Note: Descriptions are shown in the official language in which they were submitted.
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
Apparatus and Method for Light Spectrum Measurement
Field of Invention
This invention relates generally to light-spectrum-measurements and more
specifically to
a small system for light-spectrum measurement for portability and for in situ
applications.
Background of the invention
Increasingly accurate yet fast methods and instrumentation for measuring
various
quantities are required in environmental analysis and technologies, in
industrial monitoring, in
diagnostics for health care, and in pharmacology. Some of these requirements
are outlined in
Parker S. (Ed.): McGraw-Hill Encyclopedia of Chemistry, McGraw-Hill, 1983, and
in D. A.
Skoog and J. J. Leary, Principles of Instrumental Analysis, 4'h edition,
Harcourt Brace College
Publishers, New York, USA, 1992. For example, in environmental applications,
there is a need
for integrated and miniaturized measurement tools which can be used directly
at sites where
measurements are important (factory exits, waste, dumps etc.), and to transmit
continually,
without cable connection, the information necessary for real-time monitoring
to the centers for
pollution prevention or waste management. Amongst the most widespread methods
for
identification of the pollutants are those of spectrometry and in particular
of absorption
spectrophotometry. Numerous examples of existing needs for light-spectrum-
measurement based
in situ applications are described in the prior art.
Spectroscopy is an analytic technique concerned with the measurement and
characterization of the interaction of radiant energy with matter. This often
involves working
with instruments designed for this purpose, called spectrometers, and
corresponding methods of
interpreting the interaction both at the fundamental level and for practical
analysis. The
distribution of radiant energy, absorbed or emitted by a sample of a substance
under study, is
called its spectrum. If energy of ultraviolet (UV), visible (Vis) or infrared
(IR) light is used, the
corresponding spectrum is called a light-spectrum. In the description, which
follows, the term
spectrum is used in the sense of light-spectrum and the term spectrometer is
used in the sense of
spectrophotometer.
2
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
A spectrometer has a resolution associated with its design or implementation
affecting
resolution of measured spectra. As is well understood by those of skill in the
art of spectrometry,
a required resolution for UV and a required resolution for IR spectral imaging
is different.
Further, the terms high-resolution and low-resolution are related to an imaged
spectral band or to
wavelengths of light within the imaged band. For a broadband spectrometer,
either graduated
spectral resolution or a spectral resolution sufficient to properly image each
band is used.
Interpretation of spectra provides fundamental information at atomic and
molecular
energy levels. For example, the distribution of species within those levels,
the nature of processes
involving change from one level to another, molecular geometries, chemical
bonding, and
interaction of molecules in solution are all studied using spectrum
information. Practically,
comparisons of spectra provide a basis for the determination of qualitative
chemical composition
and chemical structure, and for quantitative chemical analysis as disclosed in
Parker S. (Ed.):
McGraw-Hill Encyclopedia of Chemistry, McGraw-Hill, 1983 which is hereby
incorporated by
reference.
Referring to information from that text, a general functional block diagram of
a
spectrometer is shown in Fig 1 and contains five components:
a stable source of radiant energy;
a transparent container for holding the sample of the substance for analysis;
a device that isolates a restricted region of the spectrum for measurement;
a radiation detector which converts radiant energy to a usable signal in the
form of an
electrical signal; and,
a signal processor and readout, which displays the electrical signal on a
meter scale, a
cathode-ray tube, a digital meter, or a recorder chart.
The modern spectrometers are very sophisticated and guarantee excellent
measurement
performance in a laboratory environment, but in situ applications of
spectrometers are only made
in exceptional circumstances, since they require relatively expensive
equipment, which is usually
transported in special vehicles.
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
In general, the precision of spectrometers is considered adequate for most
laboratory
applications and, therefore, recent efforts in improving spectrometers have
focused on improving
in situ usability.
The miniaturization of spectrometers is a necessary precondition for their
mass in situ
application; however, the size of a spectrometer is limited by required
precision and accuracy of
measurements because of existing relations between optical spectral
resolution, spectral range of
a spectrometer and its physical dimensions. The optical spectral resolution of
commonly
manufactured spectrometers is proportional to their dimensions. This is a
noted and important
limitation for miniaturization of spectrometers, which heretofore could not be
circumvented.
Unfortunately, since precise spectrometers for use in environmental analysis
are often bulky,
costly, and expensive to transport and install, many known and important
applications of
spectrometers remain unimplemented due to cost and/or inconvenience. A
portable spectrometer
that has a lower cost than conventional spectrometers and is preferably hand-
held would allow
the use of spectrometers in a wide range of applications to the benefit of
many industries.
Existing spectrometers, which could be adapted to in situ measurements, are
relatively
large and expensive. Companies such as Ocean Optics, CVI Laser Corporation,
and Control Data
offer miniaturized PC-compatible, on-card spectrometers whose price ranges
between $6,000 and
$20,000. These spectrometers are commonly intended for laboratory applications
and offer
interesting metrological characteristics. Some other companies offer portable
autonomous
spectrometers for measuring specific substance contents (e.g. Clean Earth).
Their dimensions are
relatively large and prices reach several thousands US Dollars. Attempts to
implement the optical
functions using semiconductor-based integration technologies have resulted in
lower quality of
operation over that obtained by means of classic discrete technologies.
Therefore, an
autonomous, integrated spectrum-measurement-based tools for UV-Vis-IR range
are still not
available.
Recently, increased research activity is directed towards developing
spectrometers for
sensing applications and for wavelength division multiplexing (WDM) in optical
communication; however, a simple low cost solution with a totally integrated
opto-electronic part
4
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
using standard technologies is still lacking. A variety of spectrometric
probes for in situ
measurement are known in the art. U.S. Patent No 5,712,710 for example,
describes a probe for
use in measuring the concentration of a specific metal ion dissolved in
liquid. The device suffers
from known problems of probe miniaturization. Either the bandwidth og the
spectrometer is
narrow to accommodate a small probe size, the quality of the spectral imaging
is poor, or the
optical processing components are large and costly.,The device comprises a
hand-held processing
unit coupled to the probe. The processing unit is programmed to calculate and
display the
concentration of a specific material. In this probe, neither the photodetector
nor the processing
unit are integrated with the light diffraction structure. Further, the use of
poor resolution in
imaging the spectrum is unacceptable for most applications when using such a
probe.
U.S. Patent 5,020,910 describesa method of forming a light diffraction
structure directly
over a photodetector. The device requires external electronic circuitry to
obtain a useful spectrum
of light and the spectral resolution is very high in comparison to that of
existing conventional
spectrometers. U.S. Patent 5,731,874 describes a spectrometer with an
integrated photodetector.
This device is sensitive only to particular spectral lines and thus is useful
over a narrow spectral
range.
In U.S. Patent 5,742,389, Zavislan et al. disclose a Spectrophotometer and
Electro-Optic
Module Especially Suitable for Use Therein. The device incorporates a grating
that is moveably
mounted within a small housing that is capable of being held. The disclosed
device concerns
itself with alignment of optical components and the detector, but does not
address resolution.
None of the above-described approaches permits manufacture a low cost high-
resolution
hand-held spectrometer. These known small spectrometric probes are frequently
of complex
design, resulting in increased manufacturing costs. It is, therefore,
desirable to provide an
autonomous simple low-cost solution where the above difficulties are
alleviated. A need remains
for a low-cost miniaturized spectrometric sensor/transducer with a spectral
resolution comparable
to that of conventional spectrometer, and capable of determining the
absorbance spectral
signature of a wide variety of substances in situ.
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
It would be advantageous to provide a small, hand-held, portable spectrometer
having
sufficient resolution and accuracy for use in applications where the
spectrometer is installed as a
sensor in a monitoring system.
Object of the Invention
It is an object of this invention to provide a tool for spectrum measurement
useful for in
situ applications.
It is an object of this invention to provide an integrated spectrometric
sensor/transducer
permitting miniaturization of spectrophotometers while maintaining a
sufficient amount of
resolution.
It is an object of this invention to provide a low cost tool for spectrum
measurement.
Summary of the invention
The resolution limitations imposed by physical size of a spectrometer are well
understood. These limitations are circumvented with the use of sophisticated
technologies for
implementing a method of resolution enhancement for use with a low resolution
grating. These
methods allow for design and manufacture of portable instruments.
In accordance with the invention, a new method for providing an integrated
spectrometric
sensor/transducer (IISS/T) is proposed enabling in situ light-spectrum-based
measurement, at a
significantly reduced cost. The new method is effective. Correspondingly,
IISS/T allows the
manufacture of a plurality of embodiments of miniature spectrometric probes
and hand-held
spectrometers adapted to the different needs. For example, some are provided
with wireless
communication for near continuous transmission of information using wireless,
or other
communication systems. This is useful, in particular, for real-time industrial
and environmental
monitoring.
The proposed new method of the light-spectrum measurement augments measurement
accuracy using digital signal processing instead of the conventional approach
of improved optics
and reduced noise.
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
According to the invention, there is provided a method of measuring a spectrum
of
incident light comprising the steps of capturing a spectrum of the incident
light at a first
resolution; digitizing the analogue measurement to provide an electrical
signal; and, processing
the electrical signal in order to obtain a spectrum having a higher resolution
than the captured
spectrum, said spectrum being an estimate of the measured light-spectrum
and/or its parameters.
Preferably, the optical hardware of a designed spectrometer is minimized.
Further preferably, the
entire method is implemented in a small hand-held device.
The proposed method of extracting information from an optical signal is more
efficient
than sophisticated analog processing and free of troubles characteristic
thereof. It has significant
advantages over optical processing. For example, though spectrometers have
seen few significant
advances in past several decades, digital processors are experiencing
significant performance
gains. With enhanced performance, more complicated and sophisticated methods
may be
implemented. This allows for improved performance during the upcoming years
and/or further
miniaturization. Further, today's semiconductor-based integration technologies
allow for VLSI
implementation of digital processors and optical components. Moreover, an
increase in accuracy
of electrical digital signal processing does not necessarily imply an increase
in technological
difficulties of its implementation, which is typical of optical analog signal
processing.
Advantageously, the IISS/T uses low cost, low-resolution optical components.
By using
low-resolution optical components in the form of gratings, overall size of the
device is
significantly reduced. However, absent significant enhancement of spectral
resolution, spectra
determined using low-resolution optical components are unacceptable for many
applications.
Preferably, the IISS/T comprises processing components that are functionally
fused.
Preferably, a specialized digital signal processor for execution of
specialized methods of
spectrum reconstruction and/or of spectrum parameter estimation ensures a
required quality of
results.
In accordance with an embodiment of the invention, there is provided a
spectrometer
comprising:
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
a transducer comprising a dipersive element for dispersing light and a
photodetector for
converting the dispersed light into an electrical signal representative of
spectral data, the
transducer having a lower spectral resolution than 4nm; and,
a processor for enhancing the resolution of the spectral data to provide
spectral data having a
resolution of at least 2 times that of the transducer.
Preferably, the spectrometer is provided with means of measuring temperature
and of
correcting spectra for temperature fluctuation induced errors.
In accordance with an embodiment of the invention, there is provided a
spectrometer
comprising:
a low resolution transducer comprising a dipersive element for dispersing
light and a
photodetector for converting the dispersed light into an electrical signal
representative of spectral
data; and,
a processor for significantly enhancing the resolution of the spectral data
using stored data, the
stored data relating a captured spectrum of a sample to a known spectrum of
the sample having
higher resolution.
In accordance with an embodiment of the invention, there is provided a
spectrometric sensor
comprising:
a low resolution transducer consisting of a port for receiving electromagnetic
radiation for
measuring a spectrum thereof; a dipersive element for receiving the
electromagnetic radiation
received at the port, for dispersing the received electromagnetic radiation,
and for providing the
dispersed electromagnetic radiation; a photodetector for receiving the
dispersed electromagnetic
radiation from the dispersive element and for converting the dispersed
electromagnetic radiation
into an electrical signal representative of spectral data;
an analog to digital converter for converting the electrical signal
representative of spectral data
into a digital electrical signal representative of spectral data; and,
a processor for significantly enhancing the resolution of the spectral data
and for correcting some
errors within the spectral data using stored data, the stored data relating a
captured spectrum of a
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
sample to a known spectrum of the sample having higher resolution.
According to another aspect of the invention, there is provided a method of
spectral
measurement comprising the steps of:
imaging a first spectrum of a sample using a spectral transducer;
comparing the first spectrum to data representative of a known spectrum for
the same sample;
determining calibration data for transforming the first spectrum into an
approximation of the
known spectrum;
imaging a spectrum of a second sample using the low-resolution spectral
transducer;
estimating an ideal spectrum for the second sample using the calibration data,
the estimation
performed using the determined transformation.
Preferably estimation of the ideal spectrum results in a t least one of a
spectrum with
enhanced resolution and a spectrum corrected for imperfections in the spectral
transducer.
In an embodiment, the first spectrum is defined by {y"°' } , the known
spectrum is defined
byx~°r(7~), and wherein the calibration data is determined by the steps
of:
choosing a form of an ideal peak vs.(~.,1) and of projection operator ~ and
reconstruction
operator ~;
pre-processing the data {y"°' } ;
determining parameters p~ of projection operator hand parameters p~ of
reconstruction
operator ~; and,
storing calibration data comprising the determined parameters in memory.
According to another embodiment of the invention, there is provided a method
of spectral
measurement comprising the steps of:
calibrating of a spectrometer comprising a spectrometric transducer, the
calibration for
determining data relating to the spectrometric transducer;
imaging a spectrum of a sample; and,
9
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
reconstructing a spectrum s(7~;1,a) based on the determined data and related
to the imaged
spectrum, the reconstructed spectrum having a higher-resolution than the
imaged spectrum.
Brief description of the drawings
Exemplary embodiment of the invention will now be described in conjunction
with the drawings
in which:
Fig. 1 presents a general functional block diagram of a spectrometer;
Fig. 2 presents the flow diagram of functioning of the light-spectrum
measuring instrument
a) existing spectrometer;
b) based on the proposed methods using IISS/T.
Fig. 3 is the illustration of the measurement principle underlying the
proposed intelligent
spectrometric transducer/sensor (IISS/T), according to the invention;
Fig. 4 is the illustration of practical gains implied by to the invention;
Fig. 5 presents the foreseen generic structure of the IISS/T;
Fig. 6 is the illustration the effectiveness of the correction of the
imperfections of the
spectrometric transducer, by means of the specialized digital signal
processor,
Fig. 7 illustrates the results of the consecutive stages of signal processing
performed by the
model of the IISS/T, according to the invention.
Fig. 8 presents the example of the hybrid, two-chip structure, of the IISS/T;
Fig. 9 is the illustration of the principle of functioning of an intelligent
spectrometric probe using
the proposed IISS/T;
Fig. 10 is a simplified diagram of a spectrometric apparatus; a computing
means in the form of a
microprocessor, such as a digital signal processor;
Figs. l la through l 1d are simplified flow diagrams of each of 4 steps
according to an exemplary
embodiment of the invention;
Fig. 12 is a simplified diagram of a measuring system according to the prior
art comprising: an
absorption spectrophotometer - model CARY-3 by VARIAN and a personal computer
PC ;
Fig. 13a is a graph of the spectrum of a standard holmium perchlorate sample;
Fig.l3b is a graph of acquired data representative of x(~,), acquired by means
of a
to
CA 02237970 1998-OS-19
Doc No. 60-3 US ~ Patent
spectrophotometer;
Fig.l4a is a graph of the spectrum of a standard holmium oxide sample;
Fig.lSa, Fig.lSb and Fig.l6 show exemplary results of spectrometric data
resolution
enhancement and spectral correction obtained by means of a method according
the invention;
and,
Fig.l7 presents a diagram illustrating applications of the IISS/T.
Detailed description of the invention
The following notation'is used for the description of the invention:
~, - wavelength; ~, E [~~n , Amax, ~
N - number of data acquired by the spectrometric apparatus;
0~, - step of wavelength discretization; 07~ _ ~7~",~ - 7~";~ )~(N -1) ;
~,n - n-th datum acquired by the spectrometric apparatus; ~,n = 7~",;" + (n -
1)~7~ for n =1, ..., N ;
x(~,) - real spectrum of a sample under study;
1- vector of the positions of peaks the spectrum x(7~) is composed of; 1= ~h
l2 .., jK ~T;
I - an estimate of 1;
a - vector of magnitudes of peaks the spectrum x(~,) is composed of; a= ~a, a2
.., aK ~T ;
a - an estimate of a;
s(7~;1, a) - an idealized spectrum of a sample under study, assumed to have
the form:
K
S(a,; I, a) _ ~ ak Vs (~1,, jk
k=1
where vs(~,, l) is an isolated, normalized peak in s(~.; I, a), whose maximum
is located at
~,-j; ~~VS(7~,,j)d7,,-1 for jE~7~,m~n~~m~~~
yn } - spectrometric data representative of x(~,) , acquired by means of the
spectrometric
apparatus; { y" } - { y" n =1, ..., N} ;
x°°r(~,) _ real spectrum of a sample used for calibration of the
spectrometric apparatus;
s(~,;1'ar, a'°r ) _ an idealized spectrum of the sample used for
calibration of the spectrometric
11
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
apparatus;
{yn°' } - spectrometric data, representative of x'°'(7~) used
for calibration of the spectrometric
apparatus; { y °°' } - ~ y cal yl = 1 , N ~°' } ;
n n
~ - an operator (algorithm) of projection mapping the idealized spectrum
s(~,;l,a) into the
space of the data:
{Yn} =~'[S(a'~l~a)~P~~
where p~ is a vector or matrix of the parameters of the operator ~, to be
determined during
calibration of the spectrometric apparatus; p~ _ ~ p~,l p~,z ...]T or:
p~,~,~ p~,l,z ..
p~ = p~,z,~ h~,z,z ..
~ - an operator of reconstruction such as a generalized deconvolution operator
for transforming
the data { yn } into an estimate s(~,) of s(~,;1, a)
s(~)=~~{Yn~~P~r~
where p~ _ ~ p~,~ p~,z ...~T are parameters of the operator ~ including
regularization
parameters, the parameters determined during calibration of the spectrometric
apparatus.
As mentioned above, increases in metrological performance of spectrometric
instrumentation are based on improvements to optical hardware. Limitations on
the size and
quality of optical hardware implementations have resulted in large and
expensive spectrometer
systems, which are generally not well suited to installation and use in a
single in situ test
environment. Described herein is a spectrometer using lower resolution optical
hardware, which
then augments resolution of the system using digital signal processing. The
solution proposed
herein is elegant and significantly advantageous. The method described herein
permits
implementation of an integrated broadband spectrometer. Further, the invention
is significant in
altering the approach to enhancing spectral resolution and thereby obviating
many known
obstacles in the design and implementation of spectrometers for in situ
applications.
12
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
One of the limitations on the size of a spectrometer is the amount of light
incident
thereon. Since a typical spectrometer divides incident light into spectral
bands of a finite
resolution, the light is thereby divided and its intensity is thereby
effected. For example, when
incident light is divided into 100 spectral bands, the resulting bands each
receive at most 1/100'h
of the incident light. These bands may each represent 1 nm of spectral
bandwidth - for a total
bandwidth of 1 OOnm, 0.1 nm - for a total bandwidth of 1 Onm, or 1 Onm for a
total bandwidth of
1004nm. Unfortunately, the larger the band, the less value the spectral
information has when
reviewed since a broad range of wavelengths are included within a single band.
For example, in
medical applications of spectrometry, spectral bands of 0.1 nm or less are
preferred since small
differences in spectra are significant. Conversely, the smaller the band, the
less light reaching a
detector within the spectrometer. To overcome this is a simple matter. For
example, in order to
increase the light within a spectral band 100 fold, one need only increase the
size of the detector
area by 100. In essence, a larger size sensor permits higher resolution
imaging of spectral data.
Referring to Fig. 2a, a simplified flow diagram of an existing spectrometer
system is
shown. Measurement accuracy depends on the performance of the optical analog
signal
processing. Conversion of the physical nature of the signal from the optical
domain to the
electrical domain is performed for display and communication purposes.
The block diagram, shown in Fig. 2b, corresponds to a spectrometer designed
according
to the proposed method of spectrum measurement of the invention. Accuracy of
measurement
significantly depends on performance of digital signal processing. According
to the invention,
optical hardware of a sensor/transducer is minimized, the optical information
is converted into
data in an electronic form using a detector and an analogue to digital
converter, and then the data
is processed using digital signal processing. In order to obtain a final
measurement result, an
estimate of the measured light-spectrum or parameters of the spectrum as
defined by a user is
performed. For example, using a digital signal processor with suitable
programming, the
spectrometer is calibrated. The calibration data relate to characterisitics of
the optical spectral
imaging of the device. For example, errors and imperfections of the transducer
and transforms
for correction thereof and relating to low optical resolution are determined.
During use, once a
spectrum is imaged and digitised, the information on the metrological
imperfections of the
13
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
optical components) is used to correct the digitised spectral data by, for
example, adapting the
parameters of the processing to the signal representing a measured light-
spectrum after
correction of the metrological imperfections of the optical component. The
processor then
determines an estimation of a measured light-spectrum or its parameters as
defined by the user.
Referring to Fig 3, measurement principles underlying the corresponding
sensor/transducer according to the invention are illustrated. In this figure,
«sample» is a sample
of a substance whose spectrum is being measured,
_x(f ) is a result of measurement of the spectrum obtained using a high-
resolution optical
s ometer;
_y(f ) is a result of measurement of the spectrum obtained using a low-
resolution optical
s ometer- thereb ermittin miniaturization - for use in a hand-held device;
and,
p Yp g
s(~,; l, a) is the measurement result from a device according to the present
invention once ~ is
corrected and its resolution enhanced using a digital signal processor. In the
figure, s(~,;1, a) is a
spectral signature, a set of characteristic peak positions and magnitudes of
the sample under
study. Alternatively, the measurement result is a reconstructed spectrum
having a same spectral
signature.
The light transformed by the sample under study is transporting measurement
information
whose extraction, in the proposed sensor/transducer, is herein described in
two steps.
First, a low-resolution spectrometric transducer performs dispersion of light,
by means,
for example, of a simple dispersing element. Photodetectors are used to
convert the dispersed
light into voltage. An A/D converter (analog-to-digital converter) is used to
convert this voltage
into a digital signal. This step essentially provides for creation of the
spectrum by illuminating a
sample, dispersing the resulting spectrum, and capturing the dispersed
spectrum to provide an
electronic data signal for use in the second step.
In the second step, a digital signal processor executes methods of spectrum
reconstruction
and enhancement of resolution on the digital data signal in order to estimate
the actual spectrum
with the desired accuracy and precision. Preferably, the digital signal
processor is a specialised
14
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
processor for use in this step. The term augmentation of spectral data is used
herein to refer to the
operations performed according to the present method for enhancing spectral
resolution and for
correcting errors of the spectral imaging transducer.
The specialized digital signal processor of the IISS/T is provided, during a
calibration
process, with information on the metrological imperfections of the
spectrometric transducer's
optical components. In essence, samples with known spectra are analyzed and
calibration data
relating to the electronic data and how it differs from known spectra for
those samples is
determined. This calibration data may include a selection of appropriate
spectral enhancement
methods that best suit the device or the type of spectrum, errors in spectral
imaging such as
attenuation curves, and other calibration information. The calibration data is
used for spectrum
reconstruction and/or for producing the final measurement result: the estimate
of the spectrum or
its parameters.
The IISS/T according t4 the invention comprises a dispersive element, a
photodetector, an
A/D converter, and a digital signal processor (DSP). The dispersive element
and the
photodetector co-operate to form a spectrum having a resolution lower than a
desired output
resolution. The DSP is used to augment the spectrum to produce an output
spectrum or output
data having sufficient resolution. Because much of the processing is performed
within the DSP,
the cost of the DSP is a significant portion of the overall sensor cost. With
current trends in
semiconductor design and manufacture, it is anticipated that the sensor cost
will be reduced in
the future as DSP processors having sufficient processing power become more
affordable.
Preferably, the IISS/T comprises a DSP and a miniature, low-cost and low-
quality
spectrometric transducer comprising, for example simple dispersive elements,
photodetectors,
and an analog-to-digital converter. This fusion of the functional blocks
enables a designer of
IISS/T to profit from advantages of each of the optical and electrical
portions. In fact,
reprogramming of the IISS/T is possible and software modifications that
improve the overall
performance are anticipated. It is well known that software distribution and
upgrading is
inexpensive relative to the costs associated with similar hardware upgrades.
Further, the use of an
integrated opto-electrical device provides excellent opportunity for automatic
correction of
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
temperature induced errors. A small temperature sensor circuit is disposed at
each of a plurality
of locations within the integrated device. The temperatures are determined and
appropriate
correction of an imaged spectrum is performed depending on the temperature of
the optical
components. Of course, the DSP is not susceptible to errors induced by
temperature fluctuations
so long as it operates within a suitable temperature range. Therefore, a
device according to the
invention is provided with an effective low-cost system of compensating for
temperature
fluctuations.
Fig. 4 illustrates the results of an experiment showing the practical gain in
the quality of
the measurement result obtained using the invention. In this figure
x(7~,) represents data acquired by means of the reference spectrophotometer
ANRITSU (MV02-
Series Optical Spectrum Analyzer) set to the resolution of 0.1 nm (which is
not available in
today's integrated spectrometers);
y~ is a raw measurement acquired by means of a same reference instrument set
to a
resolution of Snm - a typical resolution of integrated spectrometers without
internal
specialized digital signal processors; and,
x~ is an estimate of a spectrum, whose resolution is O.lnm, obtained using
digital signal
processing according to the invention.
As is evident from a review of Fig. 4, a low resolution y~ is enhanced to form
an
excellent approximation of the spectrum measured using a higher resolution
spectrometer.
Comparison of the signals x(7~,), y~ and ~~ gives an idea of practical gains
obtained using
the invention - the gain in resolution shown is of the order of 10. Therefore,
the experiment
clearly demonstrates that using a low-resolution dispersive element and a DSP,
results are typical
of a spectrometer having significantly better resolution. Since size of
spectrometers is at least
partially related to resolution, a device according to the present invention
permits spectrometers
of significantly reduced size for use in similar applications. Of course, the
reduced size and cost
of the device permit many new applications heretofore prohibited by size,
cost, and/or resolution
of prior art spectrometers.
16
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
The proposed method of extracting information from the optical signal is, in
some ways,
more efficient than sophisticated optical analog processing. Further, it is
free of some troubles
characteristic for this type of processing. As described below, it appears
more complicated
conceptually because of the use of sophisticated algorithms for digital signal
processing. The
proposed method permits modifications and selection of different processing
methods without
altering a physical sensor device. Heretofore, improvements to a spectral
sensor required
replacement or hardware modification of the sensor. Technologically, the
present invention is
adaptable and simple because, taking into account today's semiconductor-based
integration
technologies, VLSI implementation of the algorithms is easier than
miniaturized integration of
optical functions. Moreover, the increase in accuracy of electrical digital
signal processing does
not necessarily imply an increase in technological difficulties of its
implementation, as is typical
of optical analog signal processing.
An exemplary structure of the IISS/T is shown in Fig 5. The miniature,
possibly low-cost,
spectrometric transducer, shown in this figure includes a diffractive grating
5 and photodetectors
10 in the form of a CCD. Optionally, using semiconductor-based integration
technologies such as
CMOS technology, such a device is manufactured as an integrated device.
Light from a sample and representing a spectrum requiring analysis is received
at port D.
The light is characterized by the x(~,). The light is provided to dispersive
element 5 through
which it is dispersed to photodetectors 10 which provide an electronic signal
corresponding to a
captured spectrum, =~ . Since the dispersive element 5 is of small size (shown
within a single
integrated circuit), resulting resolution of the captured spectrum is low. The
electronic signal is
provided to a processor 20, in the form of a specialized DSP, where it is
digitized and augmented
to form an output spectrum or output spectral parameters.
Depending on a targeted wavelength range of the IISS/T, the optical portion of
the
IISS/T, shown in Fig. 5, is implemented in a same silicon integrated circuit
(IC) as the digital
processor or, alternatively, in a separate IC. Alternatively, it is mounted on
an IC as an external
element manufactured separately for technological reasons. Further
alternatively, it is mounted
separate from the IC and optically aligned therewith.
17
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
As mentioned above, the miniaturization of spectrometric instruments is
limited by the
required accuracy of measurement and limitations of integrated devices.
Implementation of
optical functions using semiconductor-based integration technologies does not
provide similar
performance to classical discrete optical instrumentation. This is an
important motivation for the
invention, which allows for miniaturization of spectrum-measurement-based
instrumentation.
As a further example, let us assume that a spectrometric transducer used in an
IISS/T is
characterized by the following parameters:
- range of wavelength from 450 to 650 nm (Vis),
- total surface of the spectrometric transducer: 1 cm2°
- Litrow configuration of a diffractive grating
- photodetector composed of semiconductor diodes whose diameter is 25pm;
then using the developed model of the spectrometric transducer of the IISS/T,
we obtain the
following:
- a diffractive grating with 1200 steps/mm,
- a number of diodes : 160 , with total width of the detector of 4mm
- optical resolution of obtained optical transducer 0~,=11 rlm.
Unfortunately, these results follow from the above assumptions so a higher
resolution detector
requires either different assumptions or processing of the obtained - imaged -
spectra according
to the invention.
Fig.6 illustrates the results of an experiment showing the effectiveness of
the IISS/T
designed according to the invention. In this figure:
x(~.) represents the data acquired by means of a reference spectrometer CARP-3
(Varian) set to a
resolution of 0.2nm - a resolution not commonly available in prior art
integrated spectrometers;
y~ is a measurement result at the output of the model of the spectrometric
transducer of the
IISS/T, with l5nm resolution; and,
x~ is an estimate of the spectrum x(7~), whose resolution obtained after
digital signal
processing is approximately O.lnm. This is the resolution obtainable at the
output of the IISS/T,
18
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
proposed according to the invention, satisfying the users' requirements for
many practical
applications. An enhancement of about 10 times the resolution is achieved. Of
course, as spectral
enhancement increases to for example 40 or 60 times the resolution of the
transducer, it is
expected that errors in estimation will also increase. This should be
evaluated on an application
by application basis to determine applicability and degree of miniaturization
of the invention for
a particular application.
Referring to Fig. 7, signal processing methods performed in the IISS/T and
according to
the invention are illustrated. In,this figure:
x(~,) represents data obtained by a reference spectrometer CARP-3 (Varian) set
to the resolution
of 0.2nm which is not commonly available in today's integrated spectrometers;
_y(~) is a measurement result obtained by means of a same instrument set to a
resolution of 4nm
si a resolution of the order of 1-l Onm is considered obtainable from
integrated spectrometers
having no spectral augmentation depending on technology used for implementing
optical signal
processing functions; and,
s(~,; l, a) is the spectral signature of x(~,), obtained after digital signal
processing and, therefore,
obtainable at an output of the IISS/T proposed according to the invention.
This represents an
increase in resolution of 40 times. As is seen in the diagram, the error in
the estimation is quite
pronounced. For some applications, this will be sufficiently accurate, for
other applications, a
larger transducer having a higher resolution is used.
In this example, the processing performed by the DSP includes computing of the
parameters of the analyzed light-spectrum - the positions and magnitudes of
peaks the spectrum
is composed of. Some of those parameters correspond to specific light emission
or specific
pollutants contained in the analyzed sample.
In Fig. 8, an example of a hybrid version of the IISS/T comprising two fused
chips is
shown. This embodiment employing separate integration of the spectrometric
transducer and of
the specialized dedicated digital electrical signal processor, is a natural
step towards total
integration. It is also useful forprototyping, flexibility in selecting a DSP,
when combinations of
the IISS/T with various input sensors for different spectrum-measurement-based
applications is
19
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
desired and so forth. Functionally, the device of Fig. 8 operates in
accordance with the
description relating to Fig. 5.
For monolithic implementation of a fully integrated IISS/T, shown in Fig. 5,
constraints
include those imposed by the fact that microelectronics (VLSI) and integrated
optics
technologies are not yet completely technologically compatible; however, the
monolithic
implementation of some application-specific measuring systems was successful
as described by
R. E. Kunz in "Totally Integrated Optical Measuring Sensors", SPIE
Proceedings, Vo1.1587, 1991, pp.
98-113, by L Templeton, I M; Fallahi, M; Erickson, L E; Chatenoud, F; Koteles,
E S; Champion, H G;
He, J J; Barber, Focused ion beam lithography of multiperiod gratings for a
wavelength-division-
multiplexed transmitter laser array, RPU - Journal of Vacuum Science and
Technology - Section B -
Microelectronics Nanometer Structures, 1995, v.13, n.6, p.2722, 3p. The
research progress in the
domain of multilayer silicon-based materials is very rapid and promises new
possibilities for
monolithic implementation of the IISS/T. The attainable parameters of the
IISS/T justify the
attempts to develop mass production of low-cost miniature spectrometric
sensors/transducers.
Fig. 9 shows an exemplary application of an IISS/T according to the invention
wherein
the IISS/T is for use in a remotely controlled spectrometric probe for real-
time environmental
and/or industrial monitoring. The probe is mainly composed of the IISS/T, a
light source selected
for a specific application, and a telecommunication means for real-time
communication with a
monitoring network. In some applications, other sensors and transducers
provide input light to
the IISS/T. In this case the IISS/T is used as a transducer. Where measurement
of a spectrum is
desired, the IISS/T acts as a sensory input of the probe.
Once the data are captured, interpretation of the data is not a
straightforward task. Before
the data are interpreted, the spectral data requires augmentation. For
example, when captured at a
resolution of lOnm, spectral data is not useful for most applications. In
order to produce a hand-
held broadband spectrometric sensor at a reasonable cost using current
technology, a grating
having a low resolution, such as 4nm-lOlnn, is employed . Therefore, it is
essential that the
captured spectrum is augmented prior to analysis. As proposed herein, the
method of
augmentation involves estimation of spectral values from the low-resolution
spectrum based on
existing calibration data of the sensor. The augmentation process is set out
below.
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
Referring to Fig. 10, a system is shown comprising the following: a
spectrometric apparatus,
in the form of a a spectrometric transducer for converting an analogue
electromagnetic signal,
such as light containing information of a measured spectrum, into a digital
electrical signal
representing the spectrum; a computing means in the form of a microprocessor,
a general-
purpose digital signal processor, or an application-specific digital signal
processor; and, other
functional elements necessary for measuring a spectrum of a sample of an
analyzed substance
(hereinafter referred to as sample).
The method of augmenting spectra set out below is useful in the IISS/T as a
method
implemented within the processor. It is described herein as an embodiment of a
method of
implementing spectral augmentation. Of course, the IISS/T may be provided with
another
suitable method as are known or may become known in the art. The method of
augmenting
spectra set out below is also for general application to other spectrometric
devices.
The main objective of the method of enhancing resolution and correction of
spectral data
augmenting spectra - is estimation of the positions 1 and magnitudes a of the
peaks contained in
the spectrum of a sample under study x(7~) on the basis of the acquired
spectrometric data {y~ ~ .
The feasibility of this operation is critically conditioned by an auxiliary
operation on the
reference data {yn°'} and corresponding reference spectrum
x'°'(~,), referred to as calibration of
the spectrometric apparatus. This operation is aimed at the acquisition of
information on a
mathematical model of a relationship between spectrometric data and an
idealized spectrum,
which underlies the method according to the present embodiment for estimation
of the
parameters 1 and a. Although calibration does not necessarily directly precede
augmentation of a
sequence of spectrometric data ~yn } , valid calibration results should be
available during this
process.
A significant difficulty, related to estimation of positions 1 and magnitudes
a of
spectrometric peaks, relates to blurring of those peaks caused by physical
phenomena in a sample
and by blurring of their representations in the data ~yn } caused by
imperfections in spectrometric
apparatus. This difficulty is overcome according to the present method through
application of a
21
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
process for reconstruction of an idealized spectrum s(~,;l,a) in order to
correct the spectrometric
data by, for example, reducing blurring caused by both sources; if s(~,;1, a)
is assumed to be an
approximation of x(~,), then only the instrumental blurring is corrected.
In accordance with the above general functional requirements and referring to
Figs. 11 a through
l 1d, the method comprises the following steps:
~ calibration of a spectrometer (the sub-procedure ISD cal),
~ reconstruction of a spectrum s(~,; l, a) (the sub-procedure ISD rec),
~ estimation of parameters 1 and a on the basis of an estimate s(~,) of
s(~,;I,a) (the sub-
procedure ISD est).
sub-procedure ISD cal
The sub-procedure ISD cal comprises the following steps:
a) choosing a form of ideal peak vs(7~,1) and of operators hand ~~;
b) choosing a calibration sample whose spectrum x'°'(~,) is known;
c) setting measurement parameters of the spectrometric apparatus;
d) acquiring data ~y"°' } representative of the calibration sample
whose spectrum x~°'(~,) is
known;
e) pre-processing of the data {y"°' } to eliminate outliers, to perform
baseline correction,
smoothing, acquiring a priori information in the form of a pre-estimate of the
variance of
errors in the calibration data, and normalization;
f) determining parameters p~ of the projection operator rte, and parameters
p,~ of the
reconstruction operator ~. A process for performing these estimations is
preferably tuned for
use with a specific apparatus. For example, when known variance exists in a
type of
dispersive element, this a priori knowledge is beneficial in determining the
process for
performing estimations and thereby determining a process for calibration. Of
course, this is
not necessary since some processes for estimation and calibration are
substantially universal
for spectrometric apparatuses.
22
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
sub-procedure ISD rec
The sub-procedure ISD rec comprises the following steps:
a) setting measurement parameters substantially the same as those above;
b) acquiring data { yn ~ representative of a sample under study;
c) pre-processing of the data {y" } in a similar fashion to the preprocessing
for determining the
calibration data;
d) estimating an idealized spectrum s(~,;1, a) on the basis of the data { y" ~
, by means of the
predetermined operator ~ and the parameters p~ ;
sub-procedure ISD est
The sub-procedure ISD est comprises the following steps:
a) estimating positions 1 of peaks within a spectrum on the basis of the
estimate s(~,) of s(~,;1, a)
by means of a maximum-detection algorithm;
b) estimating magnitudes a of the peaks, by means of a curve fitting algorithm
using one of the
following methods:
- the data { y" } , v S (7~, l ) , the operator ~~' with parameters p~ , and
the estimate 1;
- the estimate s(~,), vS(7~,1) and the estimate 1.
c) iteratively correcting the estimates of the parameters of peaks obtained in
(a) and (b);
d) adapting the results of parameter estimation in accordance with user
requirements, such as
transformation of parameters into some pre-defined parameters of an analyzed
substance.
A particular implementation of an exemplary embodiment has been designed for a
measuring system as shown in Fig. 12 comprising: an absorption
spectrophotometer - model
CARY-3 by VARIAN and a personal computer PC.
The following measurement parameters have been selected both for calibration
and for
acquisition of test data:
a wavelength range: 7u",;" =199.9 nm, 7~",~ = 800 nm ;
23
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
number of data acquired by the spectrophotometer: N = 6002 ;
step of wavelength discretization: 0~, _ ~~,~"~ - a.min )~(N -1) = 0.1 nm .
The test data were acquired for a standard holmium perchlorate sample; its
real
spectrum x(7~) is shown in Fig.l3a. The known parameters of this spectrum are
as follows:
the vector of the positions of peaks:
1= 382.7 386 390.1 417 422 451.2 468.1 473 479.5 485.1 491~T
the vector of the magnitudes of peaks:
a= .0483 .1492 .0938 .766 .2481 1.2513 .2292 .2595 .1475 .5419 .1073~T
The idealized spectrum of a sample under study is assumed to have the form:
S(~~ 1, a) _ ~ ak V.s (a,, 1k )
k=1
with the peaks defined by:
vs(~,,1)=g(~,-l) for l E[a,min,a,max]
The set of data representative of x(7~), acquired by means of the
spectrophotometer,
~Y"~ _ {Yn ~ n =1,...,6002} , is shown in Fig.l3b.
The calibration data were acquired for a standard holmium oxide sample; its
spectrum
x~°'(~,) is shown in Fig.l4a. The parameters of this spectrum are as
follows:
the vector of the positions of peaks:
1'°' =415.2 419.2 425.5 445.5 454.2 460.7 473.7 484 488.4~T;
the vector of the magnitudes of peaks:
a'°' =.0799 .1813 .0868 2.313 .7862 .9772 .0815 .076 .0709~T .
The idealized spectrum of a sample used for calibration
s(~,;1°°',a'°') is assumed to have
the form:
24
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
9
S a,~lcal acal acalv ~ lcal
> > )_~ k s( ~ k )
k=1
The set of data representative of x'°~(~,) , acquired by means of the
spectrophotometer,
'a~ n =1, ...,6002} , is shown in Fig.14b.
Yn Yn
The chosen operator of projection, for mapping an idealized spectrum s(7~;1,
a) into the data
space
{Yn } _ ~~s(~~ 1~ a)~ P~ J
is defined by the following operations:
x(~) = eXP[ ,~gsX (~ - ~') ln~s(~,';1, a)~ ~ ~~
Y(f) ,~~gxy(~'' a,~) x(a,~) Cl~t'r
Yn = y(~t,n) for n =1,...,N
The function g~,(7~) is estimated to have the form of the Gauss function:
1
gxy (~) - 2~ a xy eXp 2a y
Consequently, the vector of the parameters p~ of the operator ~' contains
discrete values of
gSx(~) and parameter 6~,
The chosen operator of reconstruction, for transforming the data ~ yn } into
an estimate
s(7~) of s(~,;I,a),
s(~)= '~[{Yn};P?~,
is specified by the following steps:
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
a discrete estimate {.zn} of x(~,) is found by means of a rational filter
applied to the data ~yn~ ;
and,
an estimate s(~,) of s(~,;l,a) is computed using a spline-based Kalman filter
applied to ~xn~ .
The vector p~ _ ~p~,l p~,2 ...~T of parameters of the operator ~ contains
coefficients of
the rational filter as well as discrete values of the function g.sx (7~) and
regularization parameters
for the spline-based Kalman filter.
The following operations are performed during calibration:
identification of a function gSx (~,) , using an iterative algorithm such as
the Jansson's algorithm
described in P.A. Jansson, Ed., Deconvolution of spectra and images, Academic
Press. Inc.
(1997);
estimation of parameter a xy of function gxy (7~) based on the ideal spectrum
x '"' (~, ) using an
optimization algorithm;
estimation of coefficients of the rational filter using an optimization
algorithm; and,
estimation of a regularization parameter of the spline-based Kalman filter,
using an optimization
algorithm.
The exemplary results of spectrophotometric data resolution augmentation
obtained by means of
this exemplary method are shown in Fig.lSa, Fig.lSb and Fig.l6. The estimates
of test spectrum
parameters obtained by means of the present method, are as follows:
the vector of the positions of peaks:
I = 386 390.8 395.1 410.8 417.2 421.4 451.1 468 473.1 479 485 492.2~~
the vector of the magnitudes of peaks:
a = .1276 .0738 .0376 .0243 .6932 .235 1.3142 .1729 .2593 .1239 .4937 .079~~
As is evident to those of skill in the art, application of a method as herein
described allows
for capturing of spectral information using low-resolution optical components.
This allows for
miniaturisation of optical components used in spectral sensing applications
because, through
resolution augmentation, useful information is extracted from the captured
spectra.
26
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
The proposed method of spectral resolution augmentation is applicable in
virtually all
branches of spectroscopy. The motivation for its application in a given
measurement situation is
founded on expected gains. Examples of some expected gains include the
following. Increased
accuracy of spectrometric analyses is accomplished by a given spectrometric
system. The
increased accuracy results from correction of instrumental errors and reduced
uncertainty of
estimation of parameters of measured spectra. A reduced cost of spectrometric
analyses with a
given accuracy is achieved by replacing a high-resolution spectrometric
transducer with a
functionally equivalent but low-resolution instrument. Increased reliability
and informativeness
of spectrometric analyses results from parallel utilization of a network of
low-cost spectrometers
served by a common computing resource. This replaces the conventional
autonomous
spectrometer having a dedicated processing resource. Dimensions of
spectrometers and
spectrometer-related measurement tools are reduced because software replaces
some functions
currently implemented using optical processing and because compensation of
hardware
imperfections caused by miniaturisation of optical components is achieved.
Many variations of operators and mathematical models or algorithms are useful
in a method
according to the invention. Though the above description is with respect to a
single set of
equations for augmenting resolution of a spectrum, other equations are also
applicable. Some
examples of other approaches for augmenting spectra according to the present
invention are
described below.
Optionally, the following mathematical models of the spectrometric data may be
used for
defining the operator ~ where the corresponding operators ~ ~.e set out below
a) the stationary linear model:
b) the non-stationary linear model:
y(~) _ ~'g(~, ~') s(~';1, a) ~,
and,
27
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
c) the non-linear model, e.g.:
Y(a'/ _ ,~~g(a'~ a,') Fs[S(a,';1, i1)] d~t''
Y(~)=Fy[,~~g(~~~~) Fs~s(~el~a>~ ~'~~
where g(7~) and g(~,,7~') are the apparatus functions of the spectrometric
apparatus; FS and Fy are
non-linear functions.
The corresponding operators ~ have the following forms:
a) the operator corresponding to the stationary linear model:
Yn - ~ pc~n,v l a S(a' ~ ~ 1~ a) Ci~''
''hv
where p~~ = gC ~+~2 ~ ~ 4~, ; '
b) the operator corresponding to the non-stationary linear model:
.Yn = ~ pc~nw ~+' S'a' ~ ~ 1~ a) da,'
v
where p$n,~ = g~~,n ~ ~+i2 ~ ~ ~7~ ; and,
c) the operator corresponding to the exemplary non-linear models:
Yn - ~ p~n,v ,~~' Fs [S(a,', l, a)] Cl~''
v
Yn = Fy ~ p$n,~ ~~~ FS ~s(7~', ~, a)~ ~
v
28
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
where p~n,~ - g~~,n, ~'"+' + ~" ~ a~, .
2
Optionally, the following methods of signal reconstruction in the form of
deconvolution or
generalized deconvolution are used for defining the operator ~
a) the original domain, numerical differentiation-based method as described by
Morawski & Sokolowski in 1995;
b) the iterative methods of Jansson and Gold;
c) the spectrum-domain, Tikhonov-regularization-based method;
d) the cepstrum-domain, Tikhonov-regularization-based method;
e) the original-domain, Tikhonov-regularization-based method with the
positivity constraint
imposed on the solution;
f) the Kalman-filter-based method with with the positivity constraint imposed
on the solution;
g) the Kalman-filter-based method with spline-approximation of the solution;
h) the adjoint-operator method as described by Morawski & Pawinski in 1995;
i) the entropy-based variational method;
j) the Volterra-series-based methods;
k) the rational-filter-based method as described by Szczeci ski et al. in
1997.
Moreover, many other methods developed in the domain of chemometrics such as
those of
Brown et al.; telecommunications, seismology and image processing are
applicable with the
method according to the invention for spectral resolution augmentation.
Selection of
mathematical algorithms for use in the present invention is straightforward
for one of skill in the
art without undue experimentation.
The following methods may be used for determining the regularization
parameters of the
operator ~
a) the disrepancy principle with a pre-estimate of the variance of measurement
errors in the
data as described by Tikhonov et al. In 1995;
b) the method of the L-curve as described by Hansen & O'Leary in 1991;
29
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
c) the method of additional set of calibration data as described by Szczeci
ski et al. in 1995.
Calibration is also described above with relation to an exemplary embodiment
thereof.
Optionally, the isolated peak vs(~,,1) is assumed to have the following forms:
a) the Dirac distribution 8 (~,) for all values of l ;
b) a triangle whose width is constant or varying versus l ;
c) a rectangle whose width is constant or varying versus 1;
d) a Gauss function whose width is constant or varying versus l ; and,
e) a Lorenz function whose width is constant or varying versus l .
Optionally, at least one of the following methods is used for estimation of
the apparatus function
g(7~)
a) smoothing approximation applied directly to the data ~y"°' } if the
isolated peak vs(~,,1) is
assumed to have the form of the Dirac distribution ~(~,) ;
b) deconvolution of the data {y"°'} with respect to
s(~,;1~°',a'°'); and
c) subsequent use of deconvolution and smoothing approximation.
Optionally, at least one of the following methods may be used for determining
other parameters
of the operator ~
a) a direct transformation of the parameters of the operator ~ ;
b) the minimization of any norm of the solution p.~ under constraints imposed
on another
norm of the disrepancy ~ S(a,; l~ul , a~°' ) - ~~{ ynal ~; p~
c) the minimization of any norm of the disrepancy s~7~;1 ~°' ,
a'°' ) -.'due[{ yn°' }; p~ ] under
constraints imposed on another norm of the solution p ~ I .
Optionally, at least one of the following methods is used for estimation of
magnitudes a of peaks,
given the estimates 1 of their positions 1:
a = arge inf j { y" } - ~[s~~,;1, a); per, ~ a E A ~ ; and
v
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
a = arge inf. s(~,~ - s~~,;1, a~ ~ a E A
v
with A - being a set of feasible,solutions; options: q=2 and A c R k ; q=oo
and A c R k ; q=2 and
A c R + ; q=~ and A c R + . Some examples of algorithmic solutions are given
in
Deming S. N., Morgan S. L.: Experimental Design : A Chemometric Approach,
Elsevier 1987;
Fraser R. D. B., Suzuki E.: "Biological Applications". In: Spectral Analysis -
Methods and
Techniques (ed by J. A. Balckburn), M. Dekker, 1970, pp. 171-211; Fister III
J. C., Harris J. M.:
"Multidimensional Least Squares Resolution of Excited State Raman Spectra",
Anal.
Chem.,Vol. 67, No. 4, 1995b, pp.701-709 ; Fister III J. C., Harris J. M.:
"Multidimensional Least
Squares Resolution of Raman Spectra from Intermediates in Photochemical
Reactions", Anal.
Chem., Vol. 67, No. 8, 1995a, pp.1361-1370; Goodman K. J., Brenna T.: "Curve
Fitting for
Restoration of Accuracy of Overlapping Peaks in Gas Chromatography /
Combustion Ratio Mass
Spectrometry", Anal. Chem., Vo1.66, No. 8, 1994, pp. 1294-1301; Miekina et al.
"Incorporation
of the Positivity Constraint into a Tikhonov-method-based Algorithm of
Measurand
Reconstruction". Proc. IMEKO-TCI &TC7 Colloquium (London, UK, Sept. 8-10,
1993), pp. 299-
304 and so forth. A particularly effective solution of the above optimization
problem is based on
a non-stationary Kalman filter or an adaptive LMS algorithm as described in
Ben Slima M.,
Szczecinski L., Massicotte D., Morawski R. Z., Barwicz A.: "Algorithmic
Specification of a
Specialized Processor for Spectrometric Applictions", Proc. IEEE Instrum. &
Meas. Technology
Conf. (Ottawa, Canada, May 19-21, 1997), pp. 90-95 and in Ben Slima M.,
Morawski R. Z.,
Barwicz A.: "Kalman-filter-based Algorithms of Spectrophotometric Data
Correction - Part II:
Use of Splines for Approximation of Spectra", IEEE Trans. Instrum. & Meas.,
Vol. 46, No. 3,
June 1997, pp. 685-689.
Optionally, methods for estimation of the magnitudes a are used for iterative
correction of
estimates of magnitudes a and positions 1 of the peaks. Known methods include
the following
1=arg, inf. {y"~-~[s(~,;l,a~;p~] ~ 1 EL~
v
and,
31
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
1= arg~ inf ~Ils(~,) - s(~,;1, a)II ~ 1 E L
9
with L being a set of feasible solutions; options: q=2 and L c R k ; q=oo and
L c R k ; q=2 and L
c R + ; q=oo and L c R k
According to the method set out above, the data are pre-processed. The pre-
processing is
performed according to known techniques and for known purposes with relation
to the methods
selected for augmenting resolution of the spectral data and the sensor with
which the pre-
processing is used. Optionally, one of the following methods is used for
normalization of the
data:
a) the linear or nonlinear transformation of the ~, -axis, aimed at
diminishing the non-
stationarity effects in the data;
b) the linear or nonlinear transformation of the y-axis, aimed at diminishing
the non-linearity
effects in the data;
c) the linear or nonlinear transformation of the 7~-axis and y-axis, aimed at
diminishing the
non-stationarity and non-linearity effects in the data.
Optionally, one of the following methods may be used for smoothing the data:
a) the linear, FIR-type or IIR-type, filtering;
b) the median filtering;
c) the smoothing approximation by cubic splines;
d) the deconvolution with respect to an identity operator.
Baseline correction is performed according to standard known techniques such
as those
described in Brame E. G., Grasselli J., Infrared and Raman Spectroscopy,
Marcel Dekker 1976.
Though the method of augmenting resolution and accuracy of a spectrum from a
low
resolution captured spectrum according to the invention is described with
reference to any
hardware implementation of this method, it is preferred that the method is
implemented in an
integrated hardware device as described herein.
32
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
Referring to Fig. 17, a summary of potential applications of the IISS/T in
various fields of
application is presented. The IISS/T (in the center of the figure) is applied
using different
spectrometric techniques, which are used in analytical laboratories. The use
of an IISS/T
according to the invention facilitates application of spectrum-measurement-
based methods in
real-time environmental, agricultural, medical, and industrial monitoring. It
also facilitates use of
a hand-held spectrometer designed for specific applications or for a variety
of applications.
The proposed invention permits implementation of sensors that are advantageous
in many
ways including the following. The proposed IISS/T is autonomous in the sense,
that it is capable
of producing output measurement results without external operations and/or
computing. The
IISS/T architecture supports manufacturing of various low-price intelligent
spectrometric probes
and hand-held spectrometric instruments without some of the technological
problems inherent in
high-resolution optical processing spectrometers. The proposed IISS/T is
easily adapted to
diverse applications by reprogramming the specialized digital signal
processor. The proposed
method for spectrum measurement is particularly advantageous for integrated
miniature
implementation of the IISS/T. The IISS/T is robust to mechanical,
electromagnetic, chemical and
biological influences, due to its compact packaging and integrated design.
Further, it is less
cumbersome for transport, installation, testing, and repair.
In a pre-defined specialized application, the metrological parameters -
variety and ranges
of measured quantities, as well as accuracy of measurement - of the IISS/T are
comparable to
those of a general-purpose laboratory spectrophotometers; yet, the IISS/T has
a significantly
lower manufacturing cost. Using current technology, an IISS/T is
manufacturable as small as 12
cm3. For this reason the IISS/T is naturally adapted for in situ measurements.
A network of
deployed IISS/T may replace the vehicle-based system of sampling, currently
used in the
environmental monitoring. Alternatively, a network of low-resolution sensors
coupled with a
single processor is useful for random sampling, sequential sampling, or, when
the processor is
significantly more powerful than necessary for augmenting resolution and
accuracy of a single
spectrum, for simultaneous sampling. The main advantage of this solution would
be an increase
in the reliability and informativeness of environmental monitoring due to the
continual sampling
in situ. Such a network of IISS/T is useful in chemical, pharmaceutical and
biotechnological
33
CA 02237970 1998-OS-19
Doc No. 60-3 US Patent
industries for continual monitoring of manufacturing processes. The main
advantage of this
solution within those industries is an increase in the reliability and safety
of manufacturing
processes, as well as an improvement of the quality of production.
Without the digital processor performing spectral augmentation, no useful
measurement
results are obtained. This is distinct from existing spectral transducers
having optical processing,
the results of which are provided to an external processor for spectral
analysis such as noise
filtering and so forth.
The price of an IISS/T manufactured according to an embodiment of the
invention, using
standard integration technologies, is comparable with the price of a
semiconductor device than
that of classic spectrometer. The availability of such an IISS/T will change
the approach to the
use of light-spectrum-measurement-based techniques, currently limited to the
laboratory
environment for practical purposes. This invention provides a method of
implementing a
spectrometer for use in situ in many metrological applications.
Clearly, the use of the exemplary method described herein is not limited to
the IISS/T.
The method of spectral correction and resolution augmentation described above
is useful in many
applications other than a hand-help spectrometer. For example, in high
precision measurement of
spectra or in the design of lower cost high precision spectrometers.
Similarly, the exemplary
method of spectral enhancement performed in the processor of the IISS/T as
described above, is
an exemplary method of enhancing spectral accuracy and resolution. It is
exemplary in nature
and not intended to limit the scope of the inventive apparatus.
The exemplary embodiment of the invention presented above is not intended to
limit the
applicability of the method to the presented example. Neither is it intended
to limit the variety of
algorithms that may be used to embody the operations of the specialized
digital signal processor.
Numerous other embodiments may be envisaged without departing from the spirit
or scope of the
invention.
34
