CN108254351A - For the Raman spectra detection process of article examination - Google Patents
For the Raman spectra detection process of article examination Download PDFInfo
- Publication number
- CN108254351A CN108254351A CN201611257547.5A CN201611257547A CN108254351A CN 108254351 A CN108254351 A CN 108254351A CN 201611257547 A CN201611257547 A CN 201611257547A CN 108254351 A CN108254351 A CN 108254351A
- Authority
- CN
- China
- Prior art keywords
- article
- raman spectrum
- checked
- raman
- standard
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/62—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light
- G01N21/63—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light optically excited
- G01N21/65—Raman scattering
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2201/00—Features of devices classified in G01N21/00
- G01N2201/10—Scanning
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Abstract
The invention discloses a kind of Raman spectra detection process for article examination, include the following steps：Raman spectrum acquisition step：Acquire the Raman spectrum of article to be checked；With comparison and determination step：The Raman spectrum of standard article stored in the Raman spectrum and standard database of the article to be checked of acquisition is compared, to judge whether article to be checked matches with standard article.The comparison and determination step include：Classified using support vector machines to the Raman spectrum of article to be checked, to realize the comparison of the Raman spectrum of the standard article stored in the Raman spectrum of article to be checked and standard database.
Description
Technical field
The present invention relates to Raman spectrum detection fields, particularly, are related to the Raman spectra detection process checked for article,
The Raman spectra detection process that more particularly, to entry and exit special article is quickly checked.
Background technology
Raman spectrum is a kind of molecular vibration spectrum, it can reflect the fingerprint characteristic of molecule, available for the inspection to substance
It surveys.Raman spectrum detection by detect determinand for Raman spectrum caused by the Raman scattering effect of exciting light detecting and
Identify substance.Raman spectra detection process have been widely used for liquid safety check, jewelry detection, explosive detection, illicit drugs inspection,
The fields such as drug detection, Detecting Pesticide.
The existing Site Detection for entry and exit special article, the main real time monitoring using live high definition camera device
With supervisory role, view of the scene personnel unpack detection, compare goods information and whether declaration form information is consistent, if goods information one
It causes, then it is assumed that cargo can let pass.
However, the above method has the following disadvantages：(1) view of the scene person works amount is big.It unpacks and checks examination and approval document
Whether view of the scene staff is unanimously needed to check label information one by one with label information, it is artificial to compare complicated cargo product
Whether the much informations such as name, specification, character and quantity (most of for the non-Chinese language information such as English) are consistent.(2) examination leakage
Hole is apparent.The examination mode of verification label information can not verify whether label information is consistent with actual content, Wu Fayou
Effect identification is using low-risk or no special article packing and storing or the situation of transport high risk special article.
Invention content
In terms of solving at least one of disadvantages mentioned above, it is proposed that the present invention.
The object of the present invention is to provide a kind of Raman spectra detection process for article examination, more particularly, to come in and go out
The Raman spectra detection process that border special article is quickly checked, can enhance the accuracy of the article view of the scene, and can be with
The view of the scene efficiency of article is accelerated in matching examination while realizing multiple information by one-time detection.
In order to achieve the above-mentioned object of the invention, technical scheme of the present invention is accomplished by the following way：
According to an aspect of the present invention, a kind of Raman spectra detection process for article examination is provided, including following
Step：
Raman spectrum acquisition step：Acquire the Raman spectrum of article to be checked；With
Comparison and determination step：The standard article that will be stored in the Raman spectrum and standard database of the article to be checked of acquisition
Raman spectrum compare, to judge whether article to be checked matches with standard article,
It is characterized in that, the comparison and determination step include：Using support vector machines to the Raman spectrum of article to be checked
Classify, to realize the ratio of the Raman spectrum of the standard article stored in the Raman spectrum of article to be checked and standard database
It is right.
According to some embodiments, described the step of being classified using support vector machines to the Raman spectrum of article to be checked, is wrapped
It includes：
Establish grader step：Training sample is chosen, the Raman light to obtain training sample is measured to training sample
Spectrum carries out sparse transformation to the Raman spectrum of training sample, then establishes grader using algorithm of support vector machine；With
Classifying step：Using the grader of foundation, classify to the Raman spectrum of article to be checked.
According to some embodiments, the sparse transformation includes the following steps：
Obtain the principal component M of the Raman spectral information y of training sample；
Rarefaction representation is carried out to the Raman spectral information y of training sample using formula y=Mc, obtains a series of sparse table
The vector C showni；With
The target vector c that rarefaction representation is carried out to Raman spectral information y is calculated using following objective functions：
Wherein, the nonzero element number in vector C is fixed as k, and N is the length of vector C, and k≤N, i are the serial number of vector C.
According to some embodiments, the comparison and determination step further include：
Similitude is carried out to the Raman spectrum of standard article stored in the Raman spectrum and standard database of article to be checked
Measurement；
When the result of similarity measurement is more than preset threshold value, judge that article to be measured is matched with standard article；
When the result of similarity measurement is equal to or less than preset threshold value, using drawing of the support vector machines to article to be checked
Graceful spectrum is classified, to realize the Raman spectrum of the standard article stored in the Raman spectrum of article to be checked and standard database
Comparison.
According to some embodiments, the standard article that is stored in the Raman spectrum and standard database to article to be checked
Raman spectrum carry out similarity measurement include：
Calculate the Raman spectrum of article to be checked feature vector and standard database in the Raman light of standard article that stores
The related coefficient of the feature vector of spectrum, using the related coefficient calculated as the result of similarity measurement.
According to some embodiments, the Raman spectra detection process further includes following steps：
Establish standard database step：The successively surfactant package of acquisition standard article and the Raman spectrum of actual content,
To form the normal data of the Raman spectrum of the ingredient of the Raman spectrum of the surface composition comprising standard article and actual content
Library.
According to some embodiments, the Raman spectrum acquisition step includes：Surface and inner scanning are carried out to article to be checked,
To acquire the Raman spectrum of the ingredient of the Raman spectrum of the surface composition of article to be checked and actual content；
The Raman spectrum to article to be checked carries out phase with the Raman spectrum of standard article stored in standard database
The step of being measured like property includes：Standard to being stored in the Raman spectrum and standard database of the surface composition of article to be checked respectively
The Raman spectrum and standard database of the ingredient of the actual content of the Raman spectrum of the surface composition of article and article to be checked
The Raman spectrum of the ingredient of the actual content of the standard article of middle storage carries out similarity measurement；And
The result of the similarity measurement is all higher than pre- more than the preset threshold value only result including two similarity measurements
If threshold value situation.
According to some embodiments, the standard database step of establishing further includes：The name of an article, the source for acquiring standard article are public
Department, specification size and pictorial information, and these information are stored in standard database.
According to another aspect of the present invention, a kind of electronic equipment is also provided, including：
Memory, for storing executable instruction；And
Processor, it is any in above-mentioned aspect or embodiment to perform for performing the executable instruction stored in memory
Method described in.
Detection method according to embodiments of the present invention eliminates the multinomial letter that view of the scene personnel compare article to be checked one by one
The process of breath accelerates view of the scene speed；Simultaneously using the matched recognition methods of characteristic information, the standard of the view of the scene is improved
Exactness.
Description of the drawings
Fig. 1 shows the flow chart of the detection method for special article examination of entering and leaving the border according to embodiments of the present invention；
Fig. 2 is the schematic diagram of support vector machines principle；
Fig. 3 shows the flow chart for establishing support vector machine classifier；
Fig. 4 shows the flow chart using the detection method for improving support vector machines according to embodiments of the present invention；With
Fig. 5 shows what is arranged for performing the exemplary hardware of the electronic equipment of detection method according to embodiments of the present invention
Block diagram.
Specific embodiment
Below with reference to the embodiments and with reference to the accompanying drawing the technical solutions of the present invention will be further described.Illustrating
In book, the same or similar drawing reference numeral represents the same or similar component.Following reference attached drawings are to embodiment of the present invention
Illustrate to be intended to explain the present general inventive concept of the present invention, and be not construed as a kind of limitation to the present invention.
Herein, for convenience, statements such as " first, second ", " A, B, C " or " S1/S10, S2/S20 " are used
The step of description method, still, unless otherwise specified, such statement should not be construed as the limitation to step execution sequence.
Hereinafter, mainly the technical concept of the present invention is described in detail by taking special article of entering and leaving the border as an example, generally
Ground, entry and exit special article refer to microorganism, tissue, mankind's inheritance resources, biological products, blood, blood product etc..Due to
The live special article type of inspection and quarantine examination is various, the source of goods is various, packaging is complicated, while many special articles have sample size
Less, it is visually difficult to observe, the characteristics of ingredient is unknown；In addition, the circulation of special article is compared to regular general cargo, transport difficulty and
Potential risk bigger；For the view of the scene, many special articles such as biological agents such as microorganism, blood can not realize that examination is existing
The Kaifeng detection of field.The detection that these reasons have codetermined entry and exit special article becomes inspection and quarantine examination working difficult point.
And use the Raman spectra detection process for article examination according to embodiments of the present invention, can to entry and exit special article into
The quick examination of row, while the accuracy of the view of the scene can also be improved.
According to an embodiment of the invention, it is necessary first to establish the whole Raman of a more complete entry and exit special article
Spectrogram information library, the information bank are properly termed as standard database.Then, at the scene in ping procedure, show according on cargo list
Item name, selection is present in standard article in standard database, carries out surface and inner scanning to article to be checked later,
Then the standard spectrogram in the spectrogram and standard database of the article to be checked that are obtained to scanning is compared, with reference to " similar
Property measurement " algorithm and " improve support vector machines " algorithm, complete to identify the swift nature for special article of entering and leaving the border.
It, can be first by the standard of representative entry and exit special article when establishing entry and exit special article standard database
Spectrogram is set up, and can also be accumulated addition in use or be provided by inspection and quarantine examination unit.In one example,
For each entry and exit special article, in the case of no other objects interference, can be seen that in the special article
Its surfactant package and the Raman signatures spectrogram of actual content successively are acquired at tolerant site, then arranges and establishes Raman spectrogram
Information bank, to form the standard database.
According to one embodiment of present invention, in standard database it is in store entry and exit special article Global Information.One
Aspect when naming the standard article in standard database, indicates the name of an article, source company, specification size, the figure of standard article
The information such as piece；On the other hand, each entry and exit special article has corresponding two spectrograms, and a spectrogram is to its certain bits
Point surface carry out spectrogram acquisition, it is collected be surface composition spectrogram information, the other is the inside to same site
Carry out spectrogram acquisition, it is the spectrogram information of actual constituent inside it that institute is collected.It, need to be by table when item scan to be checked compares
The collected spectrogram information in face is compared with the surface standard spectrum in spectrogram library, by collected internal spectrogram information and spectrogram
Internal standard spectrum in library is compared, and just thinks that the article to be checked can belong to same object with standard article more than threshold value twice
Product, increase matching accuracy.
Fig. 1 shows the flow chart of the detection method for special article examination of entering and leaving the border according to embodiments of the present invention,
The detection method is described in detail with reference to Fig. 1, according to embodiments of the present invention checks for special article of entering and leaving the border
Detection method mainly include the following steps：
Raman spectrum acquisition step：Article to be checked is selected on detection device, scans article to be checked to acquire article to be checked
Raman spectrum；
Raman spectrum compares and determination step：Using recognizer according to embodiments of the present invention, by the object to be checked of acquisition
The Raman spectrum of product is compared with the standard Raman spectroscopy of article indicated in the cargo list stored in above-mentioned standard database,
Whether matched with both judgements；Wherein, the recognizer of the embodiment of the present invention will be explained below；
In the comparison of above-mentioned Raman spectrum and determination step, if it is decided that article to be checked and the article phase indicated in cargo list
Matching, then data can be preserved, and terminate this examination；If it is determined that the article indicated in article to be checked and cargo list
It mismatches, then need to continue to execute following step；
Chemical substance identifies and determination step：Using conventional Raman spectrum recognition methods, the reality of article to be checked is detected
Ingredient, to identify chemical substance that article to be checked includes, then, the chemical substance that will identify that in above-mentioned standard database
Whether the chemical substance of article indicated in the cargo list of storage is compared, matched with both judgements；
In the identification of above-mentioned chemical substance and determination step, in the situation for the chemical substance that can recognize that article to be checked
Under, if it is decided that article to be checked is consistent with the information of article indicated in cargo list, then can preserve data, terminate this
Examination；If it is determined that article to be checked and the information of article indicated in cargo list are inconsistent, then need to carry out subsequent processing, example
Such as artificial examination operation of unpacking；In the case of the chemical substance for being unable to identify that article to be checked, need again to be checked
Article carries out sample examination, then preserves the data of acquisition, to carry out subsequent processing.
Raman spectrum in the above-mentioned methods is compared in determination step, needing using identification according to embodiments of the present invention
Algorithm, the mark of article that will be indicated in the cargo list stored in the Raman spectrum of the article to be checked of acquisition and above-mentioned standard database
Whether quasi- Raman spectrum is compared, matched with both judgements, is described in detail below in conjunction with attached drawing according to embodiments of the present invention
Recognizer.
According to one embodiment of present invention, using Raman spectroscopy combination similarity measurements quantity algorithm and support vector machines
Innovatory algorithm first, similarity measurement is carried out by calculating related coefficient to article to be checked and standard article；Secondly, when to be checked
When the related coefficient of article and standard article is less than threshold value, improved algorithm of support vector machine is pointedly proposed, to higher-dimension
Vector carries out dimension-reduction treatment, and classifies to be checked article of the related coefficient less than threshold value, to avoid Raman spectroscopy equipment
Influence of the noise of generation to recognition result, and promote recognition speed and accuracy.Before specific method is described, to this
Algorithm model used in method illustrates.
[algorithm model]
(1) similarity measurement
Raman spectrum identification technology is the application technology that article to be checked is classified and identified, one according to the present invention
Embodiment after completing Pretreated spectra and feature extraction, obtains the key message that can reflect material composition, extracts spectral signal
Included in spectral information, treat detection article according to spectral information difference and classified and identified, wherein, using similitude
Measure calculates spectral information difference.
It should be noted that Pretreated spectra generally comprises denoising, baseline correction, normalized etc., herein, adopt
The original spectrum that collection obtains is generally required by pretreatment, for sake of simplicity, hereafter no longer repeating one by one.
Related coefficient (Correlation coefficient) is the amount of linearly related degree between research variable, is a kind of
The method of correlation between measurement vector.Equipped with feature vector, X (x1, x2..., xn), Y (y1, y2..., yn), the correlation of the two
Coefficient is defined as follows：
Wherein, n represents feature vector, X, the length of Y, The mean value of vector X, Y are represented respectively, and i represents the i-th of vector
A data.
In the present embodiment, basis for estimation of the related coefficient as similarity measurement is selected, it can be to avoid Euclidean distance pair
The loss of information and mahalanobis distance are to the amplification of little deviation, so as to preferably judge between two feature vectors
Similitude.
(2) support vector machines (SVM)
In raman spectroscopy measurement, since there are sample homogeneity difference, noise of instrument, fluorescence backgrounds etc. so that Raman
Spectrum generates deviation；During spectral manipulation, denoising, baseline correction etc. can also generate error.In identification process only with
The accuracy rate that related coefficient carries out the feature recognition of substance is not high, therefore, in the special object of entry and exit for inspection and quarantine field
During the Raman spectrums of product carries out feature recognition, introduce support vector machines and be equal to and be less than or the slightly below object to be checked of threshold value
Product carry out taxonomy of goods.
Support vector machines is two disaggregated models, and basic model is defined as being spaced maximum linear point on feature space
Class device.Its principle is as shown in Figure 2.
If training sample set X is xi, i=1,2 ..., N, sample adheres to two classes, w separately1And w2, and linear separability.Linear discriminant
The general type of function l is：G (x)=ω x+b, but such hyperplane is not unique.Therefore, point of support vector machines
Class identification problem is converted into the Optimal Separating Hyperplane problem for finding largest interval.
Select suitable ω, b so that sample class w1And w2Meet g (x)=1 and g (x)=- 1 respectively, i.e., such as following formula：
Geometry meaning of the support vector machines in linear classifier, is exactly in hyperplane l1And l2Between distance Gap obtain most
Big value.In this process, adjust the distance hyperplane l0Nearest sample is normalized, such Gap/2=2/ | | ω | |, then
It is linear can the majorized functions of category support vector machines be shown below：
s.t.yi(ω·xi+ b) >=1, i=1,2 ..., n
It can be obtained down according to Lagrangian (Lagrange) multiplier method and Karush-Kuhn-Tucker conditions (KKT conditions)
Formula, wherein λ are Lagrange multipliers.
λi>=0, i=1,2 ..., N
λi[yi(ω x+b) -1]=1, i=1,2 ..., N
(3) support vector machines is improved
Due to Raman spectrum after pretreatment, obtaining high dimension vector to characterize substance to be identified, directly using
During classification is identified in high dimension vector, training learning time process is longer.
In order to change this problem, during being identified using support vector machines, firstly for Raman spectrum into
The certain sparse transformation of row, concrete operation include two steps：First, the principal component M of the Raman spectral information y of training sample is obtained；2nd,
Rarefaction representation is carried out to the Raman spectral information y of training sample using formula y=Mc, obtains the vector of a series of rarefaction representation
ci, and the nonzero element number in vector C is fixed as k, k≤N；And it is calculated using following objective functions and Raman spectrum is believed
Cease the target vector c that y carries out rarefaction representation：
Wherein, N is the length of vector C, and i is the serial number of vector C.
By using above-mentioned object function, find and causeThe target vector c being minimized, to use drop
Target vector c after dimension represents spectral information y, so as to fulfill the rarefaction representation of Raman spectrum.
During unmodified supporting vector is selected to carry out material classification, higher-dimension and big portion due to spectral information
The non-zero of point information, needs to be mapped to more higher-dimension using RBF kernel functions, Polynomial kernel function etc. and carries out classification processing；It improves
Support vector machines afterwards realizes the ga s safety degree for spectrum sparse expression, enhancing spectral information, and line can be used in spectral classification
Property kernel support vectors machines realize, make training and test speed faster, and required memory space is less, in training learning process
Reduce the time.
The process of establishing of grader mainly includes three steps, and by taking two classification as an example, specific Establishing process is as shown in Figure 3.
Appropriate sample is chosen as training sample, sample is measured to obtain Raman spectrum, by baseline correction, gone
It makes an uproar, normalize etc. and obtaining test sample data after pretreatments.Test sample is divided into positive sample and negative sample, wherein positive sample is
For the spectral information of certain article to be measured, negative sample is the spectral information of non-article to be measured.By carrying out rarefaction representation to spectrum,
Model foundation is carried out using support vector machines, obtains SVM classifier.
[method]
Based on computation model presented hereinbefore, with reference to Fig. 4, further describe according to embodiments of the present invention for object
The Raman spectra detection process of product examination, this method may include steps of：
Raman spectrum acquisition step：Acquire the Raman spectrum of article to be checked；With
Comparison and determination step：The standard article that will be stored in the Raman spectrum and standard database of the article to be checked of acquisition
Raman spectrum compare, to judge whether article to be checked matches with standard article.
According to one embodiment, above-mentioned comparison and determination step specifically include：Calculate the spy of the Raman spectrum of article to be checked
The related coefficient of the feature vector of the Raman spectrum of standard article stored in sign vector and standard database, when the phase calculated
When relationship number is more than preset threshold value, judge that article to be checked is matched with standard article.
Further, the comparison and determination step further include：When the related coefficient calculated is equal to or less than preset
During threshold value, classified using support vector machines to the Raman spectrum of article to be checked, with judge article to be checked whether with reference substance
Product match.
Specifically, described the step of being classified using support vector machines to the Raman spectrum of article to be checked, is included：
Establish grader step：Training sample is chosen, the Raman light to obtain training sample is measured to training sample
Spectrum carries out sparse transformation to the Raman spectrum of training sample, then establishes grader using algorithm of support vector machine；With
Classifying step：Using the grader of foundation, classify to the Raman spectrum of article to be checked.
During sparse transformation is carried out to the Raman spectrum of training sample, it may be used what above-mentioned formula (6) defined
Object function carries out sparse variation to the Raman spectrum of training sample, to obtain the rarefaction representation of Raman spectrum, to reduce training
Learning time, so that training and test speed are faster.
[experimental verification]
In the following, experimental verification is carried out to the Raman spectra detection process for article examination according to embodiments of the present invention.
(1) experiment condition and method
Experiment randomly select 380 kinds in the daily ping procedure of Entry-Exit Inspection and Quarantine Bureau of Beijing immigration special articles into
Row Raman spectrogram acquires and algorithm of support vector machine confirmatory experiment.Table 1 represents its type statistical form.
1 380 kinds of entry and exit special article substance classes statistical forms of table
Experiment instrument be Tongfangweishi Technology Co., Ltd's RT6000 handheld Raman spectrometers, excitation wavelength
For 785nm；Resolution ratio is 6~9cm-1；Wave-number range is 200~3200cm-1。
The collected Raman spectrogram of experiment institute carries out noise reduction, Baseline wander and normalization pretreatment operation, all
Spectrum analysis is obtained with contrast verification by Matlab (R2010a) and C++ programmed methods.
The Accuracy Verification for improving support vector machines includes two aspects, for the recognition result substance identical with actual value
It is judged as correct (pass), the substance different from actual value is determined as wrong (fail).Using test sample to the accurate of model
Property verified, the accuracy of decision algorithm model.Idiographic flow is as shown in Figure 4.
In experiment, the pretreatments such as baseline correction, denoising, normalization, obtained Raman are carried out to the Raman spectrum acquired
Spectrum carries out similarity system design with the sample in standard spectrum library, if related coefficient is more than threshold value, is judged to matching；To being less than
Classified in the Raman spectrum of threshold value using improved support vector machines, if classification results represent the two matching, as matched,
Conversely, then to mismatch.
(2) experimental result
All test samples are analyzed with improved support vector machines, wherein 6 class substances are shared, including blood
Product, virus, antibody etc..The ranging from 0.7460-0.7822 of the discrimination of all kinds of standard of physical databases, the average area between class
Index is 0.7623.To the accordance of substance is carried out with improving algorithm of support vector machine using similarity measurement, support vector machines
The matching rate of examination is counted, as shown in table 2.
2 test result list of table
In accordance ping procedure, the correct matching rate using three kinds of methods is as shown in table 2.It is in selected threshold successively
0.85th, it 0.86,0.87,0.88,0.89,0.90 carries out in ping procedure, comparison is understood：When choosing relatively low threshold value, use
Related coefficient also can be preferably checked as a result, as threshold value further increases, and matching rate is gradually reduced；Using supporting vector
During machine is checked with improvement algorithm of support vector machine, the matching rate of article to be checked is above directly using similarity measurements
Amount.Therefore, support vector cassification decision process is increased for the relatively low article to be checked of similitude, the possibility of missing inspection can be reduced
Property, simultaneously as the sparse expression for improving algorithm of support vector machine improves the accuracy of grader so that improve supporting vector
Machine algorithm has obtained optimal result in ping procedure.
According to another embodiment of the present invention, a kind of Raman spectra detection process for article examination, the party are also provided
Method can be without similarity measurement, and directly carries out above-mentioned comparison and determination step using improvement algorithm of support vector machine.
Specifically, this method may include steps of：
Raman spectrum acquisition step：Acquire the Raman spectrum of article to be checked；With
Comparison and determination step：Directly classified using support vector machines to the Raman spectrum of article to be checked, with judgement
Whether article to be checked matches with standard article.
Specifically, described the step of being classified using support vector machines to the Raman spectrum of article to be checked, is included：
Establish grader step：Training sample is chosen, the Raman light to obtain training sample is measured to training sample
Spectrum carries out sparse transformation to the Raman spectrum of training sample, then establishes grader using algorithm of support vector machine；With
Classifying step：Using the grader of foundation, classify to the Raman spectrum of article to be checked.
Similarly, during sparse transformation is carried out to the Raman spectrum of training sample, above-mentioned formula (6) may be used
The object function of definition carries out sparse variation to the Raman spectrum of training sample, to obtain the rarefaction representation of Raman spectrum, to subtract
Learning time is trained less.
That is, in this embodiment, using improvement algorithm of support vector machine to the Raman spectrum of all articles to be measured
Classification and Identification is carried out, and does not differentiate between its related coefficient and is greater than or less than or equal to threshold value.It should be understood that in addition to the difference it
Outside, the present embodiment can include other steps same as the previously described embodiments, such as including various pre-treatment steps etc..
According to still another embodiment of the invention, a kind of electronic equipment is also provided, Fig. 5 is the example for showing the electronic equipment
The block diagram of hardware layout 500.Hardware layout 500 includes processor 506 (for example, microprocessor (μ P), digital signal processor
(DSP) etc.).Processor 506 can be performed for the different actions of method steps described herein single treatment units or
Person is multiple processing units.Arrangement 500 can also include input unit 502, the Yi Jiyong for receiving signal from other entities
In the output unit 504 that signal is provided to other entities.Input unit 502 and output unit 504 can be arranged to single reality
The entity that body either detaches.
In addition, arrangement 500 can include having non-volatile or form of volatile memory at least one readable storage
Medium 508, e.g. electrically erasable programmable read-only memory (EEPROM), flash memory, and/or hard disk drive.Readable storage
Medium 508 includes computer program 510, which includes code/computer-readable instruction, by arrangement 500
In processor 506 perform when hardware layout 500 and/or the equipment including hardware layout 500 are performed for example
Above in conjunction with the described flow of Fig. 1,3-4 and its any deformation.
Computer program 510 can be configured with the computer journey of such as computer program module 510A~510C frameworks
Sequence code.Therefore, it in example embodiment when hardware layout 500 is used in such as equipment, arranges in 500 computer program
Code include：Module 510A, is used for ....Code in computer program further includes：Module 510B, is used for ....Computer journey
Code in sequence further includes：Module 510C, is used for ....
Computer program module can substantially perform each action in the flow shown in Fig. 1,3-4, with simulation
Equipment.In other words, when performing different computer program modules in processor 506, they can correspond to upper in equipment
State different units.
Although being implemented as computer program module above in conjunction with the code means in Fig. 5 the disclosed embodiments,
Hardware layout 500 is performed above in conjunction with the described action in Fig. 1~4 when being performed in processor 506, however alternatively implementing
In example, at least one in the code means can at least be implemented partly as hardware circuit.
Processor can be single cpu (central processing unit), but can also include two or more processing units.Example
Such as, processor can include general purpose microprocessor, instruction set processor and/or related chip group and/or special microprocessor (example
Such as, application-specific integrated circuit (ASIC)).Processor can also include the onboard storage device for caching purposes.Computer program can
To be carried by the computer program product for being connected to processor.Computer program product can include being stored thereon with computer
The computer-readable medium of program.For example, computer program product can be flash memory, random access memory (RAM), read-only deposit
Reservoir (ROM), EEPROM, and above computer program module can use the form quilt of the memory in UE in an alternative embodiment
It is distributed in different computer program products.
It will be appreciated by those skilled in the art that in an embodiment of the present invention, although the examination with special article of entering and leaving the border
The technical concept of the present invention is described in detail for example, but the present invention is not limited to examination entry and exit special article, it can be with
It is quickly detected suitable for such as drug, the fields such as biological sample screening.
Although with reference to attached drawing, the present invention is described, and the embodiment disclosed in attached drawing is intended to preferred to the present invention
Embodiment illustrates, and it is not intended that a kind of limitation of the invention.
Although some embodiments of present general inventive concept have been shown and have illustrated, those of ordinary skill in the art will manage
Solution in the case of without departing substantially from the principle of this present general inventive concept and spirit, can make a change these embodiments, of the invention
Range is limited with claim and their equivalent.
Claims (9)
1. a kind of Raman spectra detection process for article examination includes the following steps：
Raman spectrum acquisition step：Acquire the Raman spectrum of article to be checked；With
Comparison and determination step：The drawing of standard article that will be stored in the Raman spectrum and standard database of the article to be checked of acquisition
Graceful spectrum compares, to judge whether article to be checked matches with standard article,
It is characterized in that, the comparison and determination step include：The Raman spectrum of article to be checked is carried out using support vector machines
Classification, to realize the comparison of the Raman spectrum of the standard article stored in the Raman spectrum of article to be checked and standard database.
2. Raman spectra detection process according to claim 1, which is characterized in that described to use support vector machines to be checked
The step of Raman spectrum of article is classified includes：
Establish grader step：Training sample is chosen, the Raman spectrum to obtain training sample is measured to training sample, it is right
The Raman spectrum of training sample carries out sparse transformation, then establishes grader using algorithm of support vector machine；With
Classifying step：Using the grader of foundation, classify to the Raman spectrum of article to be checked.
3. Raman spectra detection process according to claim 2, which is characterized in that the sparse transformation includes following step
Suddenly：
Obtain the principal component M of the Raman spectral information y of training sample；
Rarefaction representation is carried out to the Raman spectral information y of training sample using formula y=Mc, obtains a series of rarefaction representation
Vector Ci；With
The target vector c that rarefaction representation is carried out to Raman spectral information y is calculated using following objective functions：
Wherein, the nonzero element number in vector C is fixed as k, and N is the length of vector C, and k≤N, i are the serial number of vector C.
4. Raman spectra detection process according to any one of claim 1-3, which is characterized in that the comparison and judgement
Step further includes：
Similarity measurement is carried out to the Raman spectrum of standard article stored in the Raman spectrum and standard database of article to be checked；
When the result of similarity measurement is more than preset threshold value, judge that article to be measured is matched with standard article；
When the result of similarity measurement is equal to or less than preset threshold value, using support vector machines to the Raman light of article to be checked
Spectrum is classified, to realize the ratio of the Raman spectrum of the standard article stored in the Raman spectrum of article to be checked and standard database
It is right.
5. Raman spectra detection process according to claim 4, which is characterized in that the Raman light to article to be checked
It composes and includes with the Raman spectrum of the standard article progress similarity measurement stored in standard database：
Calculate the Raman spectrum of article to be checked feature vector and standard database in the Raman spectrum of standard article that stores
The related coefficient of feature vector, using the related coefficient calculated as the result of similarity measurement.
6. Raman spectra detection process according to any one of claims 1-5, which is characterized in that further include following step
Suddenly：
Establish standard database step：The successively surfactant package of acquisition standard article and the Raman spectrum of actual content, with shape
Into the standard database of the Raman spectrum of the ingredient of the Raman spectrum and actual content of the surface composition comprising standard article.
7. Raman spectra detection process according to claim 6, which is characterized in that
The Raman spectrum acquisition step includes：Surface and inner scanning are carried out to article to be checked, to acquire the table of article to be checked
The Raman spectrum of the Raman spectrum of face ingredient and the ingredient of actual content；
The Raman spectrum of the standard article stored in the Raman spectrum and standard database to article to be checked carries out similitude
The step of measurement, includes：Standard article to being stored in the Raman spectrum and standard database of the surface composition of article to be checked respectively
Surface composition Raman spectrum and article to be checked actual content ingredient Raman spectrum and standard database in deposit
The Raman spectrum of the ingredient of the actual content of the standard article of storage carries out similarity measurement；And
The result of the similarity measurement is all higher than preset more than the preset threshold value only result including two similarity measurements
The situation of threshold value.
8. Raman spectra detection process according to claim 6, which is characterized in that described to establish standard database step also
Including：The name of an article, source company, specification size and the pictorial information of standard article are acquired, and these information are stored in normal data
In library.
9. a kind of electronic equipment, including：
Memory, for storing executable instruction；And
Processor, for performing the executable instruction stored in memory, to perform as described in any one of claim 1-8
Method.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611257547.5A CN108254351B (en) | 2016-12-29 | 2016-12-29 | Raman spectrum detection method for checking articles |
| PCT/CN2017/111624 WO2018121122A1 (en) | 2016-12-29 | 2017-11-17 | Raman spectroscopy detection method for checking goods, and electronic device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611257547.5A CN108254351B (en) | 2016-12-29 | 2016-12-29 | Raman spectrum detection method for checking articles |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108254351A true CN108254351A (en) | 2018-07-06 |
| CN108254351B CN108254351B (en) | 2023-08-01 |
Family
ID=62706829
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611257547.5A Active CN108254351B (en) | 2016-12-29 | 2016-12-29 | Raman spectrum detection method for checking articles |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN108254351B (en) |
| WO (1) | WO2018121122A1 (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110489586A (en) * | 2019-01-07 | 2019-11-22 | 公安部第一研究所 | A kind of matched method of sample detection database hierarchy |
| CN111861171A (en) * | 2020-07-08 | 2020-10-30 | 京源中科科技股份有限公司 | Special article monitoring system, method and device based on Internet of things |
| CN112686768A (en) * | 2020-12-30 | 2021-04-20 | 山东省食品药品检验研究院 | Quick identification system of medicine preparation enterprise material |
| CN113092407A (en) * | 2021-04-20 | 2021-07-09 | 深圳市国科光谱技术有限公司 | Method for identifying original place varieties of Wuchang rice fragrance rice based on near infrared spectrum |
| CN113324967A (en) * | 2020-09-18 | 2021-08-31 | 山东农业大学 | Method for rapidly identifying DFD beef |
| CN113656559A (en) * | 2021-10-18 | 2021-11-16 | 印象(山东)大数据有限公司 | Data analysis method and device based on metering platform and electronic equipment |
| CN113811880A (en) * | 2019-06-07 | 2021-12-17 | 巴斯夫涂料有限公司 | Device and method for forming at least one ground truth database for an object recognition system |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11428638B2 (en) * | 2018-01-12 | 2022-08-30 | The Regents Of The University Of California | Spectroscopic biological material characterization |
| CN111141719A (en) * | 2019-12-27 | 2020-05-12 | 梧州市食品药品检验所 | Rapid nondestructive identification method for anti-AIDS medicine |
| CN111426648B (en) * | 2020-03-19 | 2023-04-07 | 甘肃省交通规划勘察设计院股份有限公司 | Method and system for determining similarity of infrared spectrogram |
| CN112161965B (en) * | 2020-09-22 | 2022-08-05 | 复旦大学附属妇产科医院 | Method, device, computer equipment and storage medium for detecting traditional Chinese medicine property |
| CN112232673A (en) * | 2020-10-16 | 2021-01-15 | 拜耳医药保健有限公司 | Early warning method, device, equipment and medium for processing genetic resources |
| CN112834481B (en) * | 2020-12-31 | 2023-09-05 | 宁波海关技术中心 | Raman spectrum enhancement measurement system and measurement method |
| CN112834451B (en) * | 2021-01-12 | 2023-04-18 | 深圳网联光仪科技有限公司 | Sample identification method and device based on infrared spectrum and storage medium |
| CN113466206A (en) * | 2021-06-23 | 2021-10-01 | 上海仪电（集团）有限公司中央研究院 | Raman spectrum analysis system based on big data |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103903010A (en) * | 2014-03-28 | 2014-07-02 | 哈尔滨工程大学 | Method for classifying hyperspectral images based on sparse characteristics and same neighborhood properties |
| JP2015052551A (en) * | 2013-09-09 | 2015-03-19 | 株式会社かんでんエンジニアリング | Component-in-oil measuring method and oil-filled electric instrument |
| CN104764734A (en) * | 2015-04-30 | 2015-07-08 | 江苏易谱恒科技有限公司 | Identification method for fruit juice flavors and fragrances based on Raman spectrum and SVM algorithm |
| CN104865279A (en) * | 2014-02-26 | 2015-08-26 | 清华大学 | Customs supervision item rapid-inspection equipment and method |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1967846A1 (en) * | 2007-03-05 | 2008-09-10 | National University of Ireland Galway | En ensemble method and apparatus for classifying materials and quantifying the composition of mixtures |
| CN101403696A (en) * | 2008-10-21 | 2009-04-08 | 浙江大学 | Method for measuring gasoline olefin content based on Raman spectrum |
| CN103217411B (en) * | 2013-04-15 | 2015-03-25 | 浙江大学 | Raman-spectrum-based method for detecting content of oleic acid, linoleic acid and saturated fatty acid in edible vegetable oil |
| CN104215623B (en) * | 2013-05-31 | 2018-09-25 | 欧普图斯(苏州)光学纳米科技有限公司 | Laser Raman spectroscopy intelligence discrimination method and system towards conglomerate detection |
| CN105806824B (en) * | 2016-03-22 | 2019-03-19 | 武汉轻工大学 | Edible oil type method for quick identification based on multi-source optical spectrum Fusion Features |
-
2016
- 2016-12-29 CN CN201611257547.5A patent/CN108254351B/en active Active
-
2017
- 2017-11-17 WO PCT/CN2017/111624 patent/WO2018121122A1/en active Application Filing
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2015052551A (en) * | 2013-09-09 | 2015-03-19 | 株式会社かんでんエンジニアリング | Component-in-oil measuring method and oil-filled electric instrument |
| CN104865279A (en) * | 2014-02-26 | 2015-08-26 | 清华大学 | Customs supervision item rapid-inspection equipment and method |
| CN103903010A (en) * | 2014-03-28 | 2014-07-02 | 哈尔滨工程大学 | Method for classifying hyperspectral images based on sparse characteristics and same neighborhood properties |
| CN104764734A (en) * | 2015-04-30 | 2015-07-08 | 江苏易谱恒科技有限公司 | Identification method for fruit juice flavors and fragrances based on Raman spectrum and SVM algorithm |
Non-Patent Citations (1)
| Title |
|---|
| 刘芳: "结构化压缩感知研究进展", 《自动化学报》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110489586A (en) * | 2019-01-07 | 2019-11-22 | 公安部第一研究所 | A kind of matched method of sample detection database hierarchy |
| CN113811880A (en) * | 2019-06-07 | 2021-12-17 | 巴斯夫涂料有限公司 | Device and method for forming at least one ground truth database for an object recognition system |
| CN111861171A (en) * | 2020-07-08 | 2020-10-30 | 京源中科科技股份有限公司 | Special article monitoring system, method and device based on Internet of things |
| CN111861171B (en) * | 2020-07-08 | 2021-07-27 | 京源中科科技股份有限公司 | Special article monitoring system, method and device based on Internet of things |
| WO2022007974A1 (en) * | 2020-07-08 | 2022-01-13 | 京源中科科技股份有限公司 | Special article monitoring system, method, and apparatus based on internet of things |
| CN113324967A (en) * | 2020-09-18 | 2021-08-31 | 山东农业大学 | Method for rapidly identifying DFD beef |
| CN112686768A (en) * | 2020-12-30 | 2021-04-20 | 山东省食品药品检验研究院 | Quick identification system of medicine preparation enterprise material |
| CN113092407A (en) * | 2021-04-20 | 2021-07-09 | 深圳市国科光谱技术有限公司 | Method for identifying original place varieties of Wuchang rice fragrance rice based on near infrared spectrum |
| CN113092407B (en) * | 2021-04-20 | 2024-01-30 | 深圳华创智信科技有限公司 | Near infrared spectrum-based method for identifying varieties of original places of five-normal rice flower fragrance rice |
| CN113656559A (en) * | 2021-10-18 | 2021-11-16 | 印象(山东)大数据有限公司 | Data analysis method and device based on metering platform and electronic equipment |
| CN113656559B (en) * | 2021-10-18 | 2022-01-25 | 印象(山东)大数据有限公司 | Data analysis method and device based on metering platform and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108254351B (en) | 2023-08-01 |
| WO2018121122A1 (en) | 2018-07-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108254351A (en) | For the Raman spectra detection process of article examination | |
| Guillemain et al. | Performance of NIR handheld spectrometers for the detection of counterfeit tablets | |
| Custers et al. | ATR-FTIR spectroscopy and chemometrics: an interesting tool to discriminate and characterize counterfeit medicines | |
| Erdoğan et al. | COVID-19 detection with traditional and deep features on cough acoustic signals | |
| US20190187048A1 (en) | Spectroscopic systems and methods for the identification and quantification of pathogens | |
| Hao et al. | Fusing multiple deep models for in vivo human brain hyperspectral image classification to identify glioblastoma tumor | |
| Krakowska et al. | Chemometrics and the identification of counterfeit medicines—A review | |
| Bertinetto et al. | Automatic baseline recognition for the correction of large sets of spectra using continuous wavelet transform and iterative fitting | |
| CN107818298A (en) | General Raman spectral characteristics extracting method for machine learning material recognition | |
| Benchea et al. | Combining restricted boltzmann machine and one side perceptron for malware detection | |
| Pan et al. | Identification of complex mixtures for Raman spectroscopy using a novel scheme based on a new multi-label deep neural network | |
| Barburiceanu et al. | An improved feature extraction method for texture classification with increased noise robustness | |
| Bhardwaj et al. | Manoeuvre of machine learning algorithms in healthcare sector with application to polycystic ovarian syndrome diagnosis | |
| Farooq et al. | Rapid identification of breast cancer subtypes using micro-FTIR and machine learning methods | |
| JP4873483B2 (en) | Analysis method, analysis program, and analysis apparatus for time series information of signal intensity | |
| Algazinov et al. | Methods of measuring the spectral characteristics and identifying the components of grain mixtures in real-time separation systems | |
| US20210199643A1 (en) | Fluid classification | |
| Gómez-Sanchís et al. | Hyperspectral detection of citrus damage with Mahalanobis kernel classifier | |
| Dumitrache et al. | Automatic detection of skin melanoma from images using natural computing approaches | |
| Boodaghidizaji et al. | Characterizing viral samples using machine learning for Raman and absorption spectroscopy | |
| Chen et al. | Optimal multivariate method for Raman spectroscopy based diagnosis of nasopharyngeal carcinoma | |
| Vrazhnov et al. | The kernel construction for the biomedical data classification using support vector machine | |
| Li et al. | Distinguish the Value of the Benign Nevus and Melanomas Using Machine Learning: A Meta-Analysis and Systematic Review | |
| Zhang et al. | Near infrared spectroscopy drug discrimination method based on stacked sparse auto-encoders extreme learning machine | |
| Zhang et al. | Optimizing Hyperspectral Imaging Classification Performance with CNN and Batch Normalization |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |