147-1153.
[2] A. A. Gowen, C. P. Donnell, P. J. Cullen, G. Downey and
J. M. Frias, “Hyperspectral Imaging: An Emerging Proc-
ess Analytical Tool for Food Quality and Safety Control,”
Trends in Food Science & Technology, Vol. 18, No. 12,
2007, pp. 590-598.
[3] M. Liu, Q. Chen, and H. Lin, “The Study of Non-De-
structive Measurement of Fruit Internal Qualities Using
Spectral Imaging,” Acta Optica Sinica, Vol. 27, No. 11,
2007, pp. 2042-2046.
[4] G. E. Masry, N. Wang, A. E. Sayed, et al., “Early Detec-
tion of Apple Bruises on Different Background Colors
Using Hyperspectral Imaging,” Food Science and Tech-
nology, Vol. 41, No. 2, 2008, pp. 337-345.
[5] D. Wu, H. Yang and X. Chen, “Application of Image Tex-
ture for the Sorting of Tea Categories Using Multi-Spec-
tral Imaging Technique and Support Vector Machine,”
Journal of Food Engineering, Vol. 88, No. 3, 2008, pp.
474-483.
[6] J. L. Yang, T. Zhu and Y. Xu, “Study on Ultraviolet Fluo-
rescence Spectral of Monomers of Distilled Spirits,” Spec-
troscopy and Spectral Analysis, Vol. 29, No. 12, 2009, pp.
3339-3343.
[7] L. Liang, L. Wang, Q. Pang, et al., “Study of Cortex Phel-
lodendri Chinensis Decoction Experiment Based on the
Spectral Imaging Technology,” Spectroscopy and Spec-
tral Analysis, Vol. 32, No. 5, 2012, pp. 1359-1361.
[8] J. Zhao, J. Ma, Q. Pang, et al., “Identification and Analy-
sis of Five Kinds of Chinese Traditional Medicine by
Spectral Imaging Technology,” Lishizhen Medicine and
Material Medical Research, Vol. 23, No. 1, 2012, pp.
188-190.
[9] J. Zhao, Q. Pang, J. Ma, et al., “The Research on Active
Constituent Distribution of Rhizoma coptidis,” Spectros-
copy and Spectral Analysis, Vol. 31, No. 6, 2011, pp.
1692-1696.
[10] Q. Meng, Q. Pang, J. Ma, et al., “Transmission Spectral
Imaging Research on Folium Uncariae from Guangdon,”
Journal of Chinese Medicinal Material, Vol. 33, No. 5,
2010, pp. 696-699.
[11] C. Dai, Q. Pang, J. Ma, et al., “Spectral Imaging Finger-
print of Cortex Phellodendri Chinensispiece,” Laser ＆
Optoelectronics Progress, Vol. 48, 2011, Article ID: 093
001.
[12] J. Zhao, Q. Pang, J. Ma, et al., “Design of the Continuous
Spectrum Imaging Apparatu s Ba sed on LCT Fs,” A cta Pho-
tonica Sinica, Vol. 37, No. 4, 2008, pp. 758-762.