CN106950211A - A kind of explosive classifying identification method and system - Google Patents
A kind of explosive classifying identification method and system Download PDFInfo
- Publication number
- CN106950211A CN106950211A CN201710212023.2A CN201710212023A CN106950211A CN 106950211 A CN106950211 A CN 106950211A CN 201710212023 A CN201710212023 A CN 201710212023A CN 106950211 A CN106950211 A CN 106950211A
- Authority
- CN
- China
- Prior art keywords
- fluorescence signal
- explosive
- stage
- fluorescence
- restoration stage
- 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
- G01N21/64—Fluorescence; Phosphorescence
Abstract
The invention provides a kind of explosive method for identifying and classifying and system, it is intended to solves in explosive detection, it is impossible to the problem of carrying out Classification and Identification to each class material.After it is determined that detecting explosive, the change rate curve of the fluorescence signal of explosive is divided into decline stage and Restoration stage, the fluorescence signal signature of decline stage is extracted, the fluorescence signal signature of Restoration stage is extracted using compression sensing method simultaneously, then the feature of extraction is matched successively with several sample characteristics sorter models, the type most matched at last as the explosive type, it is achieved thereby that to the Classification and Identification of explosive.The effect for the Restoration stage fluorescence signal signature that method as a result of compressed sensing is extracted is optimal, and identity is strong, so that when subsequently carrying out Classification and Identification, the requirement to sorter model can be reduced, so that Classification and Identification has higher sensitivity, Classification and Identification process is simpler.
Description
Technical field
The invention belongs to explosives detection techniques field, more particularly to a kind of explosive classifying identification method and system.
Background technology
Worldwide terrorism blast activity seriously threatens the safety of human society, in all kinds of explosives, comprising
Nitro-aromatic including trinitrotoluene (Trinitrotoluene, TNT), dinitrotoluene (DNT) (Dinitro-toluen, DNT)
Class explosive is the first choice of terrorist.Up to the present, explosives detection techniques are divided into two major classes：The physical examination survey technology of explosive
With microscratch amount detection technique (Trace Detection).Wherein, the principle of explosive microscratch amount detection technique is by blast
The gas molecule of thing volatilization and the microscratch amount for the residue for adhering to explosive contactant surface are detected that the technology needs should
Use various spectroscopic techniques, chemical sensor, biosensor technique and detection technique of fluorescence etc..
Fluorescence explosives detection techniques are considered as one of best technology in current microscratch amount detection technique, extensively should
For the security field such as public security, army, government bodies.Fluorescence explosives detection techniques are to be based on explosive and chemical sensitisation
The contact of film, causes the basic optical physics such as fluorescence intensity, spectral shape and the fluorescence polarization or fluorescence anisotropy on film surface
Property changes this principle, realizes to explosive detection.The nitro-aromatic compound that many explosives contain has
Extremely low volatility, while by various noise jammings, therefore be difficult that it is accurately detected.
Equipment required for fluorescence explosives detection techniques is light, even more fast with detection speed, sensitivity high selectivity
Strong the characteristics of, and fluorescence explosives detection techniques have good Detection results for nitro arene explosive substance.But, fluorescence
Explosives detection techniques simultaneously do not have identification function, the skill that also fluorescence explosive is not classified and recognized in the market
Art, it is impossible to, it is necessary to carry out the need of careful classification and identification to each class material while to explosive detection in satisfaction society
Ask.
The content of the invention
The invention provides a kind of explosive method for identifying and classifying and system, it is intended to solves in explosive detection, it is impossible to right
The problem of each class material carries out Classification and Identification.
In order to solve the above technical problems, what the present invention was realized in.The invention provides a kind of explosive Classification and Identification
Method, methods described includes：
The fluorescence signal of the explosive to be measured of acquisition is pre-processed, the fluorescence signal of the explosive to be measured is generated
Change rate curve；
Change rate curve based on the fluorescence signal judges whether the fluorescence intensity change rate of the explosive to be measured reaches
To threshold value, if reaching threshold value, it is determined that detect explosive；
After explosive is detected, the change rate curve of the fluorescence signal based on the explosive draws the fluorescence signal
It is divided into decline stage and Restoration stage；
Fitting a straight line is carried out to generate slope to the rate of change of the fluorescence signal of the decline stage, using the slope as
Decline stage fluorescence signal signature；
Feature extraction is carried out to the fluorescence signal of the Restoration stage based on compression sensing method, it is glimmering with the stage of being restored
Optical signal feature；
By the decline stage fluorescence signal signature and the Restoration stage fluorescence signal signature, with several sample characteristics
Sorter model is matched successively, using the type of matching as the type of the explosive, is known with the classification for realizing explosive
Not.
Further, it is described that feature extraction is carried out to the fluorescence signal of the Restoration stage based on compression sensing method, with
The stage fluorescence signal signature of being restored includes：
If the fluorescence signal of the Restoration stage has limit for length's discrete signal for what a N × 1 was tieed up, by the Restoration stage
The vector x ∈ R that fluorescence signal is constitutedN×1Represent, wherein, R represents real domain, and N represents the length of the fluorescence signal of the Restoration stage
Degree；
Construct M × N calculation matrix Φ ∈ RM×N, wherein, M represents the line number of the calculation matrix, and N represents described
The columns of calculation matrix, the equal length of the columns of the calculation matrix and the fluorescence signal of the Restoration stage；
By the calculation matrix be multiplied by the Restoration stage fluorescence signal constitute vector, obtain a length with it is described
The line number identical compression measurement vector of calculation matrix, the compression measurement vector is special as the Restoration stage fluorescence signal
Levy.
Further, methods described also includes：
Fluorescence signal processing is carried out respectively to several samples of every kind of explosive sample, to obtain every kind of explosive
The fluorescence signal signature collection of sample；
Using support vector machine classifier, features training is carried out to the fluorescence signal signature collection of several explosive sample,
Several described sample characteristics sorter models of generation.
Further, the fluorescence signal of the explosive to be measured of described pair of acquisition is pre-processed, and generates the blast to be measured
The change rate curve of the fluorescence signal of thing includes：
Signal zero averaging processing is carried out to the fluorescence signal of the explosive to be measured of acquisition；
Median filter process is carried out to the fluorescence signal after the processing of signal zero averaging；
Low-pass filtering treatment is carried out to the fluorescence signal after median filter process；
Alpha-beta tracking filter processing is carried out to the fluorescence signal after the low-pass filtering treatment, to obtain at alpha-beta tracking filter
Fluorescence signal after reason；
Based on the fluorescence signal after alpha-beta tracking filter processing, the change of the fluorescence signal of the explosive to be measured is generated
Rate curve.
Further, the fluorescence signal is divided into decline by the change rate curve of the fluorescence signal based on explosive
Stage and Restoration stage include：
Using the minimum point in the change rate curve of the fluorescence signal of the explosive as separation, by the separation
Change rate curve before is divided into the decline stage, and the change rate curve after the separation is divided into the recovery
Stage.
Present invention also offers a kind of explosive classifying and identifying system, the system includes：
Fluorescence signal pretreatment module, the fluorescence signal for the explosive to be measured to acquisition is pre-processed, and generates institute
State the change rate curve of the fluorescence signal of explosive to be measured；
Explosive detection module, the glimmering of the explosive to be measured is judged for the change rate curve based on the fluorescence signal
Whether intensity rate of change reaches threshold value, if reaching threshold value, it is determined that detect explosive；
Fluorescence signal division module, for after explosive is detected, the change of the fluorescence signal based on the explosive
The fluorescence signal is divided into decline stage and Restoration stage by rate curve；
Decline stage characteristic extracting module, the rate of change for the fluorescence signal to the decline stage carries out fitting a straight line
To generate slope, the slope is regard as decline stage fluorescence signal signature；
Restoration stage characteristic extracting module, for being carried out based on compression sensing method to the fluorescence signal of the Restoration stage
Feature extraction, with the stage fluorescence signal signature that is restored；
Classification and Identification module, for the decline stage fluorescence signal signature and the Restoration stage fluorescence signal is special
Levy, matched successively with several sample characteristics sorter models, using the type of matching as the explosive type, with
Realize the Classification and Identification of explosive.
Further, the Restoration stage characteristic extracting module specifically for：
If the fluorescence signal of the Restoration stage has limit for length's discrete signal for what a N × 1 was tieed up, by the Restoration stage
The vector x ∈ R that fluorescence signal is constitutedN×1Represent, wherein, R represents real domain, and N represents the length of the fluorescence signal of the Restoration stage
Degree；
Construct M × N calculation matrix Φ ∈ RM×N, wherein, M represents the line number of the calculation matrix, and N represents described
The columns of calculation matrix, the equal length of the columns of the calculation matrix and the fluorescence signal of the Restoration stage；
By the calculation matrix be multiplied by the Restoration stage fluorescence signal constitute vector, obtain a length with it is described
The line number identical compression measurement vector of calculation matrix, the compression measurement vector is special as the Restoration stage fluorescence signal
Levy.
Further, the system also includes sample characteristics sorter model determining module, the sample characteristics grader
Model determining module includes：
Sample characteristics extraction module, is carried out at fluorescence signal respectively for several samples to every kind of explosive sample
Reason, to obtain the fluorescence signal signature collection of every kind of explosive sample；
Training module, for utilizing support vector machine classifier, to the fluorescence signal signature collection of several explosive sample
Carry out features training, several described sample characteristics sorter models of generation.
Further, the fluorescence signal pretreatment module includes：
Zero averaging processing module, the fluorescence signal for the explosive to be measured to acquisition is carried out at signal zero averaging
Reason；
Median filter process module, for carrying out median filter process to the fluorescence signal after the processing of signal zero averaging；
Low-pass filtering treatment module, for carrying out low-pass filtering treatment to the fluorescence signal after median filter process；
Alpha-beta tracking filter processing module, for carrying out alpha-beta tracking filter to the fluorescence signal after the low-pass filtering treatment
Processing, to obtain the fluorescence signal after the processing of alpha-beta tracking filter；
The change rate curve generation module of fluorescence signal, for based on the alpha-beta tracking filter processing after fluorescence signal,
Generate the change rate curve of the fluorescence signal of the explosive to be measured.
Further, the fluorescence signal division module specifically for：
Using the minimum point in the change rate curve of the fluorescence signal of the explosive as separation, by the separation
Change rate curve before is divided into the decline stage, and the change rate curve after the separation is divided into the recovery
Stage.
Compared with prior art, beneficial effect is the present invention：
Explosive method for identifying and classifying provided by the present invention and system, after it is determined that detecting explosive, by explosive
The change rate curve of fluorescence signal be divided into decline stage and Restoration stage, the fluorescence signal signature of decline stage is carried
Take, while extracted using compression sensing method to the fluorescence signal signature of Restoration stage, if then by the feature of extraction with
A dry sample characteristics sorter model is matched successively, the type most matched at last as the explosive type so that real
The Classification and Identification to explosive is showed.The effect for the Restoration stage fluorescence signal signature that method as a result of compressed sensing is extracted
It is really optimal, and identity is strong, so that when subsequently carrying out Classification and Identification, the requirement to sorter model can be reduced so that
Classification and Identification has higher sensitivity, and Classification and Identification process is simpler.
Brief description of the drawings
Fig. 1 is explosive method for identifying and classifying flow chart provided in an embodiment of the present invention；
Fig. 2 is explosive method for identifying and classifying flow chart provided in an embodiment of the present invention；
Fig. 3 is certain explosive fluorescence signal change rate curve schematic diagram provided in an embodiment of the present invention；
Fig. 4 is svm classifier recognition methods flow chart provided in an embodiment of the present invention；
Fig. 5 is explosive identification categorizing system schematic diagram provided in an embodiment of the present invention；
Fig. 6 is explosive identification categorizing system schematic diagram provided in an embodiment of the present invention.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, it is right below in conjunction with drawings and Examples
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
As one embodiment of the present invention, as shown in figure 1, the invention provides a kind of explosive Classification and Identification side
Method, this method comprises the following steps：
Step S101：The fluorescence signal of the explosive to be measured of acquisition is pre-processed, the glimmering of the explosive to be measured is generated
The change rate curve of optical signal.
Step S102：Change rate curve based on the obtained fluorescence signals of step S101 judges the fluorescence of the explosive to be measured
Whether change rate of strength reaches threshold value, if reaching threshold value, it is determined that detect explosive.Threshold value is even reached, then this is to be measured quick-fried
Fried thing determines it is explosive.
Step S103：After explosive is detected, the change rate curve of the fluorescence signal based on explosive is by fluorescence signal
It is divided into decline stage and Restoration stage.
Step S104：Fitting a straight line is carried out to generate slope to the rate of change of the fluorescence signal of decline stage, by the slope
It is used as decline stage fluorescence signal signature.
Step S105：Feature extraction is carried out to the fluorescence signal of Restoration stage based on compression sensing method, to be restored
Stage fluorescence signal signature.
Step S106：By above-mentioned decline stage fluorescence signal signature and Restoration stage fluorescence signal signature, with several samples
Eigen sorter model is matched successively, using the type of matching as the explosive type, with realize explosive point
Class is recognized.
In summary, the method that first embodiment of the invention is provided, after it is determined that detecting explosive, by explosive
The change rate curve of fluorescence signal be divided into decline stage and Restoration stage, the fluorescence signal signature of decline stage is carried
Take, while extracted using compression sensing method to the fluorescence signal signature of Restoration stage, if then by the feature of extraction with
A dry sample characteristics sorter model is matched successively, the type most matched at last as the explosive type so that real
The Classification and Identification to explosive is showed.The effect for the Restoration stage fluorescence signal signature that method as a result of compressed sensing is extracted
It is really optimal, and identity is strong, so that when subsequently carrying out Classification and Identification, the requirement to sorter model can be reduced so that
Classification and Identification has higher sensitivity, and Classification and Identification process is simpler.
As second embodiment of the present invention, as shown in Fig. 2 the invention provides a kind of explosive Classification and Identification side
Method, this method comprises the following steps：
Step S201：A variety of explosive samples are trained to obtain several sample characteristics sorter models.Step
S201 comprises the following steps：
Step S2011：Fluorescence signal processing is carried out respectively to several samples of every kind of explosive sample, it is described to obtain
The fluorescence signal signature collection of every kind of explosive sample；
Step S2012：Using SVMs (Support Vector Machine, SVM) grader, to several quick-fried
The fluorescence signal signature collection of fried thing sample carries out features training, several described sample characteristics sorter models of generation.Assuming that existing
It is respectively DNT, sulphur, TNT, nitromethane to have four kinds of sample explosives, is by DNT, sulphur, TNT, nitromethane number consecutively
Numbering 1 is to numbering 4.First, numbering 1 and numbering 2 are divided into positive class, numbering 3 and numbering 4 are divided into negative class, carry out feature instruction
Practice generation sample characteristics grader 1；Then, it is positive class by 1 point of numbering, numbering 2 is divided into negative class, carries out features training and generate sample
Eigen grader 2；Meanwhile, by numbering, 3 points are positive class, and numbering 4 is divided into negative class, carry out features training generation sample characteristics point
Class device 3.Because SVM classifier can only classify to two class materials, therefore, when needing to classify to many kinds of substance, just
Need to construct several SVM classifier models.
Step S202：The fluorescence signal of the explosive to be measured of acquisition is pre-processed, the glimmering of the explosive to be measured is generated
The change rate curve of optical signal.Step S202 specifically includes following steps：
Step S2021：Signal zero averaging processing is carried out to the fluorescence signal of the explosive to be measured of acquisition.
Step S2022：Median filter process is carried out to the fluorescence signal after the processing of signal zero averaging.
Step S2023：Low-pass filtering treatment is carried out to the fluorescence signal after median filter process.
Step S2024：Alpha-beta tracking filter processing is carried out to the fluorescence signal after low-pass filtering treatment, to obtain alpha-beta tracking
Fluorescence signal after filtering process.
Step S2025：Based on the fluorescence signal after alpha-beta tracking filter processing, the fluorescence of the explosive to be measured is generated
The change rate curve of signal.
It should be noted that the purpose that step S202 is pre-processed to the fluorescence signal of explosive to be measured is in order to smooth
The fluorescence signal of acquisition, so that influence of the system noise to the fluorescence signal of explosive to be measured is reduced, so as to avoid system noise
Detection and Classification and Identification result to explosive to be measured are impacted.In step S2021, signal zero is carried out to fluorescence signal
Equalization is handled, so as to reduce influence of the system pulses noise to fluorescence signal；In step S2023, at medium filtering
Fluorescence signal after reason carries out low-pass filtering treatment, so as to remove the noise jamming of high band；In step S2024, to glimmering
Optical signal carries out alpha-beta tracking filter processing, so as to improve signal to noise ratio, can export the filtering signal after denoising.Finally according to
The change rate curve of the fluorescence signal of filtering signal generation explosive after denoising.
Step S203：According to the change rate curve of the fluorescence signal of the explosive to be measured obtained in step S202, judging should
Whether the fluorescence intensity change rate of explosive to be measured reaches threshold value, if reaching threshold value, it is determined that detect explosive, it is determined that this is treated
It is explosive to survey explosive.When it is determined that detecting explosive, by ALM staff can be pointed out to detect blast
Thing.It should be noted that existing fluorescence explosives detection techniques are simply by simple physical quantity, the change of such as fluorescence intensity
Whether value reach threshold value to judge whether detected material is explosive, and the present embodiment is by judging the change of fluorescence intensity
Whether rate reaches threshold value to judge whether detected material is explosive.Because the change rate curve of fluorescence intensity is reflected outside
The severe degree that boundary's material reacts with fluorescence membrane, and fluorescence membrane has sensitivity high, to the selectivity of micro- trace explosive
Strong characteristic, therefore when micro- trace explosive and fluorescence membrane reaction reach to a certain degree so that the fluorescence of fluorescence signal is strong
Degree rate of change reaches threshold value.Therefore, detected material is judged by judging whether the rate of change of fluorescence intensity reaches threshold value
Whether it is explosive, improves sensitivity and the accuracy of explosive detection.
Step S204：After explosive is detected, the change rate curve of the fluorescence signal based on explosive is by fluorescence signal
It is divided into decline stage and Restoration stage.Specifically division methods are：With in the change rate curve of the fluorescence signal of explosive
Change rate curve before the separation is divided into the decline stage by minimum point as separation, after the separation
Change rate curve is divided into Restoration stage.As shown in figure 3, the A points in figure are the change rate curve of the fluorescence signal of certain explosive
In minimum point, then the change rate curve before A points be divided into the decline stage, the change rate curve after A points is divided into
Restoration stage.Minimum point is last point of decline stage, is also the starting point of Restoration stage.Now, explosive
The change rate curve of fluorescence signal is the fluorescence signal that the explosive to be measured for being explosive is had determined that in above-mentioned steps S203
Change rate curve.
Step S205：Fitting a straight line is carried out to generate slope to the rate of change of the fluorescence signal of decline stage, by the slope
It is used as decline stage fluorescence signal signature.Fitting a straight line generation slope has a variety of methods, in the present embodiment, use most
A young waiter in a wineshop or an inn multiplies the method generation slope of fitting a straight line.
Step S206：Feature extraction is carried out to the fluorescence signal of Restoration stage based on compression sensing method, to be restored
Stage fluorescence signal signature.The effect of the Restoration stage fluorescence signal signature extracted using the method for compressed sensing is optimal, and knows
Other property is strong, so that when subsequently carrying out Classification and Identification, can reduce the requirement to sorter model so that Classification and Identification has
Higher sensitivity, Classification and Identification process is simpler.Restoration stage fluorescence signal is extracted to the method based on compressed sensing below
Feature carries out simple illustration：
If the fluorescence signal of Restoration stage has limit for length's discrete signal for what a N × 1 was tieed up, believed by the fluorescence of the Restoration stage
Number vector x ∈ R constitutedN×1Represent, wherein, R represents real domain, and N represents the length of the fluorescence signal of the Restoration stage；
Construct M × N calculation matrix Φ ∈ RM×N, wherein, M represents the line number of calculation matrix, and N represents calculation matrix
Columns, therefore the equal length of the fluorescence signal of the columns of the calculation matrix and above-mentioned Restoration stage, all represented with N；
The vector that the fluorescence signal that above-mentioned calculation matrix is multiplied by into above-mentioned Restoration stage is constituted, obtains a length and the survey
The line number identical compression measurement vector of moment matrix, regard compression measurement vector as Restoration stage fluorescence signal signature.
In the present embodiment, to implement process as follows by step S206：
Construct a calculation matrix Φ=[φ1,φ2,...,φk]T, wherein k=10, φnRepresent n-th of base vector, n=
1,2 ..., k, each base vector φnThe random number for being zero by a class mean is constituted, each base vector φnLength be 300, with
The fluorescence signal length of Restoration stage is identical, i.e. the fluorescence signal length of Restoration stage is also 300.By calculation matrix with recovering rank
The fluorescence signal of section is multiplied, and the compression measurement vector of one ten dimension is obtained, as shown in following formula：
Y=Φ x
Wherein, x represents the fluorescence signal of Restoration stage, and Φ represents calculation matrix, and y represents compression measurement vector, that is, recovered
Projections of the fluorescence signal x in stage on calculation matrix Φ, its length is identical with calculation matrix Φ line number.Compression measurement vector
Y is the fluorescence signal x of Restoration stage another expression, thus can as Restoration stage fluorescence signal signature.
Step S207：By above-mentioned decline stage fluorescence signal signature and Restoration stage fluorescence signal signature, with several samples
Eigen sorter model is matched successively, using the type of matching as the explosive type, with realize explosive point
Class is recognized.
For example shown in Fig. 4, when performing step S207, by the decline stage fluorescence signal signature and Restoration stage of explosive
Fluorescence signal signature, is matched with generated grader 1 first, judges the positive class that the explosive belongs in grader 1, also
It is the negative class belonged in grader 1；If judge the positive class that the explosive belongs in grader 1, then by the lower depression of order of the explosive
Section fluorescence signal signature and Restoration stage fluorescence signal signature, are matched with generated grader 2, judge that the explosive belongs to
Positive class in grader 2, still falls within the negative class in grader 1, if judging to belong to positive class, identifies that the explosive is
DNT.In a word, the fluorescence signal signature of explosive with multiple graders by carrying out Classification and Identification, and last output result is specially
Which kind of explosive sample.
From the foregoing, the type that the premise of Classification and Identification is explosive to be identified is obtained in classification based training
A type in sample characteristics sorter model storehouse.
The method that the present embodiment is provided has carried out multiple test under portable explosive detection instrument platform.First, will
Fluorescence signal is pre-processed, by judging whether the rate of change that alpha-beta tracking filter is generated reaches that threshold value decision systems are alerted.Will
Effective fluorescence signal after alarm is divided into two processes of decline stage and Restoration stage and carries out feature extraction.Decline stage passes through glimmering
The rate of change fitting a straight line generation slope of optical signal extracts one-dimensional characteristic, then fluorescence signal is carried out by the method for compressed sensing
Feature extraction generates ten dimensional features.The total dimension of feature finally extracted is ten one-dimensional.By experiment, four kinds of explosives are carried out
Detection identification and classification, every kind of material have done 500 tests, have obtained following result as shown in table 1：
Table 1：Explosive detection and Classification and Identification result
DNT alarm rate is 99.6%, and discrimination is 97.6%；The alarm rate of sulphur is 99.2%, and discrimination is
92.4%；TNT alarm rate is 99.8%, and discrimination is 95.4%；The alarm rate of nitromethane is 98.8%, and discrimination is
93.6%.It can draw, the present invention has very strong rejection ability to the noise jamming of system, with very high alarm rate.Together
When, the present invention has good identification function, these four materials can effectively recognize DNT, sulphur, TNT and nitromethane
And classified, accuracy is all more than 90 percent.
In summary, because the species of explosive is various, and packaging, to hide situation complicated, therefore these factors cause pair
The accurate detection of latent explosive is very difficult, and the method that second embodiment of the invention is provided, it is determined that detecting quick-fried
After fried thing, the change rate curve of the fluorescence signal of explosive is divided into decline stage and Restoration stage, it is glimmering to the decline stage
Optical signal feature is extracted, while being extracted using compression sensing method to the fluorescence signal signature of Restoration stage, then
The feature of extraction is matched successively with several sample characteristics sorter models, the type most matched at last is used as the blast
The type of thing, it is achieved thereby that to the Classification and Identification of explosive.The Restoration stage that method as a result of compressed sensing is extracted
The effect of fluorescence signal signature is optimal, and identity is strong, so that when subsequently carrying out Classification and Identification, can reduce to grader
The requirement of model so that Classification and Identification has higher sensitivity, Classification and Identification process is simpler.
As the 3rd embodiment of the present invention, as shown in figure 5, the invention provides a kind of explosive Classification and Identification system
System, the system includes：
Fluorescence signal pretreatment module 101, the fluorescence signal for the explosive to be measured to acquisition is pre-processed, generation
The change rate curve of the fluorescence signal of the explosive to be measured.
Explosive detection module 102, the explosive to be measured is judged for the change rate curve based on above-mentioned fluorescence signal
Whether fluorescence intensity change rate reaches threshold value, if reaching threshold value, it is determined that detect explosive.
Fluorescence signal division module 103, for after explosive is detected, the rate of change of the fluorescence signal based on explosive
The fluorescence signal is divided into decline stage and Restoration stage by curve.
Decline stage characteristic extracting module 104, the rate of change for the fluorescence signal to the decline stage carries out fitting a straight line
To generate slope, the slope is regard as decline stage fluorescence signal signature.
Restoration stage characteristic extracting module 105, for believing by based on compression sensing method the fluorescence of the Restoration stage
Number carry out feature extraction, with the stage fluorescence signal signature that is restored.
Classification and Identification module 106, for by decline stage fluorescence signal signature and Restoration stage fluorescence signal signature, if with
A dry sample characteristics sorter model is matched successively, using the type of matching as the explosive type, to realize blast
The Classification and Identification of thing.
In summary, the explosive classifying and identifying system that third embodiment of the invention is provided, fluorescence signal divides mould
The change rate curve of the fluorescence signal of explosive is divided into decline stage and Restoration stage by block 103 after explosive is detected；
The fluorescence signal signature of decline stage characteristic extracting module 102 and Restoration stage characteristic extracting module 103 respectively to the decline stage
Extracted with the fluorescence signal signature of Restoration stage, then Classification and Identification module 104 is by the feature of extraction and several samples
Feature classifiers model is matched successively, the type most matched at last as the explosive type, it is achieved thereby that to quick-fried
The Classification and Identification of fried thing.Restoration stage characteristic extracting module 103 employs the Restoration stage fluorescence that the method for compressed sensing is extracted
The effect of signal characteristic is optimal, and identity is strong, so that when subsequently carrying out Classification and Identification, can reduce to sorter model
Requirement so that Classification and Identification has higher sensitivity, and Classification and Identification process is simpler.
As the 4th embodiment of the present invention, as shown in fig. 6, the invention provides a kind of explosive Classification and Identification system
System, the system includes sample characteristics sorter model determining module 201, fluorescence signal pretreatment module 202, explosive detection mould
Block 203, fluorescence signal division module 204, decline stage characteristic extracting module 205, Restoration stage characteristic extracting module 206 and
Classification and Identification module 207.
Before Classification and Identification is carried out to explosive, sample characteristics sorter model determining module 201 is first to a variety of blasts
Thing sample is trained to obtain several sample characteristics sorter models.Sample characteristics sorter model determining module 201 is wrapped
Include：
Sample characteristics extraction module 2011, fluorescence signal is carried out for several samples to every kind of explosive sample respectively
Processing, to obtain the fluorescence signal signature collection of every kind of explosive sample.
Training module 2012, it is special to the fluorescence signal of several explosive sample for utilizing support vector machine classifier
Collection carries out features training, several described sample characteristics sorter models of generation.Assuming that existing four kinds of sample explosives difference
It is numbering 1 to numbering 4 by DNT, sulphur, TNT, nitromethane number consecutively for DNT, sulphur, TNT, nitromethane.First, will
Numbering 1 and numbering 2 are divided into positive class, and numbering 3 and numbering 4 are divided into negative class, carry out features training generation sample characteristics grader
1；Then, it is positive class by 1 point of numbering, numbering 2 is divided into negative class, carries out features training and generate sample characteristics grader 2；Meanwhile,
By numbering, 3 points are positive class, and numbering 4 is divided into negative class, carry out features training generation sample characteristics grader 3.Due to SVM classifier
Two class materials can only be classified.Therefore when needing to classify to many kinds of substance, it is necessary to construct several svm classifiers
Device model.
Fluorescence signal pretreatment module 202 is pre-processed to the fluorescence signal of explosive to be measured, to obtain blast to be measured
The change rate curve of the fluorescence signal of thing, fluorescence signal pretreatment module 202 includes：
Zero averaging processing module 2021, the fluorescence signal for the explosive to be measured to acquisition carries out signal zero averaging
Processing.
Median filter process module 2022, for being carried out to the fluorescence signal after the processing of signal zero averaging at medium filtering
Reason.
Low-pass filtering treatment module 2023, for carrying out low-pass filtering treatment to the fluorescence signal after median filter process.
Alpha-beta tracking filter processing module 2024, for carrying out alpha-beta tracking to the fluorescence signal after the low-pass filtering treatment
Filtering process, to obtain the fluorescence signal after the processing of alpha-beta tracking filter.
The change rate curve generation module 2025 of fluorescence signal, for based on alpha-beta tracking filter processing after fluorescence signal,
Generate the change rate curve of the fluorescence signal of explosive to be measured.
It should be noted that what fluorescence signal pretreatment module 202 was pre-processed to the fluorescence signal of explosive to be measured
Purpose is the fluorescence signal in order to smoothly obtain, so that influence of the system noise to the fluorescence signal of explosive to be measured is reduced, from
And avoid detection and Classification and Identification result of the system noise to explosive to be measured from impacting.Zero averaging processing module 2021
Signal zero averaging processing is carried out to fluorescence signal, so as to reduce influence of the system pulses noise to fluorescence signal；Low pass filtered
Ripple processing module 2023 carries out low-pass filtering treatment to the fluorescence signal after median filter process, so as to remove making an uproar for high band
Acoustic jamming；Alpha-beta tracking filter processing module 2024 carries out alpha-beta tracking filter processing to fluorescence signal, so that signal to noise ratio is improved,
The filtering signal after denoising can be exported, final 2025 fluorescence for generating explosive to be measured according to the filtering signal after the denoising are believed
Number change rate curve.
Explosive detection module 203, for the change rate curve of the fluorescence signal according to explosive to be measured, judges that this is to be measured
Whether the fluorescence intensity change rate of explosive reaches threshold value, if reaching threshold value, it is determined that detect explosive, that is, determines that this is to be measured
Explosive is explosive.When it is determined that detecting explosive, by ALM staff can be pointed out to detect explosive.
It should be noted that existing fluorescence explosives detection techniques are simply by simple physical quantity, the change value of such as fluorescence intensity
Threshold value whether is reached to judge whether detected material is explosive, and the present embodiment is by judging the rate of change of fluorescence intensity
Threshold value whether is reached to judge whether detected material is explosive.Because the change rate curve of fluorescence intensity reflects the external world
The severe degree that material reacts with fluorescence membrane, and fluorescence membrane has sensitivity high, the selectivity to micro- trace explosive is strong
Characteristic, therefore when micro- trace explosive and fluorescence membrane reaction reach to a certain degree when so that the fluorescence intensity of fluorescence signal
Rate of change reaches threshold value.Therefore, judge that detected material is by judging whether the rate of change of fluorescence intensity reaches threshold value
No is explosive, improves sensitivity and the accuracy of explosive detection.
Fluorescence signal division module 204, for after explosive is detected, the rate of change of the fluorescence signal based on explosive
The fluorescence signal is divided into decline stage and Restoration stage by curve.Fluorescence signal division module 204 is with the fluorescence of explosive
Change rate curve before separation is divided into lower depression of order by the minimum point in the change rate curve of signal as separation
Section, Restoration stage is divided into by the change rate curve after separation.Now, the change rate curve of the fluorescence signal of explosive is
For the change rate curve for the fluorescence signal that the explosive to be measured for being explosive is had determined that in above-mentioned 203.
Decline stage characteristic extracting module 205, the rate of change for the fluorescence signal to the decline stage carries out fitting a straight line
To generate slope, the slope is regard as decline stage fluorescence signal signature.
Restoration stage characteristic extracting module 206, for based on fluorescence signal of the compression sensing method to the Restoration stage
Feature extraction is carried out, with the stage fluorescence signal signature that is restored.
In the present embodiment, Restoration stage characteristic extracting module 206 specifically for：
If the fluorescence signal of Restoration stage has limit for length's discrete signal for what a N × 1 was tieed up, believed by the fluorescence of the Restoration stage
Number vector x ∈ R constitutedN×1Represent, wherein, R represents real domain, and N represents the length of the fluorescence signal of the Restoration stage；
Construct M × N calculation matrix Φ ∈ RM×N, wherein, M represents the line number of calculation matrix, and N represents calculation matrix
Columns, therefore the equal length of the fluorescence signal of the columns of the calculation matrix and above-mentioned Restoration stage, all represented with N；
The vector that the fluorescence signal that above-mentioned calculation matrix is multiplied by into above-mentioned Restoration stage is constituted, obtains a length and the survey
The line number identical compression measurement vector of moment matrix, regard compression measurement vector as Restoration stage fluorescence signal signature.
For example, in the present embodiment, constructing a calculation matrix Φ=[φ1,φ2,...,φk]T, wherein k=10, φnTable
Show n-th of base vector, n=1,2 ..., k, each base vector φnThe random number for being zero by a class mean is constituted, each base vector
φnLength be 300, identical with the fluorescence signal length of Restoration stage, i.e. the fluorescence signal length of Restoration stage is also 300.
Calculation matrix is multiplied with the fluorescence signal of Restoration stage, the compression measurement vector of one ten dimension is obtained, as shown in following formula：
Y=Φ x
Wherein, x represents the fluorescence signal of Restoration stage, and Φ represents calculation matrix, and y represents compression measurement vector, that is, recovered
Projections of the fluorescence signal x in stage on calculation matrix Φ, its length is identical with calculation matrix Φ line number.Compression measurement vector
Y is the fluorescence signal x of Restoration stage another expression, thus can as Restoration stage fluorescence signal signature.
Classification and Identification module 207, for by decline stage fluorescence signal signature and Restoration stage fluorescence signal signature, if with
A dry sample characteristics sorter model is matched successively, using the type of matching as the explosive type, to realize blast
The Classification and Identification of thing.For example shown in Fig. 4, according to above-mentioned 201 illustrated examples of sample characteristics sorter model determining module, classification
Identification module 207 is when being identified, by the decline stage fluorescence signal signature and Restoration stage fluorescence signal signature of explosive,
Matched first with generated grader 1, judge the positive class that the explosive belongs in grader 1, still fall within grader 1
In negative class；If judge the positive class that the explosive belongs in grader 1, then the decline stage fluorescence signal of the explosive is special
Seek peace Restoration stage fluorescence signal signature, matched with generated grader 2, judge that the explosive belongs in grader 2
Positive class, still fall within the negative class in grader 1, if judge belong to positive class, identify the explosive be DNT.In a word, explode
The fluorescence signal signature of thing with several graders by carrying out Classification and Identification, and which kind of explosive is last output result be specially
Sample.
In summary, the explosive classifying and identifying system that four embodiment of the invention is provided, fluorescence signal divides mould
The change rate curve of the fluorescence signal of explosive is divided into decline stage and Restoration stage by block after explosive is detected；Under
Depression of order section characteristic extracting module and Restoration stage characteristic extracting module to the fluorescence signal signature of decline stage and recover rank respectively
The fluorescence signal signature of section is extracted, and then Classification and Identification module is by the feature of extraction and several sample characteristics grader moulds
Type is matched successively, the type most matched at last as the explosive type, it is achieved thereby that knowing to the classification of explosive
Not.The effect for the Restoration stage fluorescence signal signature that the method that Restoration stage characteristic extracting module employs compressed sensing is extracted is most
It is excellent, and identity is strong, so that when subsequently carrying out Classification and Identification, the requirement to sorter model can be reduced so that classification
Identification has higher sensitivity, and Classification and Identification process is simpler.
Presently preferred embodiments of the present invention is the foregoing is only, not to limit invention, all spirit in the present invention
With any modification, equivalent and the improvement made within principle etc., it should be included within the scope of the present invention.
Claims (10)
1. a kind of explosive classifying identification method, it is characterised in that methods described includes：
The fluorescence signal of the explosive to be measured of acquisition is pre-processed, the change of the fluorescence signal of the explosive to be measured is generated
Rate curve；
Change rate curve based on the fluorescence signal judges whether the fluorescence intensity change rate of the explosive to be measured reaches threshold
Value, if reaching threshold value, it is determined that detect explosive；
After explosive is detected, the fluorescence signal is divided into by the change rate curve of the fluorescence signal based on the explosive
Decline stage and Restoration stage；
Fitting a straight line is carried out to generate slope to the rate of change of the fluorescence signal of the decline stage, the slope is regard as decline
Stage fluorescence signal signature；
Feature extraction is carried out to the fluorescence signal of the Restoration stage based on compression sensing method, with the stage fluorescence letter that is restored
Number feature；
By the decline stage fluorescence signal signature and the Restoration stage fluorescence signal signature, classify with several sample characteristics
Device model is matched successively, using the type of matching as the explosive type, to realize the Classification and Identification of explosive.
2. the method as described in claim 1, it is characterised in that it is described based on compression sensing method to the glimmering of the Restoration stage
Optical signal carries out feature extraction, and to be restored, stage fluorescence signal signature includes：
If the fluorescence signal of the Restoration stage has limit for length's discrete signal for what a N × 1 was tieed up, by the fluorescence of the Restoration stage
The vector x ∈ R that signal is constitutedN×1Represent, wherein, R represents real domain, and N represents the length of the fluorescence signal of the Restoration stage；
Construct M × N calculation matrix Φ ∈ RM×N, wherein, M represents the line number of the calculation matrix, and N represents the measurement
Matrix column number, the equal length of the columns of the calculation matrix and the fluorescence signal of the Restoration stage；
The vector that the fluorescence signal that the calculation matrix is multiplied by into the Restoration stage is constituted, obtains a length and the measurement
The line number identical compression measurement vector of matrix, regard the compression measurement vector as the Restoration stage fluorescence signal signature.
3. the method as described in claim 1, it is characterised in that methods described also includes：
Fluorescence signal processing is carried out respectively to several samples of every kind of explosive sample, to obtain every kind of explosive sample
Fluorescence signal signature collection；
Using support vector machine classifier, features training, generation are carried out to the fluorescence signal signature collection of several explosive sample
Several described sample characteristics sorter models.
4. the method as described in claim 1, it is characterised in that the fluorescence signal of the explosive to be measured of described pair of acquisition carries out pre-
Processing, generating the change rate curve of the fluorescence signal of the explosive to be measured includes：
Signal zero averaging processing is carried out to the fluorescence signal of the explosive to be measured of acquisition；
Median filter process is carried out to the fluorescence signal after the processing of signal zero averaging；
Low-pass filtering treatment is carried out to the fluorescence signal after median filter process；
Alpha-beta tracking filter processing is carried out to the fluorescence signal after the low-pass filtering treatment, to obtain after the processing of alpha-beta tracking filter
Fluorescence signal；
Based on the fluorescence signal after alpha-beta tracking filter processing, the rate of change of the fluorescence signal of the explosive to be measured is generated
Curve.
5. the method as described in claim 1, it is characterised in that the change rate curve of the fluorescence signal based on explosive will
The fluorescence signal, which is divided into decline stage and Restoration stage, to be included：
Using the minimum point in the change rate curve of the fluorescence signal of the explosive as separation, before the separation
Change rate curve be divided into the decline stage, the change rate curve after the separation is divided into the recovery rank
Section.
6. a kind of explosive classifying and identifying system, it is characterised in that the system includes：
Fluorescence signal pretreatment module, the fluorescence signal for the explosive to be measured to acquisition is pre-processed, and is treated described in generation
Survey the change rate curve of the fluorescence signal of explosive；
Explosive detection module, judges that the fluorescence of the explosive to be measured is strong for the change rate curve based on the fluorescence signal
Whether degree rate of change reaches threshold value, if reaching threshold value, it is determined that detect explosive；
Fluorescence signal division module, for after explosive is detected, the rate of change of the fluorescence signal based on the explosive to be bent
The fluorescence signal is divided into decline stage and Restoration stage by line；
Decline stage characteristic extracting module, the rate of change for the fluorescence signal to the decline stage carries out fitting a straight line with life
Into slope, the slope is regard as decline stage fluorescence signal signature；
Restoration stage characteristic extracting module, for carrying out feature to the fluorescence signal of the Restoration stage based on compression sensing method
Extract, with the stage fluorescence signal signature that is restored；
Classification and Identification module, for by the decline stage fluorescence signal signature and the Restoration stage fluorescence signal signature, with
Several sample characteristics sorter models are matched successively, using the type of matching as the explosive type, with realize
The Classification and Identification of explosive.
7. system as claimed in claim 6, it is characterised in that the Restoration stage characteristic extracting module specifically for：
If the fluorescence signal of the Restoration stage has limit for length's discrete signal for what a N × 1 was tieed up, by the fluorescence of the Restoration stage
The vector x ∈ R that signal is constitutedN×1Represent, wherein, R represents real domain, and N represents the length of the fluorescence signal of the Restoration stage；
Construct M × N calculation matrix Φ ∈ RM×N, wherein, M represents the line number of the calculation matrix, and N represents the measurement
Matrix column number, the equal length of the columns of the calculation matrix and the fluorescence signal of the Restoration stage；
The vector that the fluorescence signal that the calculation matrix is multiplied by into the Restoration stage is constituted, obtains a length and the measurement
The line number identical compression measurement vector of matrix, regard the compression measurement vector as the Restoration stage fluorescence signal signature.
8. system as claimed in claim 6, it is characterised in that the system also determines mould including sample characteristics sorter model
Block, the sample characteristics sorter model determining module includes：
Sample characteristics extraction module, fluorescence signal processing is carried out for several samples to every kind of explosive sample respectively, with
Obtain the fluorescence signal signature collection of every kind of explosive sample；
Training module, for utilizing support vector machine classifier, is carried out to the fluorescence signal signature collection of several explosive sample
Features training, several described sample characteristics sorter models of generation.
9. system as claimed in claim 6, it is characterised in that the fluorescence signal pretreatment module includes：
Zero averaging processing module, the fluorescence signal for the explosive to be measured to acquisition carries out signal zero averaging processing；
Median filter process module, for carrying out median filter process to the fluorescence signal after the processing of signal zero averaging；
Low-pass filtering treatment module, for carrying out low-pass filtering treatment to the fluorescence signal after median filter process；
Alpha-beta tracking filter processing module, for being carried out to the fluorescence signal after the low-pass filtering treatment at alpha-beta tracking filter
Reason, to obtain the fluorescence signal after the processing of alpha-beta tracking filter；
The change rate curve generation module of fluorescence signal, for based on the fluorescence signal after alpha-beta tracking filter processing, generation
The change rate curve of the fluorescence signal of the explosive to be measured.
10. system as claimed in claim 6, it is characterised in that the fluorescence signal division module specifically for：
Using the minimum point in the change rate curve of the fluorescence signal of the explosive as separation, before the separation
Change rate curve be divided into the decline stage, the change rate curve after the separation is divided into the recovery rank
Section.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710212023.2A CN106950211B (en) | 2017-04-01 | 2017-04-01 | A kind of explosive classifying identification method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710212023.2A CN106950211B (en) | 2017-04-01 | 2017-04-01 | A kind of explosive classifying identification method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106950211A true CN106950211A (en) | 2017-07-14 |
| CN106950211B CN106950211B (en) | 2019-08-16 |
Family
ID=59475159
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710212023.2A Active CN106950211B (en) | 2017-04-01 | 2017-04-01 | A kind of explosive classifying identification method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106950211B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112505244A (en) * | 2020-11-05 | 2021-03-16 | 福建美营自动化科技有限公司 | Circuit and method for quickly distinguishing low-concentration gasoline and alcohol in air |
| CN113075180A (en) * | 2021-03-24 | 2021-07-06 | 临海市鸥巡电子科技有限公司 | Method and system for detecting change trend of fluorescence data |
| CN115728276A (en) * | 2022-11-14 | 2023-03-03 | 中船重工安谱（湖北）仪器有限公司 | Explosive detection method and detection system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012110791A1 (en) * | 2011-02-14 | 2012-08-23 | Laser Optical Engineering Limited | Improvements in or relating to detecting illicit substances in a target area |
| CN102879375A (en) * | 2012-10-12 | 2013-01-16 | 陕西师范大学 | Identification detection method for nitro-aromatic explosives |
| CN103293139A (en) * | 2013-06-28 | 2013-09-11 | 陕西师范大学 | Sensing device of a fluorescent exploder detector |
| US9013686B2 (en) * | 2011-11-10 | 2015-04-21 | Cdex, Inc | Chemical and molecular identification and quantification system utilizing enhanced photoemission spectroscopy |
-
2017
- 2017-04-01 CN CN201710212023.2A patent/CN106950211B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012110791A1 (en) * | 2011-02-14 | 2012-08-23 | Laser Optical Engineering Limited | Improvements in or relating to detecting illicit substances in a target area |
| US9013686B2 (en) * | 2011-11-10 | 2015-04-21 | Cdex, Inc | Chemical and molecular identification and quantification system utilizing enhanced photoemission spectroscopy |
| CN102879375A (en) * | 2012-10-12 | 2013-01-16 | 陕西师范大学 | Identification detection method for nitro-aromatic explosives |
| CN103293139A (en) * | 2013-06-28 | 2013-09-11 | 陕西师范大学 | Sensing device of a fluorescent exploder detector |
Non-Patent Citations (2)
| Title |
|---|
| XIANGCHENG SUN ET AL: "Fluorescence based explosive detection: from mechanisms to sensory materials", 《CHEM SOC REV》 * |
| 王超宇 等: "爆炸物检测用化学传感器的研究进展_爆炸物检测用化学传感器的研究进展", 《理化检验-化学分册》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112505244A (en) * | 2020-11-05 | 2021-03-16 | 福建美营自动化科技有限公司 | Circuit and method for quickly distinguishing low-concentration gasoline and alcohol in air |
| CN112505244B (en) * | 2020-11-05 | 2023-03-14 | 福建美营自动化科技有限公司 | Circuit and method for quickly distinguishing low-concentration gasoline and alcohol in air |
| CN113075180A (en) * | 2021-03-24 | 2021-07-06 | 临海市鸥巡电子科技有限公司 | Method and system for detecting change trend of fluorescence data |
| CN115728276A (en) * | 2022-11-14 | 2023-03-03 | 中船重工安谱（湖北）仪器有限公司 | Explosive detection method and detection system |
| CN115728276B (en) * | 2022-11-14 | 2024-01-23 | 中船重工安谱（湖北）仪器有限公司 | Explosive detection method and detection system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106950211B (en) | 2019-08-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106950211B (en) | A kind of explosive classifying identification method and system | |
| CN103544392B (en) | Medical science Gas Distinguishing Method based on degree of depth study | |
| CN104331436A (en) | Rapid classification method of malicious codes based on family genetic codes | |
| CN111436944B (en) | Falling detection method based on intelligent mobile terminal | |
| EP3798924A1 (en) | System and method for classifying manufactured products | |
| Bedoui et al. | Electronic nose system and principal component analysis technique for gases identification | |
| US7531137B2 (en) | Substance detection system | |
| Cho et al. | Sensitive and fast recognition of explosives using fluorescent polymer sensors and pattern recognition analysis | |
| Karkra et al. | Analysis of heavy metal ions in potable water using soft computing technique | |
| CN112580486B (en) | Human behavior classification method based on radar micro-Doppler signal separation | |
| Kim et al. | Graph convolution networks for seismic events classification using raw waveform data from multiple stations | |
| CN107358248A (en) | A kind of method for improving fall detection system precision | |
| CN103530373A (en) | Mobile application classifying method under imbalanced perception data | |
| Dickey et al. | Beyond correlation: A path‐invariant measure for seismogram similarity | |
| CN102362282A (en) | Signal classification method and signal classification device | |
| Johnsson | Structures in high-dimensional data: Intrinsic dimension and cluster analysis | |
| Kundu et al. | Classification of tea samples using SVM as machine learning component of E-tongue | |
| Ku et al. | UAV payload detection using deep learning and data augmentation | |
| Madureira et al. | A neural network seismic detector | |
| Tan et al. | A sparse representation-based classifier for in-set bird phrase verification and classification with limited training data | |
| CN115078327A (en) | Rapid detection method for dangerous chemicals based on neural network architecture search | |
| CN101884045A (en) | Mixed statistical and numerical model for sensor array detection and classification | |
| CN104123538A (en) | Network malicious image detection method based on visual word bag | |
| Karunaratna et al. | Sound Event Recognition and Classification Using Machine Learning Techniques | |
| Omijeh et al. | Optimizing a Sensor to Detect Ammonium Nitrate Based IEDS in Vehicles Using Artificial Neural Networks |
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 | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 518060 3688 Nanhai Road, Nanshan District, Shenzhen, Guangdong Applicant after: Shenzhen University Applicant after: Shenzhen sword Defense Technology Co. Ltd. Address before: 518060 3688 Nanhai Road, Nanshan District, Shenzhen, Guangdong Applicant before: Shenzhen University Applicant before: Shaanxi Normal University |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20200121 Address after: 518000 Guangdong Province Nanshan Street Nanxin Road Sunshine Creation Center Phase I A Block 701, Nanshan District, Shenzhen City Patentee after: Shenzhen sword Defense Technology Co. Ltd. Address before: 518060 Nanhai Road, Guangdong, Shenzhen, No. 3688, No. Co-patentee before: Shenzhen sword Defense Technology Co. Ltd. Patentee before: Shenzhen University |