CN106056339B - A kind of article identification method using electromagnetic wave penetrability - Google Patents
A kind of article identification method using electromagnetic wave penetrability Download PDFInfo
- Publication number
- CN106056339B CN106056339B CN201610388891.1A CN201610388891A CN106056339B CN 106056339 B CN106056339 B CN 106056339B CN 201610388891 A CN201610388891 A CN 201610388891A CN 106056339 B CN106056339 B CN 106056339B
- Authority
- CN
- China
- Prior art keywords
- label
- signal
- article
- chest
- training set
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06K—GRAPHICAL DATA READING; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
- G06K7/00—Methods or arrangements for sensing record carriers, e.g. for reading patterns
- G06K7/10—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation
- G06K7/10009—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation sensing by radiation using wavelengths larger than 0.1 mm, e.g. radio-waves or microwaves
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
- G06Q10/087—Inventory or stock management, e.g. order filling, procurement or balancing against orders
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Quality & Reliability (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Human Resources & Organizations (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Toxicology (AREA)
- Electromagnetism (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses a kind of article identification methods using electromagnetic wave penetrability, by pasting RF tag common in logistics system in chest bottom, capture the signal that different labels are transmitted under adjacent time point, the influence of moving articles (article of the people or movement that such as walk about) in ambient enviroment and environment is eliminated, article is to influence caused by electromagnetic wave signal in acquisition chest.Using this influence, identify whether article is illegally stolen or destroyed in chest.Identification method proposed by the present invention, the method of article or package is identified and checked before being different from by the appraisement label true and false, by the truth of article and attribute load on electromagnetic wave signal, it is truly realized identification article itself, security risk in logistics system is made up, relative to existing method, safety is had more.
Description
Technical field
The present invention relates to Internet of Things field of sensing technologies, and in particular to a kind of article identification side using electromagnetic wave penetrability
Method.
Background technique
In logistics now, storage field, Radio Frequency Identification Technology gradually is widely applied.Compared to two before
The advantages of technologies such as dimension code, bar code, Radio Frequency Identification Technology, is: (1) in addition to label can be utilized to carry certain information, also
What be can be convenient is modified information using reader.(2) it is not limited to for label to be placed on specified orientation, as long as extremely by label
In the read range of reader, can be written and read.
In existing radio frequency identification equipment, it is divided into hyperfrequency, high frequency etc., is respectively used under different scene and application.
In the warehousing system of logistics, the method that tradition identifies article or package, is on the packing box of package or article
Print the mode of two dimensional code, bar code.When identifying and identifying article, by scanning the two-dimensional code and bar code, read therein
Information is identified article and is identified.But security risk under cover very big under such mode: (1) firstly, two dimension
Code and bar code are easy to be replicated.It is only necessary to replicate the two dimensional code and bar code can substitute or pretend to be just to attacker
Normal cargo.(2) secondly, scanning the two-dimensional code with bar code is a kind of mode for identifying article appended by it, and in logistics and storage
In environment, two dimensional code or bar code are usually printed on external packing box, in this way, attacker is not destroying two dimensional code and bar code
Under the premise of, the means that can be disassembled by violence substitute or take away the article in external packing box.The two kinds of above security risks,
Under the premise of not unpacking mounted box, it is not easy very much to be found and discover.
Summary of the invention
For above-mentioned problems of the prior art, the object of the present invention is to provide one kind to be based on EPC Global
Whether the application of hyperfrequency under Class1 Gen2 agreement can identify the interior article of package again while identifying package
The method for being replaced and taking away is suitable for logistics, situations such as warehousing system.
In order to realize above-mentioned task, the invention adopts the following technical scheme:
A kind of article identification method using electromagnetic wave penetrability, comprising the following steps:
Step 1, the data training of article to be measured
Tag array is attached to the bottom in chest, and article to be measured is put into chest and is packaged by step S10；
Step S11 is provided with Wireless RF identifier and general software radio peripheral hardware outside chest, wherein wirelessly penetrating
The read antenna of frequency identifier is located at the lower section of chest, and the antenna of general software radio peripheral hardware is located at the top of chest；
Step S12 monitors the read antenna and chest interior label array using general software radio peripheral hardware antenna
Communication process, obtain original signal；
The EPC signal that tag array in original signal is replied is intercepted out by step S13, and letter is extracted from EPC signal
Number feature, determines label belonging to EPC signal using classifier；
Step S14 is removed in external environment and environment after the EPC signal alignment of labels different in tag array
The influence of mobile article obtains training set, and saves training set；
Step 2, the identification of article to be measured
Step S20 when actually detected, is equipped with one in the chest of article to be measured, and the bottom in chest has and step S10
The tag array of same procedure setting；The chest equipped with article to be measured is carried out according to step S11 to the identical method of step S14
Processing, the difference is that finally obtaining test set at this time rather than training set；
Whether test set is compared step S21 with training set, be replaced and destroyed with determination article to be measured.
Further, the tag array includes two labels, respectively the first label and the second label.
Further, the classifier uses K-means classifier.
Further, the detailed process of acquisition training data includes: in the step S14
Remember that the first label, the second label obtain corresponding original signal S1、S2It is respectively as follows:
S1=F1+Nen+Neq+Nw (1)
S2=F2+N′en+N′eq+N′w (2)
In above formula, F1、F2Respectively indicate the first label, the signal when signal of the second label penetrates article to be measured in chest
Feature, Nen、N′enIt is influence of the external environment locating for chest to the first label, the second label signal, Neq、N′enIt is using not
Influence of the same equipment to the first label, the second label signal, Nw、N′wIt is the Gauss white noise of the first label, the second label signal
Sound；
By original signal S1、S2Subtracting each other can obtain:
S1-S2=(F1-F2)+(Nen-N′en)+(Neq-N′eq)+(Nw-N′w) (3)
In above formula, the external environment as locating for chest be for the first label and the second label it is similar,
(Nen-N′en)≈0；In same primary test, the equipment that the first label and the second label use is the same, therefore (Neq-N′eq)
≈0；First label, the second label signal white Gaussian noise be independent from each other, enable G=Nw-N′w；Then according to white Gaussian noise
Probability distribution can obtainIt obeys and is desired for 0, variance isDistribution；Enable H=F1-F2, by
It is unrelated with article in chest in Gaussian noise, it can be considered that G and H are independent from each other, are in this way asked to formula (3) both sides the phase
It hopes are as follows:
E(S1-S2)=E (H)+E (G) (4)
By the signal replied for the first label every time, the second label, a sampled points up to ten thousand can be collected, because
This, by Chebyshev's law of great number:
In above formula, n indicates sampled point；
The variance of known G is 0, therefore can be obtained:
Therefore, the training set is F1、F2Difference constitute eigenmatrix F；In eigenmatrix F, each behavior
First label, the second label EPC signal it is corresponding subtract each other after obtained feature vector.
Further, after step S14 obtains training set, characteristics extraction is carried out to training set, in order to the ratio of step 2
To process；Specific features value extracting method is as follows:
Wavelet transformation is carried out to each row of eigenmatrix F, approximate characteristic vector is denoted as ca, minutia vector is remembered
For cd, by the approximate characteristic vector c of every rowaThe new eigenmatrix D constitutedm×k′, wherein m is the number of feature vector, and k' is
Feature vector c after wavelet transformationaLength；
To new eigenmatrix Dm×k'In each row vector extract trend vector Vt, and to new eigenmatrix Dm×k'It is whole
Body extracts detail matrices Vd, extraction process is as shown in formula 7:
Vd(j, i)=Dm×k′(j,i)-Vt(i),0<i≤k',0<j≤m
Further, method test set and training set being compared in the step S21 includes:
Step S210 when actually detected, carries out feature to test set according to the method for carrying out characteristics extraction to training set
Value is extracted, and the trend vector V of test set is obtainedt' and detail matrices Vd′；
Step S211 calculates the trend vector V of training set and test sett、Vt' cross-correlation coefficient x；
Step S212, judges whether cross-correlation coefficient x is greater than threshold value T, if more than step S213 being carried out, otherwise, it is determined that knot
Fruit R=0 exports R；
Step S213, by the detail matrices V of training set and test setdAnd Vd' it is put into Naive Bayes Classifier, it obtains pre-
Result R is surveyed, R is exported.
The present invention has following technical characterstic compared with prior art:
1. ultrasonic inspection is primarily used to the irregular deformation such as crack in detection article, and ultrasonic inspection is to object
The influence or damage that product may cause cannot be estimated, therefore it is not suitable for logistics, the environment of storage；And the present invention uses electromagnetic wave
Penetrability article is identified, can detect package in article whether be replaced or take away, further improve logistics, storage
Safety detection measure in environment, has great importance to real life and production；
2.X optical check usually apply with the transport hubs places such as airport, subway, for checking that closed case is intracorporal violated
Article.However, user usually focuses on personal privacy protection very much, it is undesirable that article quilt in chest in many logistics environments
Stranger checks.Also, since delivery side and recipient are often different user group, by checking article in chest
The mode of concrete condition can not confirm whether article is replaced in chest.And due to the constant magnitude of x-ray examination machine, institute
Not to be suitable for the environment such as logistics, storage yet.Therefore the present invention can be good at solving the above problems, and identification effect is effective and can
It leans on.
Detailed description of the invention
Fig. 1 is the flow chart of embodiment in the present invention；
Fig. 2 is the setting schematic diagram of signal acquisition module in Fig. 1；
Fig. 3 is influence of all kinds of articles to electromagnetic signal, wherein (a) is carton, is (b) empty van, (c) is fiber system
Product are (d) spitball (filler)；
Fig. 4 is EPC first kind second generation agreement flow chart；
Fig. 5 is K-means classifier result, wherein variance and average value relationship that (a) is Alien 964X, (b) are sound
The variance and average value relationship of frequency outstanding person EH47 is (c) audio outstanding person E41C variance and average value relationship, (d) for Alien 964X's
Variance and amplitude relation, (e) variance and amplitude relation for being audio outstanding person EH47, (f) is audio outstanding person E41C variance and amplitude relation；
Fig. 6 is the approximate characteristic vector c of five kinds of unlike materialsaFigure；
Figure label represents: 1-article to be measured, the antenna of 2-general software radio peripheral hardwares, 3-chests, 4-labels
Array, the read antenna of 5-Wireless RF identifiers.
Specific embodiment
Integral Thought of the invention is a pair of of RF tag to be put into chest (package) bottom center, at a distance of 4cm (label
Spacing can be according to the big minor adjustment of cabinet, and chest size used in the present embodiment is 23cm*24cm*40cm), after being put into article,
Trained or test data is acquired in case top, since the dielectric constant of different materials is different, electromagnetic wave is penetrating different articles
When suffered influence it is also different, by the data characteristics of contrastive test collection and training set, come judge to wrap up interior article whether by
It replaces or takes away, specific method is as follows:
A kind of article identification method using electromagnetic wave penetrability, comprising the following steps:
Step 1, the data training of article to be measured
Tag array is attached to the bottom in chest, and article to be measured is put into chest during the training period by step S10
And it is packaged；
Step S11 is provided with Wireless RF identifier and general software radio peripheral hardware outside chest, wherein wirelessly penetrating
The read antenna of frequency identifier is located at the lower section of chest, and the antenna of general software radio peripheral hardware is located at the top of chest；
Step S12 monitors the read antenna and chest interior label array using general software radio peripheral hardware antenna
Communication process, obtain original signal；In this step, general software radio peripheral hardware and wireless radio frequency identification reader are same
When work, obtain the complex signal that label and reader signal penetrate article to be measured in chest；
The EPC signal that tag array in original signal is replied is intercepted out by step S13, and letter is extracted from EPC signal
Number feature, determines label belonging to EPC signal using classifier；According in EPC Global Class1Gen2 to label and
The regulation of reader communication modes and reader command prefix code, label signal needed for intercepting are bad since label signal is too weak
Identification, therefore the information of tag return is judged and intercepted by signal transmitted by identification reader.By EPC Global
Class1Gen2 agreement by the ACK order of reader transmission, and is returned it is found that before the EPC signal of tag return immediately in label
It is QueryRep signal transmitted by reader after multiple EPC signal.Since ack signal and QueryRep signal have fixation
And it is different from the prefix code of other reader signals, therefore, by identifying and looking in original signal transmitted by reader
Ack signal and QueryRep signal, so that it may find and intercept the EPC signal of tag return.
Since the label signal that wireless universal electricity peripheral hardware can obtain contains many noises, and due to penetrating cabinet
A large amount of energy losses caused by middle article, therefore acquired label signal is not easy correctly to decode, it is not recommended that with decoding
Mode determines the label for issuing the signal.But due to label in the production process caused by hardware differences, cause each mark
Replied signal is signed in backscattering with different link frequencies (BLF) and signal characteristic, is incited somebody to action by Fourier transformation
There is the time-domain signal of a large amount of noises to be converted into frequency-region signal, the subtle difference that can be more clearly observed on this link frequency
Not.By K-means classifier, signal clearly can be divided into two classes.In this way, can be accurately by K-means classifier
Judge label belonging to the label signal received in wireless universal electricity peripheral hardware.
In the present embodiment, tag array includes two labels, respectively the first label and the second label.
Step S14 is removed in external environment and environment after the EPC signal alignment of labels different in tag array
The influence of mobile article obtains training set, and saves training set；
Obtain training data detailed process include:
Remember that the first label, the second label obtain corresponding original signal S1、S2It is respectively as follows:
S1=F1+Nen+Neq+Nw (1)
S2=F2+N′en+N′eq+N′w (2)
In above formula, F1、F2Respectively indicate the first label, the signal when signal of the second label penetrates article to be measured in chest
Feature, Nen、N′enIt is influence of the external environment locating for chest to the first label, the second label signal, in external environment
Reflection etc. of the facility to signal；Neq、N′enIt is the influence pair to signal using different equipment, such as different antennas, reader
The influence of first label, the second label signal；Nw、N'wIt is the white Gaussian noise of the first label, the second label signal；
By original signal S1、S2Subtracting each other can obtain:
S1-S2=(F1-F2)+(Nen-N′en)+(Neq-N′eq)+(Nw-N′w) (3)
In above formula, the external environment as locating for chest is similar (label for the first label and the second label
Distance to spacing much smaller than label to external environment facility, therefore think that external environment is considered as the signal reflex of label pair
It is similar), therefore (Nen-N′en)≈0；Also, in same primary test, the equipment that the first label is used with the second label is
It is the same, therefore (Neq-N′eq)≈0；First label, the second label signal white Gaussian noise be independent from each other, enable G=Nw-
N′w；It can then be obtained according to the probability distribution of white Gaussian noiseIt obeys and is desired for 0, variance is
DistributionRespectively the first label, the second label correspond to white Gaussian noise variance of probability distribution)；Enable H=F1-F2,
Since Gaussian noise is unrelated with article in chest, it can be considered that G and H are independent from each other, formula (3) both sides are asked in this way
It is expected that are as follows:
E(S1-S2)=E (H)+E (G) (4)
By the signal replied for the first label every time, the second label, a sampled points up to ten thousand can be collected, because
This, by Chebyshev's law of great number:
In above formula, n indicates sampled point, and ε is any positive number；
The variance of known G is 0, therefore can be obtained:
Therefore, when extracting feature using numerous sampled points, the difference of label pair can be regarded as to signal characteristic difference
Approximation.The training set is F1、F2Difference constitute eigenmatrix F；In eigenmatrix F, each behavior first is marked
It signs, the EPC signal of the second label corresponds to the feature vector obtained after subtracting each other.
For the ease of subsequent comparison process, it can use wavelet transformation here and feature further extracted to eigenmatrix F
Value.Wavelet transformation is a kind of new transform analysis method, it inherits and developed the thought of short time discrete Fourier transform localization, together
When overcome the disadvantages of window size does not change with frequency again, be capable of providing " T/F " window with frequency shift,
It is the ideal tools for carrying out signal time frequency analysis and processing.It is mainly characterized by by transformation can sufficiently outstanding problem it is certain
The feature of aspect can analyze the localization of time (space) frequency, by flexible shift operations to signal (function) gradually into
Row multi-scale refinement is finally reached high frequency treatment time subdivision, and frequency is segmented at low frequency, can adapt to wanting for time frequency signal analysis automatically
It asks, so as to focus on any details of signal.Using wavelet transformation, the trend of signal and fine feature will all can be extracted
Come.Specific features value extracting method is as follows:
Wavelet transformation is carried out to each row of eigenmatrix F, approximate characteristic vector is denoted as ca, minutia vector is remembered
For cd, it is found through experiments that, approximate characteristic vector caIt can preferably reflect the feature of object, therefore, to the every of eigenmatrix F
After one row vector carries out wavelet transformation, by the approximate characteristic vector c of every rowaThe new eigenmatrix D constitutedm×k', wherein m
For the number of feature vector, k' is feature vector c after wavelet transformationaLength；
To new eigenmatrix Dm×k'In each row vector extract trend vector Vt(i) (V is hereafter usedtIndicate), and to new
Eigenmatrix Dm×k'It is whole to extract detail matrices Vd(j, i) (hereafter uses VdIndicate), i, j are variable parameter；Its extraction process is such as
Shown in formula 7:
Vd(j, i)=Dm×k′(j,i)-Vt(i),0<i≤k',0<j≤m
Step 2, the identification of article to be measured
Step S20 when actually detected, is equipped with one in the chest of article to be measured, and the bottom in chest has and step S10
The tag array of same procedure setting；The chest equipped with article to be measured is carried out according to step S11 to the identical method of step S14
Processing, the difference is that finally obtaining test set at this time rather than training set；I.e. this step when the object that handles be one encapsulated
Good chest has article to be measured in chest, and tag array is arranged at the bottom in chest.In the method mentioned in step 1,
Wireless RF identifier and general software radio peripheral hardware are set, keep chest positioned there between, obtains original signal, interception
EPC signal extracts signal characteristic, determines label described in EPC signal；EPC signal alignment processing to different labels, except going
Boundary influences to obtain test set；
Whether test set is compared step S21 with training set, be replaced and destroyed with determination article to be measured, specifically
Method is as follows:
Algorithm input:
(1) when the training stage chest trend vector Vt；
(2) when test phase chest trend vector Vt′；
(3) when the training stage chest detail matrices Vd；
(4) when test phase chest detail matrices Vd′；
(5) test threshold T；
Algorithm output: the judgement result R whether package is destroyed or replaced；
Step S210 when actually detected, carries out feature to test set according to the method for carrying out characteristics extraction to training set
Value is extracted, and the trend vector V of test set is obtainedt' and detail matrices Vd′；That is the step is to be with test set (replacement training set)
Research object carries out processing to test set according to above-mentioned specific features value extracting method to obtain Vt' and Vd′；
Step S211 calculates the trend vector V of training set and test sett、Vt' cross-correlation coefficient x；
Step S212, judges whether cross-correlation coefficient x is greater than threshold value T, if more than step S213 being carried out, otherwise, it is determined that knot
Fruit R=0 exports R；
Step S213, by the detail matrices V of training set and test setdAnd Vd' it is put into Naive Bayes Classifier, it obtains pre-
Result R is surveyed, R is exported.
Wherein, test threshold T can be obtained by specific experiment, or be dynamically determined according to the user demand in actual scene.
The algorithm is first compared the trend phasor of test result twice, and the test set data higher than threshold value think to may be normal
Data wrap up and are not destroyed and replace.Test set data lower than threshold value then directly regard as improper data, i.e., package can
Destruction or replacement can be have passed through.Test set higher than threshold value is further compared detail matrices, with naive Bayesian two-value
Classifier finally determines whether package status is normal.
Embodiment:
Referring to Fig. 1, process of the invention is divided into four big modules, respectively signal acquisition module, preprocessing module, feature
Extraction module and feature comparison module.Whole flow process is further elaborated below:
1. signal acquisition module
Reader antenna referring to fig. 2, is placed in chest bottom, two labels is put by collection process and the system arrangement of signal
Enter chest bottom, cabinet interior is packed into institute's shipped item, and general software radio peripheral hardware antenna is placed in case top, collects general
Software radio peripheral hardware collected label reader communication process input of the signal as preprocessing module.
Referring to Fig. 3, the signal label EPC signal discovery of the invention collected by signal acquisition module is intercepted, is being passed through
In chest after different articles, electromagnetic wave signal can generate different signal characteristics, be in particular on signal be signal decaying,
The formation of burr.
2. preprocessing module
(1) the EPC signal segment of cutting label: signal collected by general software radio peripheral hardware is entire RFID system
System communication process, whole process is referring to fig. 4.Therefore firstly the need of interception cutting label E PC segment as signal characteristic abstraction value
The source of extraction.
(2) label is distinguished using classifier:
Since the label E PC segment being syncopated as is original not decoded signal, firstly the need of distinguishing the segment signal
It is to be issued by which label；Since signal is excessively noisy, decoded mode is easy to appear mistake, therefore utilizes label itself
Hardware differences judge preferably；K-means classifier is selected in this programme, the label E PC segment of interception is carried out discrete
Result after Fourier transformation is distinguished with classifier, distinguishes result referring to Fig. 5；Three classes label is used in the experiment of Fig. 5,
In respectively Alien 964X, Impinj H47 and Impinj E41, Fig. 5, (a) (b) (c) is to set the EPC number of two labels
It is set to an only difference, and (d) (e) (f) is by the entirely different of the EPC number setting of two labels.It can by classification results
It was found that the method for distinguishing label with hardware differences is i.e. accurate and efficient.
(3) it is aligned and subtracts each other label E PC segment: being described in detail eliminate ambient noise, equipment otherness, multipath in front
The method of effects by the EPC signal alignment of two labels and subtracts each other, resulting signal difference vector extracts characteristic value.
3. characteristic extracting module
Characteristic value is extracted to signal difference obtained in preprocessing module, this programme carries out small echo change to signal difference matrix first
Change, extracts approximate characteristic vector ca.The approximate characteristic vector c extracted under unlike materialaTool is very different, in order to illustrate this
Point, uses 5 class experimental materials: metal product (alloy) in this programme, fibre (clothes), carton (paperbox),
Water (water), woodwork (wood) are respectively put into chest, approximate characteristic vector caSignal as shown in fig. 6, its from trend,
The features such as signal strength have apparent difference.
By approximate characteristic vector caNew eigenmatrix D is constituted, two category feature values are extracted to new feature matrix D, respectively
It is trend vector VtWith detail matrices Vd, in the database by the storage of these two types of features, to subsequent comparison.
4. when actually detected, the trend vector sum detail matrices of training set and test set are extracted, using mentioning in step S21
The algorithm arrived carries out Characteristic Contrast, can determine that whether the article in chest is replaced by attacker's malice by comparing result.
Claims (6)
1. a kind of article identification method using electromagnetic wave penetrability, which comprises the following steps:
Step 1, the data training of article to be measured
Tag array is attached to the bottom in chest, and article to be measured is put into chest and is packaged by step S10；
Step S11 is provided with Wireless RF identifier and general software radio peripheral hardware outside chest, and wherein less radio-frequency is known
The read antenna of other device is located at the lower section of chest, and the antenna of general software radio peripheral hardware is located at the top of chest；
Step S12 monitors the logical of the read antenna and chest interior label array using general software radio peripheral hardware antenna
News process obtains original signal；
The EPC signal that tag array in original signal is replied is intercepted out by step S13, and signal spy is extracted from EPC signal
Sign, determines label belonging to EPC signal using classifier；
Step S14 is removed and is moved in external environment and environment after the EPC signal alignment of labels different in tag array
The influence of article obtains training set, and saves training set；
Step 2, the identification of article to be measured
Step S20 when actually detected, is equipped with one in the chest of article to be measured, and the bottom in chest has identical as step S10
The tag array of method setting；According to step S11 to the identical method of step S14 to equipped with article to be measured chest at
Reason, the difference is that finally obtaining test set at this time rather than training set；
Whether test set is compared step S21 with training set, be replaced and destroyed with determination article to be measured.
2. utilizing the article identification method of electromagnetic wave penetrability as described in claim 1, which is characterized in that the label battle array
Column include two labels, respectively the first label and the second label.
3. utilizing the article identification method of electromagnetic wave penetrability as described in claim 1, which is characterized in that the classifier
Using K-means classifier.
4. utilizing the article identification method of electromagnetic wave penetrability as claimed in claim 2, which is characterized in that the step
The detailed process of acquisition training data includes: in S14
Remember that the first label, the second label obtain corresponding original signal S1、S2It is respectively as follows:
S1=F1+Nen+Neq+Nw (1)
S2=F2+N′en+N′eq+N′w (2)
In above formula, F1、F2Respectively indicate the first label, the signal characteristic when signal of the second label penetrates article to be measured in chest,
Nen、N′enIt is influence of the external environment locating for chest to the first label, the second label signal, Neq、N′enIt is to be set using different
The standby influence to the first label, the second label signal, Nw、N′wIt is the white Gaussian noise of the first label, the second label signal；
By original signal S1、S2Subtracting each other can obtain:
S1-S2=(F1-F2)+(Nen-N′en)+(Neq-N′eq)+(Nw-N′w) (3)
In above formula, the external environment as locating for chest is similar, (N for the first label and the second labelen-
N′en)≈0；In same primary test, the equipment that the first label and the second label use is the same, therefore (Neq-N′eq)≈0；
First label, the second label signal white Gaussian noise be independent from each other, enable G=Nw-N′w；Then according to the general of white Gaussian noise
Rate distribution can obtainIt obeys and is desired for 0, variance isDistribution；Enable H=F1-F2, due to height
This noise is unrelated with article in chest, it can be considered that G and H are independent from each other, asks expectation to formula (3) both sides in this way
Are as follows:
E(S1-S2)=E (H)+E (G) (4)
By the signal replied for the first label every time, the second label, a sampled points up to ten thousand can be collected, therefore,
By Chebyshev's law of great number:
In above formula, n indicates sampled point, and ε is any positive number；
The variance of known G is 0, therefore can be obtained:
Therefore, the training set is F1、F2Difference constitute eigenmatrix F；In eigenmatrix F, each behavior first
Label, the second label EPC signal it is corresponding subtract each other after obtained feature vector.
5. utilizing the article identification method of electromagnetic wave penetrability as claimed in claim 4, which is characterized in that step S14 is obtained
After training set, characteristics extraction is carried out to training set, in order to the comparison process of step 2；Specific features value extracting method is such as
Under:
Wavelet transformation is carried out to each row of eigenmatrix F, approximate characteristic vector is denoted as ca, minutia vector is denoted as cd,
By the approximate characteristic vector c of every rowaThe new eigenmatrix D constitutedm×k', wherein m is the number of feature vector, and k' is small echo
Feature vector c after transformationaLength；
To new eigenmatrix Dm×k'In each row vector extract trend vector Vt, and to new eigenmatrix Dm×k'Entirety mentions
Take detail matrices Vd, extraction process is as shown in formula 7:
6. utilizing the article identification method of electromagnetic wave penetrability as claimed in claim 5, which is characterized in that the step
The method that test set and training set are compared in S21 includes:
Step S210 when actually detected, carry out characteristic value to test set according to the method for carrying out characteristics extraction to training set and mentions
It takes, obtains the trend vector V of test sett' and detail matrices Vd′；
Step S211 calculates the trend vector V of training set and test sett、Vt' cross-correlation coefficient x；
Step S212, judges whether cross-correlation coefficient x is greater than threshold value T, if more than step S213 being carried out, otherwise, it is determined that result R
=0, export R；
Step S213, by the detail matrices V of training set and test setdAnd Vd' it is put into Naive Bayes Classifier, obtain prediction knot
Fruit R exports R.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610388891.1A CN106056339B (en) | 2016-06-02 | 2016-06-02 | A kind of article identification method using electromagnetic wave penetrability |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610388891.1A CN106056339B (en) | 2016-06-02 | 2016-06-02 | A kind of article identification method using electromagnetic wave penetrability |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106056339A CN106056339A (en) | 2016-10-26 |
| CN106056339B true CN106056339B (en) | 2019-07-12 |
Family
ID=57170102
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610388891.1A Expired - Fee Related CN106056339B (en) | 2016-06-02 | 2016-06-02 | A kind of article identification method using electromagnetic wave penetrability |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106056339B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107122810A (en) * | 2017-03-10 | 2017-09-01 | 杨玉芹 | A kind of label authentic authentication method based on RFID physical layer feature |
| CN111259679B (en) * | 2020-01-16 | 2021-08-13 | 西安交通大学 | Non-binding type article identification method based on radio frequency signal characteristics |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101170312A (en) * | 2006-10-23 | 2008-04-30 | 国际商业机器公司 | Article case with rfid tag and rfid system |
| CN101356543A (en) * | 2005-08-10 | 2009-01-28 | 洛克威尔自动控制技术股份有限公司 | Enhanced controller utilizing RFID technology |
| CN101359357A (en) * | 2007-07-30 | 2009-02-04 | 日电（中国）有限公司 | Label recognizing system, label accessing device and label sequent determining method |
| CN101751581A (en) * | 2009-07-29 | 2010-06-23 | 中国科学院自动化研究所 | System and method for testing radio frequency identification device label data fraud risk |
| CN103106376A (en) * | 2012-11-07 | 2013-05-15 | 无锡成电科大科技发展有限公司 | Method for identifying radio frequency tag |
| CN104268510A (en) * | 2014-09-17 | 2015-01-07 | 西安电子科技大学 | SAR image target recognition method based on non-negative matrix factorization of sparse constraint |
| CN204791086U (en) * | 2015-07-14 | 2015-11-18 | 小驴科技（北京）有限公司 | Case with function is checked to article |
-
2016
- 2016-06-02 CN CN201610388891.1A patent/CN106056339B/en not_active Expired - Fee Related
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101356543A (en) * | 2005-08-10 | 2009-01-28 | 洛克威尔自动控制技术股份有限公司 | Enhanced controller utilizing RFID technology |
| CN101170312A (en) * | 2006-10-23 | 2008-04-30 | 国际商业机器公司 | Article case with rfid tag and rfid system |
| CN101359357A (en) * | 2007-07-30 | 2009-02-04 | 日电（中国）有限公司 | Label recognizing system, label accessing device and label sequent determining method |
| CN101751581A (en) * | 2009-07-29 | 2010-06-23 | 中国科学院自动化研究所 | System and method for testing radio frequency identification device label data fraud risk |
| CN103106376A (en) * | 2012-11-07 | 2013-05-15 | 无锡成电科大科技发展有限公司 | Method for identifying radio frequency tag |
| CN104268510A (en) * | 2014-09-17 | 2015-01-07 | 西安电子科技大学 | SAR image target recognition method based on non-negative matrix factorization of sparse constraint |
| CN204791086U (en) * | 2015-07-14 | 2015-11-18 | 小驴科技（北京）有限公司 | Case with function is checked to article |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106056339A (en) | 2016-10-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Bertoncini et al. | Wavelet fingerprinting of radio-frequency identification (RFID) tags | |
| US20180047128A1 (en) | Event-driven authentication of physical objects | |
| Wang et al. | Verifiable smart packaging with passive RFID | |
| US20110248827A1 (en) | System and Method for Radio-Frequency Fingerprinting as a Security Layer in RFID Devices | |
| CN104751093B (en) | Method and apparatus for obtaining the video identification code that host equipment is shown | |
| CN102024141A (en) | Face recognition method based on Gabor wavelet transform and local binary pattern (LBP) optimization | |
| CN107491799B (en) | A kind of switch gate condition detection method and device based on RFID | |
| Patil et al. | A survey on Iris recognition system | |
| CN106056339B (en) | A kind of article identification method using electromagnetic wave penetrability | |
| US9996716B2 (en) | Modulation index (depth) based grouping, addressing, and fingerprinting RFID tags | |
| Liya et al. | An overview of radio frequency identification systems | |
| US11983957B2 (en) | Irreversible digital fingerprints for preserving object security | |
| CN205334511U (en) | General identity recognition device | |
| Das et al. | Pedestrian detection using dense LDB descriptor combined with HOG | |
| Prasad et al. | Image quality augmented intramodal palmprint authentication | |
| Sethy et al. | A 3-bit chipless RFID tag for IOT applications | |
| Hinders et al. | Classification of RFID tags with wavelet fingerprinting | |
| Choong et al. | Investigation of Features for Classification RFID Reading Between Two RFID Reader in Various Support Vector Machine Kernel Function | |
| US11687746B2 (en) | Methods and apparatuses for determining frequency response of a security tag | |
| Zhang et al. | A fusion iris feature extraction method based on fisher linear discriminant | |
| EP3438885B1 (en) | Wireless identification system and wireless identification method | |
| Weldemedhin | RFID based Anti-theft System for Metropolia UAS Electronics laboratories | |
| EP3989102B1 (en) | Classification of objects in the proximity of an nfc reader device | |
| Jing et al. | RFID access authorization by face recognition | |
| Shen et al. | A mango biological fingerprint anti-counterfeiting method based on Fuzzy C-means clustering |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190712 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |