Method and system for detecting and authenticating taggants in markings via surface-enhanced Raman spectroscopy - Patent Application 20070233334

Abstract

The present invention relates to a method and corresponding system that can check the presence or absence of authentic SERS or SERRS taggants having unique, characteristic surface-enhanced scattering features in a machine-readable marking applied to a document of value by using a Raman spectrometer configured to perform Raman spectroscopic analysis of the marking. The method according to the invention allows reliable and fast detection of the presence of SERS / SERRS taggants and is particularly suitable for checking the authenticity of documents of value, such as banknotes, that are moving relative to the Raman spectrometer at a given speed, possibly even at high speed, or that are only slightly exposed to the Raman spectrometer.

Description

Technical Field

[0001]

[0001] The present invention relates to the technical field of detecting taggants present in markings on substrates (e.g., banknotes) by surface-enhanced Raman spectroscopy (SERS) or surface-enhanced resonance Raman spectroscopy (SERRS). The taggants are of the SERS type or the SERRS type and thus have unique characteristic surface-enhanced features (i.e., surface-enhanced Raman scattering features or surface-enhanced resonance Raman scattering features, respectively) that enable the use of a standard Raman spectrometer for the detection of the taggants. Background of the Invention

[0002]

[0002] As is well known to those skilled in the art, SERS or SERRS taggants comprise an assembly of nanoparticles presenting a plasmonic surface and Raman-active reporter molecules adsorbed on the surface of the nanoparticles. The nanoparticles presenting a plasmonic surface play a role in generating the electric field necessary for Raman amplification, while the Raman-active reporter molecules provide the unique vibrational fingerprint of the SERS taggant. The SERS or SERRS taggant may further comprise an outer coating layer that separates the nanoparticle assembly with the Raman-active molecules adsorbed on the surface from the external medium. Thus, the outer coating layer serves to a) separate the SERS / SERRS taggant from the external medium, thereby preventing the Raman-active reporter molecules from leaching out of the SERS / SERRS taggant and protecting the SERS / SERRS taggant from contamination by the external medium that may result in false peaks, b) increase the colloidal stability of the SERS / SERRS taggant, and c) provide a surface favorable for further chemical functionalization. The outer coating layer includes silica and polymers such as poly(ethyleneimine) (PEI), sodium poly(styrene-alt-maleic acid) salt (PSMA), poly(diallyldimethylammonium chloride) (PDADMAC).

[0003]

[0003] Raman spectroscopy is widely used for quantitative pharmaceutical analysis. However, when using Raman spectroscopy, generally, as shown in FIG. 1, due to the fact that the duration of the Raman signal is much shorter than that of the fluorescence signal, sample fluorescence generally obscures the scattered Raman signal, which is an obstacle. In FIG. 1, the Raman intensity signal (10) (relative intensity value) caused by irradiation with a 600 ps laser pulse (the 1 ns gate is indicated by the vertical dotted line), and various emission (fluorescence) intensity signals (11, 12, 13, and 14) are shown (the lifetimes are 1 ns, 5 ns, 10 ns, and 50 ns respectively). Time gating, which provides an instrument-based method for removing most of the fluorescence signal through the time resolution of the spectral signal, is known, enabling the Raman spectrum of the fluorescent substance to be obtained. An additional practical advantage is that spectral signal analysis is possible even under ambient illumination. Conventional partial least squares (PLS) regression enables spectral signal quantification by Raman-active time domain selection (based on visual inspection) to improve performance. The model performance is further improved by using kernel-based regularized least squares (RLS) regression that uses greedy feature selection (i.e., "forward selection" by selecting the best feature one by one, or "backward selection" by removing the worst feature one by one), where the data usage in both the Raman shift and the time dimension is statistically optimized. The overall time-gated Raman spectroscopy using data analysis optimized in both the spectral dimension and the time dimension in particular shows the possibility of highly sensitive and relatively routine analysis of luminescent substances (e.g., drugs during drug development and manufacturing).

[0004]

[0004] A Raman spectrum is obtained by measuring the intensity distribution of Raman scattered photons received from a substrate containing a substance of interest and irradiated with a monochromatic light source as a function of wavelength. Quantitative determination is based on the fact that the concentration of the substance of interest is proportional to the integrated intensity of the characteristic Raman band of the substance. However, overlapping peaks of different compounds in a mixture present on the substrate and experimental effects not related to the sample concentration generally complicate signal analysis. In such cases, multivariate analysis, which may involve a large number of spectral data, is more reliable than methods that consider only one or a few spectral features. Several multivariate methods for interpreting Raman spectra have been established. The objectives of the methods are (i) to extract spectral information for quantifying the substance of interest, (ii) to estimate the uncertainty of the quantification, and (iii) to evaluate the performance of the constructed model.

[0005]

[0005] Partial least squares (PLS) regression is one of the most widely used chemometric methods for quantitative analysis of spectra. PLS links information within two data matrices X (e.g., spectral variations) and Y (e.g., sample composition) in a multivariate model by maximizing the covariance of the matrices. Kernel-based regularized least squares (kernel-based RLS) regression, when combined with a feature selection algorithm such as greedy feature selection, is another approach that has the ability to learn a function from non-linear data features that optimizes the use of features provided by the data features. PLS and RLS are very similar in that they shrink the solution from the least squares solution towards the direction of the variable space of larger sample variance with lower variability.

[0006]

[0006] Known sources of error in the quantitative analysis of powder mixtures using Raman spectroscopy include day-to-day and daily variations in the Raman instrument, changes in room temperature and humidity, sample fluorescence, mixing, packing, and positioning, as well as sample particle size and compactness. Most problems can be addressed by appropriate spectral processing and data analysis techniques, but it is difficult to completely subtract fluorescence without using any instrument-based methods, even with sophisticated algorithms.

[0007]

[0007] Furthermore, the measured Raman spectra are masked by a strong fluorescence background in many possible applications. The reason above is that the probability of Raman (cross-section) scattering is much lower than that of fluorescence. In other words, Raman scattering and fluorescence emission are two competing phenomena, and the spectrum is typically affected by the most likely phenomenon, which is fluorescence, thus inducing a continuous background for the residual spectrum, and in particular, increasing the photon shot noise that degrades the signal-to-noise ratio, resulting in uncertainty in both cases of material identification and concentration measurement.

[0008] However, Raman scattered photons and fluorescence scattered photons have different lifetimes. Raman photons are observed very instantaneously during excitation (by the laser light), while fluorescence photons can still be detected several nanoseconds or even several milliseconds later. Thus, when scattered photons are collected only during the short Raman scattering phase, the fluorescence background can be suppressed. This can be achieved by irradiating the sample with short, intense laser pulses (having a pulse width much smaller than the fluorescence lifetime), rather than with conventional continuous wave ("CW") radiation, and recording the sample response only during the short pulses. Thus, since fluorescence photons are mainly emitted after Raman scattered photons, the probability of fluorescence photons being detected can be reduced by synchronizing the measurement to the period of the laser pulses. Further, the accuracy of the baseline of the Raman spectrum is improved, which also results in greater accuracy in both material identification and quantitative analysis. A synchronization (or gating) signal is a digital signal or pulse (sometimes called a "trigger") that provides a time window such that a particular event or signal is selected from among many events or signals, and other events or signals are excluded or discarded.

[0009]

[0009] Synchronization can be achieved by various detection systems such as a time-resolved photomultiplier tube, a high-speed optical shutter based on a charge-coupled device (ICCD, "intensified charge-coupled device"), a quantum dot resonant tunneling diode, and a complementary metal-oxide-semiconductor single-photon avalanche diode (CMOS SPAD). One of the essential advantages of CMOS SPAD is that it can remove both optical luminescence tails and photon noise. The SPAD is realized in standard CMOS technology and includes a pn junction reverse-biased above the breakdown voltage of the SPAD. That is, even a single photon, when it enters, can trigger an avalanche breakdown that can be recorded later. The width and position of the time gate need to be appropriately selected. Current CMOS single-photon avalanche diodes are small and inexpensive while being able to achieve sufficient time resolution (sub-nanoseconds). CMOS SPAD detectors have been used to evaluate fluorescence lifetimes. More recently, it has also been demonstrated that CMOS SPAD can be applied to fluorescence removal in the Raman spectroscopy of pharmaceutical products.

[0010]

[0010] Some previous studies have implemented this "time-gating" technique using a high-speed optical shutter or spectrometer based on a charge-coupled device and a mode-locked laser with an intensified CCD (ICCD, "intensified charge-coupled device"). In addition, some analysis has been done to investigate the appropriate gate positions of ICCD and CCD to achieve the best fluorescence removal efficiency. However, these devices are highly sophisticated and either physically large and effective or can only measure a single wavelength of the spectrum at a time, resulting in a long measurement time and thus being unsuitable for on-site applications and unable to be used when the sample moves relative to the Raman spectrometer. To overcome the above problems, CCD and ICCD must be replaced with more appropriate detectors.

[0011]

[0011] For example, problems arise when using a Raman spectrometer to authenticate SERS tags or SERRS tags present in markings (e.g., patterns printed with an ink containing SERS / SERRS tags) adhered to a valuable document such as a banknote. More specifically, the spectrum measured by the Raman spectrometer includes the tag “fingerprint” (i.e., the spectral characteristics that uniquely identify the tag) and additional interference or background information. The SERS or SERRS tag (spectral) fingerprint includes vibration bands represented by a plurality of peaks having a Gaussian / Lorentzian distribution shape at different locations and different widths in the spectrum. The location of the peaks in the spectrum is not absolute but depends on the wavelength of the laser excitation light (due to the shift from the laser wavelength). Raman and SERS / SERRS signals are physical effects different from fluorescence. The substrate of the valuable document (e.g., the paper of a banknote) and the marking (e.g., the ink present on the banknote) have a fluorescence spectrum that can be measured by a Raman spectrometer. When different inks (e.g., multiple prints on a banknote), substrates (e.g., paper), and tags are present in the same measurement track of the spectrometer, the resulting spectral content is accumulated. Thus, the measurement values from the Raman spectrometer generally consist of multiple spectral information resulting from the cumulative effect. Some of the spectral information of the ink, paper, tag, etc. is known (the “known spectral data”) and is stable over time (depending on the banknote design). However, some of the spectral information is unknown (the “unknown spectral data”) and is caused by (varying) external conditions during the measurement process, such as the presence of contaminant gases (e.g., the presence of human perspiration, or even beer, or food traces…) or the presence of dirt spots on the support of the tag. The above unknown spectral information is added during the circulation of the banknote and cannot be anticipated.Furthermore, the above problem is more relevant when measurements are carried out on valuable documents moving at high speeds that require a very short integration time (e.g., 100 - 500 μs), such as when having a high spatial resolution (e.g., several millimeters) while the banknotes are being transported within a banknote sorting device at several m / s (e.g., 10 - 12 m / s or more).

[0012]

[0012] Under such extreme conditions, existing prior art solutions involve, as disclosed in, for example, U.S. Patent No. 10,417,856, using a large number (e.g., 100 or more) of spectral channels for measuring the entire Raman spectrum, together with an absorption wall (for partially absorbing disturbing Rayleigh scattering excitation light) within the Raman spectrometer and a small entrance slit (the smaller the slit, the higher the spectral resolution, and thus the less light hitting the CCD sensor). The problem addressed in this patent document is a situation where the banknotes being processed need to be authenticated by detecting the SERS spectrum of a security taggant. The disclosed solution is to carefully profile all banknotes by using multiple small measurements along the banknote transport path. This profiling requires an integration time of several hundred microseconds, and as a result, the signal readable in this regime is very low (since a compromise on spectral resolution is necessary). An improved discrimination between the Raman spectrum of the taggant and the spectra due to other components of the banknote is disclosed in U.S. Patent Application Publication No. 2007 / 0165209. However, there is still a need for a faster detection of Raman spectra with higher signal levels to provide more reliable diagnostics.

Summary of the Invention

[0013]

[0013] The present invention relates to a method and a corresponding system capable of checking whether a genuine SERS or SERRS taggant having a unique characteristic surface enhancement feature is present in a machine-readable marking on a valuable document (for example, a banknote or a label marked with ink containing the taggant) by using a Raman spectrometer configured to perform Raman spectroscopy (RS) analysis of the marking. The present invention can be used, for example, to authenticate a valuable document or an item marked with an SERS or SERRS taggant according to various processes as follows. The taggant(s) can be present within a specific region, within a part of the substrate of the valuable document or item. For example, in the case of a paper substrate (e.g., a banknote), the taggant can be fixed to the paper fibers within the above region. In this case, the marking containing the taggant(s) is the part of the substrate where the substrate is impregnated with the above taggant(s). The taggant(s) can be mixed with an ink printed on a specific region of the substrate of the valuable document or item. In this case, the marking containing the taggant(s) is the part on the substrate printed with the ink containing the above taggant(s). The taggant(s) can be mixed with a material (e.g., a varnish) deposited (e.g., as a layer) on a specific region of the substrate of the valuable document or item. In this case, the marking containing the taggant(s) is the part on the substrate where the material is deposited. The taggant(s) can be mixed with a specific material of a coating layer deposited on a plastic support. In all cases, the marking applied to the valuable document or item includes a material containing the SERS or SERRS taggant(s) (e.g., a part of the substrate itself containing tagged fibers, or ink printed on the substrate, or a layer of varnish deposited on the substrate...). The method according to the invention enables a highly reliable and fast detection of the presence of genuine SERS or SERRS tagants, and at a given speed, possibly at high speed (e.g., 10 m / s or more), is moving relative to a Raman spectrometer or is only slightly exposed to the Raman spectrometer (e.g., as in a sorting machine), and is particularly suitable for checking the authenticity of valuable documents, such as banknotes, marked with said tagants.

[0014]

[0014] To overcome the above-mentioned drawbacks of the prior art, the present invention is a method for authenticating a marking having a composition comprising a first material deposited on a substrate and containing a SERS tagant or a SERRS tagant, defining an overall model of the Raman spectrum of a genuine marking having a composition deposited on a genuine substrate and containing a genuine first material containing a genuine SERS tagant or a genuine SERRS tagant as a first weighted sum of the reference Raman spectrum of the genuine tagant, the reference Raman spectrum of the reference genuine substrate not marked by the genuine tagant, and the reference Raman spectrum of the reference genuine first material not containing the genuine tagant, wherein the reference Raman spectra are collected upon being irradiated by the excitation light respectively for the genuine tagant, the reference genuine substrate, and the reference genuine first material, the defining step; defining a reduced model of the Raman spectrum of a reduced marking that differs from the genuine marking only in that it does not contain the genuine tagant as a second weighted sum of the reference Raman spectrum of the reference genuine substrate and the reference Raman spectrum of the reference genuine first material; irradiating the marking with excitation light and measuring the corresponding Raman optical signal scattered by the marking via a Raman spectrometer to obtain a measured Raman spectrum of the marking; fitting the measured Raman spectrum to the overall model of the Raman spectrum by calculating a value of a weight, which is a value of the weight within the overall model, and minimizing the difference between the overall model and the measured Raman spectrum under the non-negativity constraint of the weight to obtain a corresponding first residual; a value of a weight in a reduction model, and calculating a value of the weight to minimize a difference between the reduction model and a measured Raman spectrum under a non-negativity constraint of the weight, fitting the measured Raman spectrum to the reduction model of the Raman spectrum to obtain a corresponding second residual; calculating an F value corresponding to an F-test for comparing an overall model of the measured Raman spectrum and the reduction model from the obtained first residual and second residual; determining whether a tagant exists in the marking based on the calculated F value relating to a method including. Accordingly, if the F value corresponds to the presence of a genuine SERS or SERRS tagant in the marking being tested, the marking is considered genuine. If the F value does not correspond to the presence of a genuine SERS or SERRS tagant in the marking being tested, the marking may be considered forged or at least suspicious. The reference genuine substrate differs from the genuine substrate only by not being marked by a genuine (SERS or SERRS) tagant. Similarly, the reference genuine first material differs from the genuine first material only by not containing a genuine (SERS or SERRS) tagant. Of course, if the marking being checked is actually genuine, the first material of the marking and the tagant of the marking also correspond to a genuine first material containing a genuine tagant. The reference genuine substrate mentioned above refers to the corresponding genuine substrate without a marking (for example, the paper substrate of a banknote before printing), and the reference genuine first material refers to the corresponding genuine first material that does not contain any tagant.

[0015]

[0015] The method according to the present invention is particularly configured for cases where the marking is moving relative to the Raman spectrometer during the operation of measuring the Raman optical signal scattered by the marking.

[0016]

[0016] In the above method, the composition of the marking may include a second material, and the weighted sum of each of the overall model and the reduced model is a reference spectrum of the corresponding true second material having a corresponding weight, which is collected by irradiating the true second material with excitation light, and may further include a reference spectrum of the corresponding true second material. The second material (for example, ink) is generally separate from the first material containing the tagant and does not contain the tagant.

[0017]

[0017] In a preferred mode, the Raman spectrometer has a plurality of spectral channels, and the operation of measuring the Raman optical signal scattered by the marking disperses the collected Raman light into a plurality of spectral channels, and an imaging unit acquires a two-dimensional digital image of the dispersed spectral data; a processing unit with a memory performs the following operations on the acquired two-dimensional digital image, namely converting the two-dimensional spectral data into one-dimensional spectral data through line binning and conversion of the binned data into wavelength data; resampling the one-dimensional spectral data to obtain a one-dimensional spectrum with data points equidistant in wavelength; calibrating the one-dimensional spectrum against a reference white light spectrum stored in the memory to obtain a calibrated spectrum; filtering the calibrated spectrum by a low-pass filter to obtain a filtered spectrum; and aligning the filtered spectrum with the reference spectrum of the tagant stored in the memory in terms of wavelength to obtain a pre-processed spectrum, where the aligning operation is performed to pre-process; and the operation of calculating the first residual and the second residual is performed by using the pre-processed spectrum as the measured Raman spectrum. is included. The optical elements and gratings of the Raman spectrometer cause typical (two-dimensional) deformations of the Raman lines formed in the two-dimensional image (the Raman lines are curved and compressed). The line binning and calibration operations are performed to compensate for the above deformations of the Raman lines. The calibration operation is generally performed by the (reference) excitation light delivered by an argon lamp to calculate the two-dimensional deformation of the Raman lines by comparison with the observed image of the argon lines.

[0018]

[0018] According to the above preferred mode, the method is defining a spectral measurement vector as a vector corresponding to the obtained preprocessed spectrum, defining a first spectral vector as the product of a first weight vector and an overall design matrix, and determining non-negative components of the first weight vector that minimize, via the least squares method, a first residual vector corresponding to the difference between the first spectral vector and the spectral measurement vector, wherein the overall design matrix has columns representing reference spectral data of the overall model, respectively, the steps of defining and determining, defining a second spectral vector as the product of a second weight vector and a reduced design matrix, and determining non-negative components of the second weight vector that minimize, via the least squares method, a second residual vector corresponding to the difference between the second spectral vector and the spectral measurement vector, wherein the reduced design matrix has columns representing reference spectral data of the reduced model, respectively, the steps of defining and determining, calculating a first residual sum of squares RSS1 of the error corresponding to the first weight vector having non-negative components of number p1, calculating a second residual sum of squares RSS2 of the error corresponding to the second weight vector having non-negative components of number p2, The step of calculating the F value as the ratio F = ((RSS2 - RSS1) / (p1 - p2)) / (RSS1 / (N - p1)), where the difference between the second residual sum of squares RSS2 and the first residual sum of squares RSS1 is divided by the difference between the numbers p2 and p1, and the first residual sum of squares RSS1 is divided by the difference between the number N of components of the spectral measurement value vector and the number p1 may be included.

[0019]

[0019] Furthermore, the operation of determining the non - negative components of each of the first weight vector and the second weight vector is representing the first weight vector that minimizes the first residual vector as the product of the pseudo - inverse matrix of the overall design matrix and the spectral measurement value vector, and representing the second weight vector that minimizes the second residual vector as the product of the pseudo - inverse matrix of the reduced design matrix and the spectral measurement value vector, and when each component of the first weight vector or the second weight vector has a negative value, respectively modifying the overall design matrix or the reduced design matrix by removing the spectral vector corresponding to the negative component from the above - mentioned matrix, and setting the negative value component to zero, and recalculating the pseudo - inverse matrix of the modified overall design matrix or the modified reduced design matrix until the obtained components of the first weight vector and the second weight vector have only non - negative values may be included.

[0020]

[0020] The present invention further relates to a system operable to perform the steps of the method mentioned above for authenticating a marking having a composition comprising a first material deposited on a substrate and containing an SERS tagant or an SERRS tagant, the system comprising a light source, a Raman spectrometer, an imaging unit, and a control unit having a processing unit and a memory, the light source being controlled by the control unit via a current loop to deliver calibrated excitation light, and the system the operation of irradiating the marking with excitation light delivered by the light source controlled by the control unit The resulting Raman light is collected from the marking, dispersed in a Raman spectrometer having a plurality of spectral channels, and an imaging unit acquires a two-dimensional digital image of the corresponding spectral data, and the acquired spectral data is stored in a memory as the measured Raman spectrum of the marking, and is configured to perform The memory is attached to a genuine substrate and stores, as a first weighted sum of the reference Raman spectra of the genuine tagant, the reference Raman spectrum of the reference genuine substrate not marked by the genuine tagant, and the reference Raman spectrum of the reference genuine first material not containing the genuine tagant, a global model of the Raman spectrum of a genuine marking having a composition containing a genuine first material containing the genuine SERS tagant or the genuine SERRS tagant, wherein the reference Raman spectra are collected upon being irradiated by the excitation light with respect to the genuine tagant, the reference genuine substrate, and the reference genuine first material, respectively. The memory stores, as a second weighted sum of the reference Raman spectrum of the reference genuine substrate and the reference Raman spectrum of the reference genuine first material, a reduction model of the Raman spectrum of a reduced marking, wherein only the composition not containing the genuine tagant is different from the genuine marking. The system, via a processing unit, Calculating a value of a weight that is a value of the weight in the global model and minimizing the difference between the global model and the measured Raman spectrum under the non-negativity constraint of the weight, fitting the measured Raman spectrum stored in the memory to the stored global model of the Raman spectrum to obtain a corresponding first residual, and storing the first residual in the memory; and Calculating a value of a weight that is a value of the weight in the reduction model and minimizing the difference between the reduction model and the measured Raman spectrum under the non-negativity constraint of the weight, fitting the measured Raman spectrum stored in the memory to the stored reduction model of the Raman spectrum to obtain a corresponding second residual, and storing the second residual in the memory; and From the stored first residual and second residual, calculate an F value corresponding to an F-test for comparing the overall model and the reduced model of the measured Raman spectrum, and store it in memory, and Based on the stored F value, determine whether a tagant exists in the marking, and transmit a signal indicating the result of the determination. The system is further configured to perform.

[0021]

[0021] In a preferred embodiment of the system, during the operation of measuring the Raman optical signal scattered by the marking, the marking is moving relative to the Raman spectrometer, and the control unit synchronizes the irradiation of the marking with the light source and synchronizes the acquisition of the measured Raman spectrum via the Raman spectrometer and the imaging unit with the movement of the marking.

[0022]

[0022] In the above system, when the composition of the marking includes a second material, each weighted sum of the overall model and the reduced model is a reference spectrum of the corresponding true second material having a corresponding weight, which is collected by being irradiated with the excitation light of the true second material and stored in memory, and further includes the reference spectrum of the corresponding true second material. For example, in the case of a printed marking, the true second material may correspond to a set of inks that are used to print the marking but do not contain SERS or SERRS tagants.

[0023]

[0023] In the above system, the processing unit Converts the stored two-dimensional digital image into one-dimensional spectral data through line binning and conversion of the binned data into wavelength data, Resamples the one-dimensional spectral data to obtain a one-dimensional spectrum with data points equidistant in wavelength, Calibrates the one-dimensional spectrum against the reference white light spectrum stored in memory to obtain a calibrated spectrum, Filtering the calibrated spectrum by a low-pass filter to obtain a filtered spectrum, and aligning the filtered spectrum with the reference spectrum of the tagant stored in the memory at the wavelength, thereby obtaining a preprocessed spectrum and storing it in the memory, the aligning; performing an operation of calculating a first residual and a second residual by using the preprocessed spectrum stored in the memory as a measurement Raman spectrum may be configured to perform an operation of preprocessing by.

[0024]

[0024] Furthermore, the processing unit defining the spectrum measurement value vector as a vector corresponding to the obtained preprocessed spectrum, and defining a first spectral vector as a product of a first weight vector and an overall design matrix, and determining non-negative components of each of the first weight vectors that minimize a first residual vector corresponding to a difference between the first spectral vector and the spectrum measurement value vector through the least squares method, where the overall design matrix has columns representing reference spectral data of the overall model, respectively, defining and determining; defining a second spectral vector as a product of a second weight vector and a reduced design matrix, and determining non-negative components of each of the second weight vectors that minimize a second residual vector corresponding to a difference between the second spectral vector and the spectrum measurement value vector through the least squares method, where the reduced design matrix has columns representing reference spectral data of the reduced model, respectively, defining and determining; calculating a first residual sum of squares RSS1 of an error corresponding to the first weight vector having non-negative components of number p1, and storing the calculated first residual sum of squares RSS1 and number p1 in the memory; calculating a second residual sum of squares RSS2 of an error corresponding to the second weight vector having non-negative components of number p2, and storing the calculated second residual sum of squares RSS2 and number p2 in the memory; Calculate the F value as the ratio F = ((RSS2 - RSS1) / (p1 - p2)) / (RSS1 / (N - p1)), where RSS2 is the stored second residual sum of squares, RSS1 is the stored first residual sum of squares, p2 and p1 are the stored numbers, and N is the number of components of the spectral measurement value vector. It may be further configured to perform.

[0025]

[0025] The processing unit makes each non - negative component of the first weight vector and the second weight vector represent the first weight vector that minimizes the first residual vector as the product of the pseudo - inverse matrix of the overall design matrix and the spectral measurement value vector, represent the second weight vector that minimizes the second residual vector as the product of the pseudo - inverse matrix of the reduced design matrix and the spectral measurement value vector, If each component of the first weight vector or the second weight vector has a negative value, respectively correct the overall design matrix or the reduced design matrix by removing the spectral vector corresponding to the negative component from the above matrix, set the negative value component to zero, recalculate the pseudo - inverse matrix of the corrected overall design matrix or the corrected reduced design matrix until each obtained component of the first weight vector and the second weight vector has only non - negative values, and store the obtained components in the memory. It may be further configured to determine by.

Brief Description of the Drawings

[0026]

[0026] The present invention will be more fully described hereinafter with reference to the accompanying drawings in which the prominent aspects and features of the present invention are illustrated.

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

[0027]

[0027] To overcome the above-mentioned drawbacks of the prior art, detect the presence of a SERS tagant or a SERRS tagant in a marking adhered to an authenticated substrate, and also determine the amount of a signal (i.e., a very specific peak on the Raman spectrum of the fingerprint) derived from the SERS / SERRS tagant fingerprint in the raw spectral data from the marking measured by a Raman spectrometer, the method according to the present invention compares the measured spectral data from the marking being tested with reference Raman spectral models of various individual materials forming the corresponding genuine marking and the reference Raman spectrum of the reference genuine substrate, and uses a robust quality model capable of reliably determining whether the SERS / SERRS tagant is identified within the marking. If the tagant is identified as genuine within the marking, the marking itself is considered genuine, and more generally, the valuable document containing the marking (adhered to the substrate of the valuable document) is considered genuine.

[0028]

[0028] Additional / undesirable spectral information within the raw spectral data acquired by the Raman spectrometer is divided into two sub-spectrum categories related to the "known spectral data" and "unknown spectral data" respectively mentioned above, in order to improve the signal-to-noise ratio (SNR) and provide a fast and reliable check for the presence of SERS / SERRS tagants on markings attached to valuable documents, corresponding to a high-speed sorting device. The known spectral data is used to model the measured spectral information, while the unknown spectral data, which is expected to be "low-frequency" data, is simply modeled by polynomials (e.g., Legendre polynomials, Jacobi, Gegenbauer, Zernike, Chebyshev, Romanovsky polynomials).

[0029]

[0029] The spectral enhancement effect due to the structure of exactly an example of a SERS tagant is shown in FIG. 2 having a Raman spectrum and a SERS spectrum (the scattering intensity is plotted against the Raman shift in cm -1 units), and the spectral enhancement effect due to the structure of exactly an example of a SERRS tagant is shown in FIG. 3 (the Raman spectrum is scaled by a factor of 8 so as to overlap better with the SERRS spectrum). On both drawings, characteristic enhanced Raman scattering intensity peaks are clearly visible and are specific to the structure of the nanoparticles exactly forming the tagant, so as to constitute distinctive features (i.e., the "fingerprint" of the tagant).

[0030]

[0030] According to an exemplary embodiment of the present invention, the authentication marking (pattern) is printed on the paper substrate of the banknote with several inks. If the marking (and thus the banknote) is genuine, each (genuine) ink composition is known, and a genuine SERS taggant with a known (reference) Raman spectrum is added to one of the inks so as to be printed on the banknote. The ink containing the SERS taggant corresponds to the first material mentioned above, and the second material mentioned above corresponds to the other ink(s). In this particular embodiment, there are four distinct inks (each having a specific composition) present within the marking, and each ink has a known (reference) Raman spectrum if it is genuine and does not contain a taggant. The reference Raman spectra of the genuine SERS taggant, the reference genuine paper substrate (of the corresponding genuine banknote), and each of the four reference genuine inks can be measured by a Raman spectrometer upon irradiation with the excitation light (here laser light) of the genuine SERS taggant, the reference genuine paper substrate, and each of the four reference genuine inks, respectively. Thereafter, the reference Raman spectra are used to derive an overall model of the Raman spectrum of a general genuine marking as a linear combination of different reference spectra. Each reference spectrum corresponds to the acquisition of a specific number of values of the scattered high intensity at different wavelengths via the Raman spectrometer. Thus, an interpolated reference spectral curve giving the measured scattered intensity I as a function of wavelength λ, i.e., I(λ), can be obtained for each of the genuine SERS taggant, the reference genuine paper substrate, and the four reference genuine inks mentioned above. For simplicity, it is assumed that the same number n (for example, n = 1024) of reference intensity values (corresponding to n distinct wavelength values) are extracted from each reference spectral curve.

[0031]

[0031] In the overall model of the Raman spectrum of a general genuine marking (by four genuine inks) applied to the (genuine) substrate, the (separate) representation of the spectral curve includes n Raman intensity values I i (i = 1, …, n) (taken along the spectral curve), and each intensity value I iis modeled as a linear combination of (p1 - 1) reference Raman intensity values X i2 , …, Xi p1 (for i = 1, …, n, when X i1 ≡ 1). In a particular embodiment, p1 = 7 (p1 is the number of independent variables in the model). Thus, I i = β 1 X i1 + β 2 X i2 + … + β 7 X i7 where β 1 , …, β 7 are weights, and X i2 (i = 1, …, n) are the n intensity values at selected representative points along the (normalized) reference Raman spectrum of the genuine SERS tagant. The selected points are within a wavelength band of approximately 150 nm width within the NIR range (near infrared, 750 - 1400 nm). Normalization of the spectral curve is obtained by taking the difference between the measured value and the minimum of the measured values and removing the offset value (data that is generally not centered at 0 on the vertical axis) by setting the highest peak point to, for example, 1000. X i3 (i = 1, …, n) are the n intensity values at selected representative points along the (normalized) reference Raman spectrum of the reference genuine paper. X i4 , …, X i7 (i = 1, …, n) are the respective n intensity values at selected representative points along the (normalized) reference Raman spectrum of each of the four reference genuine inks used for printing the genuine marking (each of the four reference genuine inks is considered to be standalone, i.e., not containing the SERS tagant). In vector notation, the vector I can be associated with n scalar components I i (i = 1, …, n), and the vector β is associated with p1 (here, p1 = 7) scalar weights β 1 , β 2 , …, β p1can be associated with, and the (nXp1) matrix X can be associated with the overall model. The first column of the matrix contains n values X i1 = 1 (i = 1, …, n), and the second to p1-th columns are the components X i2 (i = 1, …, n), …, X ip1 (i = 1, …, n), respectively. Thus, the expression of the Raman spectrum in the overall model is I = Xβ.

[0032]

[0032] According to the present invention, the "reduced" marking is a marking applied to a (true) paper substrate that is different from the true marking in that it does not contain a (true) SERS tagant. Thus, in the reduced model of the Raman spectrum of the reduced marking, (p2 - 1) reference Raman intensity values (for i = 1, …, n, when Z i1 ≡ 1) Z i2 , …, Z ip2 along the spectral curve modeled as n Raman intensity values J i (i = 1, …, n), where p2 = 6. Thus, J i = μ 1 Z i1 + μ 2 Z i2 + … + μ 6 Z i6 where μ 1 , …, μ 6 are weights, and Z i2 (i = 1, …, n) are the n intensity values at the selected representative points along the (normalized) reference Raman spectrum of the reference true paper. Z i3 , …, Z i6 (i = 1, …, n) are the respective n intensity values at the selected representative points along the (normalized) reference Raman spectra of the four inks (of course, not containing SERS tagant) used to print the reduced marking. In fact, by the definition of the reduced model, here (for i = 1, …, n), Z i1 = X i1equals 1, and for k = 2, …, p2, Z ik = X i(k+1) is. In vector notation, the vector J can be associated with n scalar components J i (i = 1, …, n), and the vector μ can be associated with p2 (here, p2 = 6) scalar weights μ 1 , μ 2 , …, μ p2 and the (n X p2) matrix Z can be associated with the reduced model, and the first column of the matrix contains n values Z i1 = 1 (i = 1, …, n), and the second to p2-th columns are formed by the components Z i2 (i = 1, …, n), …, Z ip2 (i = 1, …, n), respectively. Thus, the expression of the Raman spectrum in the reduced model is J = Zμ.

[0033]

[0033] The marking on the authenticated banknote is irradiated by the excitation light, and the corresponding Raman optical signal scattered by the marking is measured by a Raman spectrometer to obtain the measured Raman spectrum of the marking. Preferably, a Raman spectrometer equipped with a multimode laser source (MML) is used. In fact, even when it is common practice to use a single mode laser (SML) source to obtain the best possible resolution, it has been shown empirically that, in fact, the detection speed is improved by using an MML source. For example, the laser output can be increased by a factor of 10 (when there is no compromise) compared to an SML source, while the integration time of the measurement is reduced to one tenth (e.g., it can reach 0.2 ms instead of 2 ms). This is due to two main differences between the SML source and the MML source, namely, the laser output (e.g., when the MML is much higher, e.g., about 1 W, the SML is about 100 mW at 760 nm) and the linewidth (when the MML is 0.08 nm, the SML is about 0.02 nm).

[0034]

[0034] The measured Raman spectrum gives the measured (Raman) scattered light intensity Y, i.e., Y(λ), as a function of the scattered light wavelength λ. The Raman spectrometer has a plurality of spectral channels, and the Raman optical signal scattered by the marking and collected by the spectrometer is first dispersed (via a grating) into the spectral channels, and the imaging unit (CCD) acquires a two-dimensional digital image of the corresponding dispersed spectral data as a two-dimensional array of intensity values versus wavelength, i.e., two-dimensional spectral data. Since the acquired two-dimensional spectral data from the Raman spectrometer is raw, it is further preprocessed mainly by the processing unit (to reduce the amount of data to be analyzed later (to reduce the processing time and to correspond to banknote detection in a high-speed sorting device)), the signal-to-noise ratio (SNR) is improved, and the Raman bands of the taggant fingerprint are precisely located.

[0035]

[0035] The preprocessing steps of the two-dimensional digital image acquired by the imaging unit are performed by a processing unit equipped with a memory and include the following operations. 1) An operation of converting the acquired two-dimensional spectral data into one-dimensional spectral data by line binning and conversion of the binned data into wavelength data. This conversion greatly reduces the amount of data to be processed and improves the SNR (the noise is typically reduced by a factor equal to the square root of the amount of pixels within a column of the two-dimensional digital image). 2) An operation of resampling the obtained one-dimensional spectral data to form a one-dimensional spectrum in which the data points are equidistant in wavelength. This operation is performed via spline or polynomial interpolation of the spectral data. This resampling has the advantage that the spectral compression along the abscissa axis is reduced, and also provides a linear derivation of the spectrum that enables the use of well-known signal processing tools (such as low-pass filtering by FFT convolution, FIR convolution, etc.). 3) Operating to calibrate the resampled one-dimensional spectrum against a reference white light spectrum stored in memory (for example, from a quartz tungsten halogen lamp to balance the sensitivity of the Raman spectrometer) to obtain a calibrated (one-dimensional) spectrum. This operation enables the balancing of the light intensity delivered by the Raman spectrometer (generally, spectrometers output different values for the same light intensity at different wavelengths). 4) Operating to filter the calibrated spectrum with a low-pass filter to obtain a filtered spectrum. In fact, very high-frequency noise in the spectral data mainly originates from the imaging unit (i.e., the image sensor of the imaging unit and the circuit of the imaging unit) and is known to be a measurement artifact. This filtering can be performed through various methodologies such as using a moving average filter, or an FFT (Fast Fourier Transform) filter, or a Savitzky-Golay filter. Preferably, FFT filtering is used (since this method can also be used for spectral alignment). 5) Operating to align the filtered spectrum with the reference spectrum of the genuine tagant stored in memory at the wavelength. In fact, the stored reference Raman spectrum of the genuine tagant is generally not aligned with the Raman spectrum measured from the marking due to many possible causes such as, for example, expansion of the spectrometer, fluctuations in the light source wavelength and / or temperature affecting the grating, and mechanical perturbations due to vibrations. Therefore, for the best possible verification of the tagant fingerprint, the measured Raman spectrum obtained from the marking is aligned with the reference spectrum at the wavelength. This alignment can be achieved by different methods as follows. Executing the algorithm at different shift increments and selecting the best position on the wavelength axis. Executing the algorithm at different shift increments and interpolating to find the best position on the wavelength axis. Preferably, in the frequency domain, convolution with the tagant fingerprint is performed. Monitoring the position of the light source during measurement from the marking. As a result of the above operation, a preprocessed Raman spectrum is obtained from the two-dimensional spectral data acquired by the imaging unit. The (separate) representation of the spectral curve Y(λ) of the preprocessed Raman spectrum includes n (preprocessed) Raman intensity values Y i (i = 1, …, n) (taken along the spectral curve), and the (n-dimensional) vector Y can be associated with n scalar components Y i (i = 1, …, n).

[0036]

[0036] To fit the (preprocessed) Raman spectrum to the global model, the spectral (measurement) vector Y is decomposed as Y = I + ε (linear regression analysis), the first spectral vector is I = Xβ, X is the nXp1 (design) matrix of the global model, β is the corresponding first weight vector, and ε is the error vector or residual vector having components ε i (i = 1, …, n). The components β k (k = 1, …, p1, where p1 = 7) of the first weight vector β that minimizes the error vector ε can be determined through various known optimization methods. For example, it is possible to (iteratively) calculate the residual vectors for multiple selected values of the components of the vector β and select the vector β corresponding to the residual vector having a lower norm. Another method is to use a well-known optimization algorithm such as Dantzig's simplex algorithm. Preferably, a method of least squares residual (LSR) with lower computational intensity for CPU calculation and thus more suitable for the authentication of markings on banknotes in a high-speed sorting machine is used.

[0037]

[0037] Similarly, to fit the (preprocessed) Raman spectrum to the reduction model, the vector Y is decomposed as Y = J + ε’, the second spectral vector is J = Zμ, Z is the nXp2 (design) matrix of the reduction model, μ is the corresponding second weight vector, and ε’ is the error vector or residual vector having components ε’ i (i = 1, …, n). The components μ m (m = 1, …, p2, where p2 = (p1 - 1) = 6) of the second weight vector μ that minimizes the error vector ε’ can be determined via the method of least squares residuals (LSR).

[0038]

[0038] According to the LSR method, the least squares parameter estimates of β (or μ of the reduction model) of the overall model with respect to the measured values Y are obtained from p1 (or p2) normal equations. ε i = Y i - β 1 X i1 + β 2 X i2 + … + β p1 X ip1 (i = 1, …, n), that is, ε = Y - Xβ, and

Equation

Equation

Equation

Number

Number

Number

Number

Number

[0039]

[0039] However, the LSR method mentioned above has a problem that it does not consider whether the obtained solution is "feasible". In fact, when the solution includes negative values of the weight component β j (j ∈ {2, …, 7}) of the vector β or the weight component μ r (r ∈ {2, …, 6}) of the vector μ, the intensity of the related spectral component becomes negative, which is not physically possible (this constitutes an infeasible solution). It has been observed that the authentication method becomes much more robust when a specific minimization method is used to comply with the non-negativity constraint (NNC) on the weight values. For example, some methods incorporating the above non-negativity constraint are known, such as the active constraint method (detailed in the book "Solving Least Square Problems" by Charles L. Lawson and Richard J. Hanson (SIAM 1995)), or Landweber's gradient descent method. According to the present invention, the LSR method is combined with the following method shown in FIG. 5 in order to comply with the non-negativity constraint. This is explained in the case of the overall model having p1 = 7 weights, and can be directly replaced with the case of the reduced model (having p2 = 6 weights) by making changes where necessary. The method for calculating the values of the p1 components β [Number] of the weight vector 1 β 7 β, …, is to calculate the values from the pseudo-inverse matrix X + stored in the memory of the processing unit and the spectral measurement vector Y, i.e., [Number] is started by (S1), and then the first calculated weight vector [Number] a check (S2) is performed to determine whether there are negative weight values in 2 and β 6 has negative values (corresponding to SERS tagant and the third ink respectively), and the value of weight β 2 is set to zero (S3), and the corresponding column of the design matrix X, that is, the column corresponding to the Raman spectrum of the (true) SERS tagant (components X 12 , …, X n2 having) is removed from the (initial) design matrix X (S4), and thus a new nX(p1 - 1) design matrix X’ is obtained. Then, the corresponding new pseudo-inverse matrix X’ + is calculated (S5), and the new weight vector [Number] is used to calculate by (S6), and the new weight vector has only (p1 - 1) components β’ [Number] β’ 1 , β’ 3 , β’ 4 , β’ 5 , β’ 6 and β’ 7 only (because β 2 is set to zero). Then, a check (S7) is performed to determine whether there are negative weight values (Yes "Y") or not (No "N") in the calculated weight vector [Number] In the example shown in Fig. 5, one weight β’6 has a negative value (corresponding to the third ink), and the weight β’ 6 is set to zero (S8), and the corresponding column of the design matrix X’, i.e., the column corresponding to the Raman spectrum of the (reference true) third ink (component X 16 , …, X n6 having) is removed from the design matrix X’ (S9), and thus a new nX(p1 - 2) design matrix X’’ is obtained. Then, the corresponding new pseudo-inverse matrix X’’ + is calculated (S10), and the new weight vector

Number

Number

Number

Number

[0040]

[0040] The first weight vector

Number

Number

Number

Number

[0041]

[0041] Generally, the value of F depends on the SNR as follows. When the SNR is low and there is a SERS tagant in the marking, the value of F is low. This is normal because random noise affects the fingerprint for distinguishing SERS tagants in the same way. When the SNR is low and there is no (genuine) SERS tagant in the marking, the value of F is low. When the SNR is high and there is a (genuine) SERS tagant in the marking, the value of F is high. When the SNR is high and there is no (genuine) SERS tagant in the marking, the value of F is low. Since the trend between the SNR and the value F is linear, when the value of F is between 8000 and 1000000, it is not very suitable for determining the authenticity of the marking. In order to create a plateau on the curve representing the dependence of the value F on the SNR, a further "compression" step can be applied to correct the value of F. In this embodiment, the corrected ( "compressed") value F' is obtained through the transformation F' = constant x Log(F). For example, the value of the constant coefficient is 5. From a series of experiments, the following can be reliably concluded. Values of F' (e.g., 1 - 20) below about 20 low threshold values (LTV) correspond to the absence of (genuine) SERS taggants in the marking and indicate a negative determination D that the corresponding banknote is not genuine. - is delivered. Values of F' (e.g., 50 - 80) above about 50 high threshold values (LTV) indicate the presence of SERS taggants in the marking and indicate an affirmative determination D that the corresponding banknote is genuine. + is delivered. However, intermediate values of F' (e.g., between the low threshold LTV and the high threshold HTV) cannot enable a conclusion to be drawn (the result depends greatly on the level of the SNR). In the latter case, it is not possible to determine whether a SERS taggant is present in the marking, and thus it is not possible to determine whether the banknote is genuine. Therefore, the banknote is reserved (R) for more detailed (e.g., forensic) analysis.

[0042] The steps of the above - preferred embodiment of the method for authenticating a marking having a composition deposited on a substrate and containing an ink and a SERS taggant (or a SERRS taggant) are summarized in FIG. 4. The method starts at (M0), where the values p1 (p1≧4) and p2=(p1 - 1) of the number of reference Raman spectra in the global model and the reduced model are specified and stored in the memory of the processing unit (M1), and the number n of points on the measured Raman spectrum is taken. In step (M2), the Raman spectra X of each of the global model and the reduced model i2 , …, Xip1 (i = 1, …, n) and Z i2 …, Z ip2 is specified, and the corresponding full design matrix X and reduced design matrix Z are stored. In step (M3), the corresponding pseudo-inverse X + of the full design matrix and the pseudo-inverse Z + of the reduced design matrix are calculated and stored. Next, in step (M4), the measured Raman spectrum is obtained from the marking through a two-dimensional image obtained by the imaging unit of the Raman spectrometer (upon receiving that the marking is irradiated by the excitation laser beam), preprocessed, and a one-dimensional spectrum is obtained, and a corresponding spectral measurement value vector Y having n components is formed. The LSR method with the NNC method (i.e., LSR-NNC) is implemented in step (M5) to obtain the first weight vector

Number

Number

Number

Number

[0043]

[0043] When the marking includes a plurality of SERS tagants or SERRS tagants, the determination of authenticity based on only a single F value may not be sufficiently reliable. According to the present invention, it is possible to calculate different F values in order to use a plurality of different reduction models and compare the overall model of the genuine marking (i.e., including a plurality of reference spectra of various tagants) with each of the reduction models. For example, different reduction models may correspond to markings different from the genuine marking only by the absence of one of the plurality of different tagants of the genuine marking. These F values are obtained from the (pre-processed) spectral vector Y obtained from the measured Raman spectrum of the marking authenticated by applying the LSR-NNC method mentioned above to find different weight vectors that minimize the squares of the corresponding residual vectors. The determination of the authenticity of the marking must include different threshold rules for each of the calculated F values, resulting in a certain degree of complexity. In this case, the determination of authenticity may preferably be based on a decision tree incorporating the above threshold rules.

[0044]

[0044] The present invention also relates to a system (60) in which a particular embodiment is shown in FIG. 6, the system (60) comprising a light source (61), a Raman spectrometer (62), an imaging unit (63), a processing unit (64), a memory unit (65), and a control unit (66). The control unit (66) controls the light source (61) (here a laser) via a current loop to deliver calibrated excitation light and irradiate the marking (67) when the (moving) marking (67) on the banknote (68) reaches the level of the imaging unit (65). The laser excitation light is delivered to the marking (67) via a dichroic mirror (69). In response to the irradiation, Raman light is scattered from the marking and collected via the dichroic mirror (69) and dispersed via a grating (70) towards the CCD sensor (71) of the imaging unit (63). A corresponding two-dimensional digital image of the collected Raman spectrum is formed by the imaging unit (63) and constitutes a 2D measured Raman spectrum stored in the memory unit (65). The memory unit (65) stores the overall model of the genuine marking (applied to the genuine substrate of the genuine banknote), i.e., the number n of points of the selection criterion spectrum, the number p1 of weights forming the first weight vector β, and the number p2 of weights forming the second weight vector μ as described above. The reference spectrum of the overall model is stored as a component of the overall (design) matrix X, and the reference spectrum of the reduced model is stored as a component of the reduced (design) matrix Z. The memory unit (65) further stores the reduced model, the pre-computed pseudo-inverses X + and Z + of the matrices X and Z respectively. The stored two-dimensional measured Raman spectrum is pre-processed via the processing unit (64) to obtain a (one-dimensional) pre-processed spectrum in the form of a spectral measurement value vector Y having n components stored in the memory unit (65) as described above. The processing unit 64 then minimizes the square of the first residual vector ε = Y - Xβ, the first weight vector corresponding to the overall model

Number

Number

Number

Number

[0045] The subject matter disclosed above should be considered illustrative rather than limiting and serves to provide a better understanding of the invention as defined by the independent claims. [Claims of the Invention] [Claim 1] A method for authenticating a marking having a composition comprising a first material deposited on a substrate and containing a SERS tagant or a SERRS tagant, defining an overall model of the Raman spectrum of a genuine marking having a composition deposited on a genuine substrate and containing a genuine first material containing a genuine SERS tagant or a genuine SERRS tagant as a first weighted sum of the reference Raman spectrum of the genuine tagant, the reference Raman spectrum of a reference genuine substrate not marked by the genuine tagant, and the reference Raman spectrum of a reference genuine first material not containing the genuine tagant, wherein the reference Raman spectra are collected upon being irradiated by excitation light with respect to the genuine tagant, the reference genuine substrate, and the reference genuine first material, respectively; defining a reduction model of the Raman spectrum of a reduced marking, in which only the composition not containing the genuine tagant is different from the genuine marking, as a second weighted sum of the reference Raman spectrum of the reference genuine substrate and the reference Raman spectrum of the reference genuine first material; irradiating the marking with the excitation light, measuring a corresponding Raman optical signal scattered by the marking via a Raman spectrometer to obtain a measured Raman spectrum of the marking; fitting the measured Raman spectrum to the overall model of the Raman spectrum by calculating a value of a weight, which is a value of the weight in the overall model and minimizes the difference between the overall model and the measured Raman spectrum under the non-negativity constraint of the weight, to obtain a corresponding first residual; fitting the measured Raman spectrum to the reduction model of the Raman spectrum by calculating a value of a weight, which is a value of the weight in the reduction model and minimizes the difference between the reduction model and the measured Raman spectrum under the non-negativity constraint of the weight, to obtain a corresponding second residual; calculating an F value corresponding to an F-test for comparison between the overall model and the reduction model of the measured Raman spectrum from the obtained first residual and second residual; Determining whether the tagant exists in the marking based on the calculated F value A method characterized by including the above steps [Item 2] The method according to item 1, wherein during the operation of measuring the Raman optical signal scattered by the marking, the marking is moving relative to the Raman spectrometer [Item 3] The composition of the marking includes a second material, and the weighted sums of the overall model and the reduced model are reference spectra of corresponding true second materials with corresponding weights, which are collected by irradiating the true second materials with the excitation light. The method according to item 1 or 2, further including reference spectra of corresponding true second materials [Item 4] The Raman spectrometer has a plurality of spectral channels, and the operation of measuring the Raman optical signal scattered by the marking includes Dispersing the collected Raman light into the plurality of spectral channels, and acquiring a two-dimensional digital image of the dispersed spectral data by an imaging unit By a processing unit with a memory, the acquired two-dimensional digital image is subjected to the following operations, namely Converting the two-dimensional spectral data into one-dimensional spectral data through line binning and conversion of the binned data into wavelength data Resampling the one-dimensional spectral data to obtain a one-dimensional spectrum with data points equidistant in wavelength Calibrating the one-dimensional spectrum against a reference white light spectrum stored in the memory to obtain a calibrated spectrum Filtering the calibrated spectrum by a low-pass filter to obtain a filtered spectrum, and Aligning the filtered spectrum with the reference spectrum of the tagant stored in the memory in wavelength to obtain a preprocessed spectrum, the aligning operation By performing the above operations, a preprocessing step Calculating the first residual and the second residual by using the preprocessed spectrum as the measured Raman spectrum The method according to any one of items 1 to 3, including the above steps [Item 5] Defining a spectral measurement value vector as a vector corresponding to the obtained preprocessed spectrum Define the first spectral vector as the product of the first weight vector and the overall design matrix, and determine the non - negative components of each of the first weight vectors that minimize the first residual vector corresponding to the difference between the first spectral vector and the spectral measurement value vector via the least - squares method, wherein the overall design matrix has columns representing the reference spectral data of the overall model, the steps of defining and determining; Define the second spectral vector as the product of the second weight vector and the reduced design matrix, and determine the non - negative components of each of the second weight vectors that minimize the second residual vector corresponding to the difference between the second spectral vector and the spectral measurement value vector via the least - squares method, wherein the reduced design matrix has columns representing the reference spectral data of the reduced model, the steps of defining and determining; Calculate the first residual sum of squares RSS1 of the error corresponding to the first weight vector having p1 non - negative components; Calculate the second residual sum of squares RSS2 of the error corresponding to the second weight vector having p2 non - negative components; Calculate the F - value as the ratio F = ((RSS2 - RSS1) / (p1 - p2)) / (RSS1 / (N - p1)), where the numerator is the value obtained by dividing the difference between the second residual sum of squares RSS2 and the first residual sum of squares RSS1 by the difference between p2 and p1, and the denominator is the value obtained by dividing the first residual sum of squares RSS1 by the difference between the number N of components of the spectral measurement value vector and p1; The method according to item 4, comprising; [Item 6] The steps of determining the non - negative components of each of the first weight vector and the second weight vector are; Represent the first weight vector that minimizes the first residual vector as the product of the pseudo - inverse matrix of the overall design matrix and the spectral measurement value vector, and represent the second weight vector that minimizes the second residual vector as the product of the pseudo - inverse matrix of the reduced design matrix and the spectral measurement value vector; If any of the components of the first weight vector or the second weight vector has a negative value; Modify each of the overall design matrix or the reduced design matrix by removing the spectral vector corresponding to the negative component from the matrix; the step of setting the negative value component to zero; the step of recomputing the pseudo-inverse matrix of the modified overall design matrix or the modified reduced design matrix until the obtained components of the first weight vector and the second weight vector have only non-negative values; The method according to item 5, comprising the above steps. [Item 7] A system for authenticating a marking having a composition including a first material deposited on a substrate and including an SERS tagant or an SERRS tagant, the system comprising a light source, a Raman spectrometer, an imaging unit, and a control unit having a processing unit and a memory, the light source being controlled by the control unit via a current loop to deliver calibrated excitation light, the system irradiating the marking with the excitation light delivered by the light source controlled by the control unit; collecting Raman light resulting therefrom from the marking, dispersing the collected Raman light in the Raman spectrometer having a plurality of spectral channels, and obtaining, by the imaging unit, a two-dimensional digital image of corresponding spectral data, and storing the obtained spectral data in the memory as the measured Raman spectrum of the marking; is configured to perform the above operations; the system wherein the memory stores a global model of the Raman spectrum of a genuine marking having a composition including a genuine first material deposited on a genuine substrate and including a genuine SERS tagant or a genuine SERRS tagant as a first weighted sum of a reference Raman spectrum of the genuine tagant, a reference Raman spectrum of the reference genuine substrate not marked by the genuine tagant, and a reference Raman spectrum of the reference genuine first material not including the genuine tagant, the reference Raman spectra being collected upon being irradiated by the excitation light with the genuine tagant, the reference genuine substrate, and the reference genuine first material, respectively; wherein the memory stores a reduced model of the Raman spectrum of a reduced marking that differs from the genuine marking only in that the composition does not include the genuine tagant as a second weighted sum of the reference Raman spectrum of the reference genuine substrate and the reference Raman spectrum of the reference genuine first material. The system, via the processing unit, fits the measured Raman spectrum stored in the memory to the stored overall model of the Raman spectrum by calculating a value of a weight that is a value of a weight in the overall model and that minimizes a difference between the overall model and the measured Raman spectrum under a non-negativity constraint of the weight, obtains a corresponding first residual, and stores the first residual in the memory; fits the measured Raman spectrum stored in the memory to the stored reduced model of the Raman spectrum by calculating a value of a weight that is a value of a weight in the reduced model and that minimizes a difference between the reduced model and the measured Raman spectrum under a non-negativity constraint of the weight, obtains a corresponding second residual, and stores the second residual in the memory; calculates an F value corresponding to an F-test for comparison between the overall model and the reduced model of the measured Raman spectrum from the stored first residual and the stored second residual, and stores the F value in the memory; determines whether or not the tagant is present in the marking based on the stored F value, and delivers a signal indicating a result of the determination The system is further configured to perform. [Item 8] During the operation of measuring the Raman optical signal scattered by the marking, the marking is moving relative to the Raman spectrometer, and the control unit synchronizes the irradiation of the marking with the light source and synchronizes the acquisition of the measured Raman spectrum via the Raman spectrometer and the imaging unit with the movement of the marking. The system according to Item 7. [Item 9] The composition of the marking includes a second material, and each weighted sum of the overall model and the reduced model is a reference spectrum of a corresponding true second material having a corresponding weight, the reference spectrum being collected by being irradiated with the excitation light of the corresponding true second material and stored in the memory. The system according to Item 7 or 8, further including a reference spectrum of a corresponding true second material. [Item 10] The processing unit converts the stored two-dimensional digital image into one-dimensional spectral data via line binning and conversion of the binned data into wavelength data of the spectral data; Resampling the one-dimensional spectral data to obtain a one-dimensional spectrum in which data points are equidistant in wavelength; Calibrating the one-dimensional spectrum with respect to the reference white light spectrum stored in the memory to obtain a calibrated spectrum; Filtering the calibrated spectrum by a low-pass filter to obtain a filtered spectrum; Aligning the filtered spectrum with the reference spectrum of the tagant stored in the memory at the wavelength, thereby obtaining a preprocessed spectrum and storing it in the memory; By using the preprocessed spectrum stored in the memory as the measured Raman spectrum, performing an operation of calculating the first residual and the second residual The system according to any one of items 7 to 9, which is configured to perform an operation of preprocessing by [Item 11] The processing unit Defining the spectral measurement value vector as a vector corresponding to the obtained preprocessed spectrum; Defining the first spectral vector as the product of the first weight vector and the overall design matrix, and determining each non-negative component of the first weight vector that minimizes the first residual vector corresponding to the difference between the first spectral vector and the spectral measurement value vector through the least squares method, wherein the overall design matrix has columns representing the reference spectral data of the overall model respectively; defining and determining; Defining the second spectral vector as the product of the second weight vector and the reduced design matrix, and determining each non-negative component of the second weight vector that minimizes the second residual vector corresponding to the difference between the second spectral vector and the spectral measurement value vector through the least squares method, wherein the reduced design matrix has columns representing the reference spectral data of the reduced model respectively; defining and determining; Calculating the first residual sum of squares of the error corresponding to the first weight vector having p1 non-negative components, and storing the calculated first residual sum of squares RSS1 and the number p1 in the memory; Calculate a second residual sum of squares RSS2 of the error corresponding to the second weight vector having non-negative components of the number p2, and store the calculated second residual sum of squares RSS2 and the number p2 in the memory, calculate the F value as the ratio F = ((RSS2 - RSS1) / (p1 - p2)) / (RSS1 / (N - p1)), where the F value is the value obtained by dividing the difference between the stored second residual sum of squares RSS2 and the stored first residual sum of squares RSS1 by the difference between the stored numbers p2 and p1, and the stored first residual sum of squares RSS1 is divided by the difference between the number of components N of the spectral measurement value vector and the number p1, The system according to item 10, further configured to perform. [Item 12] The processing unit, for each of the non-negative components of the first weight vector and the second weight vector, represent the first weight vector that minimizes the first residual vector as the product of the pseudo-inverse matrix of the overall design matrix and the spectral measurement value vector, represent the second weight vector that minimizes the second residual vector as the product of the pseudo-inverse matrix of the reduced design matrix and the spectral measurement value vector, if the components of the first weight vector or the second weight vector respectively have negative values, modify the overall design matrix or the reduced design matrix respectively by removing the spectral vector corresponding to the negative component from the matrix, set the negative value component to zero, recalculate the pseudo-inverse matrix of the modified overall design matrix or the modified reduced design matrix until the obtained components of the first weight vector and the second weight vector have only non-negative values, and store the obtained components in the memory, The system according to item 11, further configured to determine by.

Claims

1. A method for authenticating a marking deposited on a substrate and having a composition comprising a first material including a SERS taggant or a SERRS taggant, comprising a system comprising a light source, a Raman spectrometer, an imaging unit, and a control unit having a processing unit and a memory, the light source being controlled by the control unit via a current loop to deliver calibrated excitation light, the system performing the following steps: storing in said memory an overall model of a Raman spectrum of a genuine marking deposited on a genuine substrate and having a composition including a genuine SERS taggant or a genuine first material including a genuine SERRS taggant as a first weighted sum of a reference Raman spectrum of the genuine taggant, a reference Raman spectrum of a reference genuine substrate not marked with the genuine taggant, and a reference Raman spectrum of a reference genuine first material not including the genuine taggant, said reference Raman spectra being collected in response to the genuine taggant, the reference genuine substrate, and the reference genuine first material being each illuminated with excitation light; storing in the memory a reduction model of the Raman spectrum of a reduced marking that differs from the authentic marking only by its composition that does not include the authentic taggant as a second weighted sum of the reference Raman spectrum of the reference authentic substrate and the reference Raman spectrum of the reference authentic first material; illuminating the marking with the excitation light and measuring a corresponding Raman light signal scattered by the marking via a Raman spectrometer to obtain a measured Raman spectrum of the marking; fitting, by the processing unit, the measured Raman spectrum to the global model of the Raman spectrum by calculating weight values ​​in the global model that minimize a difference between the global model and the measured Raman spectrum under a non-negativity constraint on the weights, to obtain a corresponding first residual; fitting, by the processing unit, the measured Raman spectrum to the reduced model of the Raman spectrum by calculating values ​​of weights in the reduced model that minimize a difference between the reduced model and the measured Raman spectrum under a non-negativity constraint on the weights, to obtain a corresponding second residual; calculating, by the processing unit, from the obtained first residual and the second residual, an F-value corresponding to an F-test of a comparison between the full model and the reduced model of the measured Raman spectrum; determining, by the processing unit, whether the taggant is present in the marking based on the calculated F-value; A method comprising:

2. The method of claim 1 , wherein the marking is moving relative to the Raman spectrometer during an operation of measuring the Raman optical signal scattered by the marking.

3. 3. The method of claim 1 or 2, wherein the composition of the marking includes a second material, and the weighted sum of each of the overall model and the reduced model further includes a reference spectrum of a corresponding authentic second material having a corresponding weight, the reference spectrum being collected upon illumination of the authentic second material by the excitation light.

4. The Raman spectrometer has a plurality of spectral channels, and the operation of measuring the Raman optical signal scattered by the marking comprises: dispersing the collected Raman light into the plurality of spectral channels and acquiring, with the imaging unit, a two-dimensional digital image of the dispersed spectral data; The processing unit processes the acquired two-dimensional digital image through the following operations: converting the two-dimensional digital image of the dispersed spectral data into one-dimensional spectral data via line binning and converting the binned data into wavelength data; resampling the one-dimensional spectral data to obtain a one-dimensional spectrum with data points equidistant in wavelength; calibrating the one-dimensional spectrum against a reference white light spectrum stored in the memory to obtain a calibrated spectrum; filtering the calibrated spectrum with a low pass filter to obtain a filtered spectrum; and aligning the filtered spectrum in wavelength with a reference spectrum for the taggant stored in the memory, thereby obtaining a pre-processed spectrum; and pre-processing the image by performing performing an operation of calculating the first residual and the second residual by using the preprocessed spectrum as the measured Raman spectrum; The method according to any one of claims 1 to 3, comprising:

5. defining a spectral measurement vector as a vector corresponding to the obtained pre-processed spectrum; defining a first spectral vector as the product of a first weight vector and an overall design matrix and determining non-negative components of each of the first weight vectors that minimize a first residual vector corresponding to a difference between the first spectral vector and the spectral measurement vector via a least squares method, the overall design matrix having columns each representing reference spectral data for the overall model; defining a second spectral vector as the product of a second weight vector and a reduction design matrix and determining non-negative components of each of the second weight vectors that minimize a second residual vector corresponding to a difference between the second spectral vector and the spectral measurement vector via a least squares method, the reduction design matrix having columns each representing the reference spectral data of the reduction model; calculating a first residual sum of squares of errors RSS1 corresponding to said first weight vector having a number p1 of non-negative components; calculating a second residual sum of squares of errors RSS2 corresponding to said second weight vector having a number p2 of non-negative components; calculating said F-measure as a ratio F=((RSS2-RSS1) / (p1-p2)) / (RSS1 / (N-p1)) of a difference between said second sum of squared residuals RSS2 and said first sum of squared residuals RSS1 divided by a difference between said numbers p2 and p1 and said first sum of squared residuals RSS1 divided by a difference between a number N of components of said spectral measurement vector and said number p1; The method of claim 4 , comprising:

6. The step of determining a non-negative component of each of the first weight vector and the second weight vector comprises: expressing the first weighting vector that minimizes the first residual vector as a product of a pseudo-inverse of the overall design matrix and the spectral measurement vector, and expressing the second weighting vector that minimizes the second residual vector as a product of a pseudo-inverse of the reduced design matrix and the spectral measurement vector; When a component of the first weight vector or the second weight vector has a negative value, the component is a negative component. modifying the global design matrix or the reduced design matrix, respectively, by removing from the matrix spectral vectors corresponding to the negative components; setting the negative components to zero; recalculating the pseudo-inverse of the modified overall design matrix or the modified reduced design matrix, respectively, until the resulting components of the first weight vector and the second weight vector have only non-negative values; The method of claim 5 , comprising:

7. 1. A system for authenticating a marking deposited on a substrate and having a composition including a first material including a SERS taggant or a SERRS taggant, the system comprising: a light source, a Raman spectrometer, an imaging unit, and a control unit having a processing unit and a memory, the light source being controlled by the control unit via a current loop to deliver a calibrated excitation light, the system comprising: illuminating the marking with the excitation light delivered by the light source controlled by the control unit; collecting the resulting Raman light from the marking, dispersing the collected Raman light in the Raman spectrometer having a plurality of spectral channels, acquiring a two-dimensional digital image of corresponding spectral data by the imaging unit, and storing the acquired spectral data in the memory as a measured Raman spectrum of the marking. [0023] The present invention is configured to The system comprises: the memory stores an overall model of a Raman spectrum of a genuine marking deposited on a genuine substrate and having a composition including a genuine SERS taggant or a genuine first material including a genuine SERRS taggant as a first weighted sum of a reference Raman spectrum of the genuine taggant, a reference Raman spectrum of a reference genuine substrate not marked with the genuine taggant, and a reference Raman spectrum of a reference genuine first material not including the genuine taggant, the reference Raman spectra being collected in response to the genuine taggant, the reference genuine substrate, and the reference genuine first material being each illuminated with excitation light; the memory stores a reduction model of the Raman spectrum of the reduced marking, which differs from the authentic marking only in its composition not including the authentic taggant, as a second weighted sum of the reference Raman spectrum of the reference authentic substrate and the reference Raman spectrum of the reference authentic first material; The system, via the processing unit, fitting the measured Raman spectrum stored in the memory to the stored global model of the Raman spectrum by calculating weight values ​​in the global model that minimize a difference between the global model and the measured Raman spectrum under a non-negativity constraint on the weights, to obtain a corresponding first residual, which is stored in the memory; fitting the measured Raman spectrum stored in the memory to the stored reduced model of the Raman spectrum by calculating values ​​of weights in the reduced model that minimize a difference between the reduced model and the measured Raman spectrum under a non-negativity constraint on the weights, to obtain a corresponding second residual, which is stored in the memory; calculating an F-value corresponding to an F-test of a comparison between the full model and the reduced model of the measured Raman spectrum from the stored first residual and the stored second residual, and storing the F-value in the memory; determining whether the taggant is present in the marking based on the stored F-value and transmitting a signal indicative of the result of said determination; The system is further configured to:

8. 8. The system of claim 7, wherein during the operation of measuring the Raman light signal scattered by the marking, the marking is moving relative to the Raman spectrometer, and the control unit synchronizes the illumination of the marking with the light source and synchronizes the acquisition of the measured Raman spectrum via the Raman spectrometer and the imaging unit with the movement of the marking.

9. 9. The system of claim 7 or 8, wherein the composition of the marking includes a second material, and the weighted sum of each of the overall model and the reduced model further includes a reference spectrum of a corresponding authentic second material having a corresponding weight, the reference spectrum being collected upon illumination of the authentic second material by the excitation light and stored in the memory.

10. The processing unit processes the stored two-dimensional digital image. converting the two-dimensional spectral data to one-dimensional spectral data via line binning and converting the binned data to wavelength data; resampling the one-dimensional spectral data to obtain a one-dimensional spectrum with data points equidistant in wavelength; calibrating the one-dimensional spectrum against a reference white light spectrum stored in the memory to obtain a calibrated spectrum; filtering the calibrated spectrum with a low pass filter to obtain a filtered spectrum; aligning the filtered spectrum in wavelength with a reference spectrum for the taggant stored in the memory, thereby obtaining a pre-processed spectrum, which is stored in the memory; and performing an operation of calculating the first residual and the second residual by using the preprocessed spectrum stored in the memory as the measured Raman spectrum; The system according to any one of claims 7 to 9, configured to perform an operation of pre-processing by:

11. The processing unit includes: defining a spectral measurement vector as a vector corresponding to the obtained pre-processed spectrum; defining a first spectral vector as the product of a first weight vector and an overall design matrix, and determining non-negative components of each of the first weight vectors that minimize via a least squares method a first residual vector corresponding to a difference between the first spectral vector and the spectral measurement vector, the overall design matrix having columns each representing reference spectral data for the overall model; defining a second spectral vector as a product of a second weight vector and a reduction design matrix, and determining non-negative components of each of the second weight vectors that minimize a second residual vector corresponding to a difference between the second spectral vector and the spectral measurement vector via a least squares method, the reduction design matrix each having columns representing the reference spectral data of the reduction model; Calculating a first residual sum of squares of an error corresponding to the first weight vector having a number p1 of non-negative components, and storing the calculated first residual sum of squares RSS1 and the number p1 in the memory; Calculating a second residual sum of squares RSS2 of the error corresponding to the second weight vector having a number p2 of non-negative components, and storing the calculated second residual sum of squares RSS2 and the number p2 in the memory; calculating the F-measure as a ratio F=((RSS2-RSS1) / (p1-p2)) / (RSS1 / (N-p1)) of a difference between the stored second sum of squared residuals RSS2 and the stored first sum of squared residuals RSS1 divided by the difference between the stored numbers p2 and p1 to a difference between the number N of components of the spectral measurement vector and the number p1; The system of claim 10 , further configured to:

12. The processing unit converts each of the non-negative components of the first weight vector and the second weight vector into expressing the first weighting vector that minimizes the first residual vector as a product of the pseudo-inverse of the overall design matrix and the spectral measurement vector; expressing the second weighting vector that minimizes the second residual vector as a product of a pseudo-inverse of the reduced design matrix and the spectral measurement vector; When a component of the first weight vector or the second weight vector has a negative value, the component is a negative component. modifying the global design matrix or the reduced design matrix, respectively, by removing from the matrix spectral vectors corresponding to the negative components; setting said negative components to zero; recalculating the pseudo-inverse of the modified overall design matrix or the modified reduced design matrix until the resulting components of the first weight vector and the second weight vector have only non-negative values, respectively, and storing the resulting components in the memory; The system of claim 11 , further configured to determine by:

Images