CN110210373A - A method of trailing behavioral value - Google Patents
A method of trailing behavioral value Download PDFInfo
- Publication number
- CN110210373A CN110210373A CN201910456973.9A CN201910456973A CN110210373A CN 110210373 A CN110210373 A CN 110210373A CN 201910456973 A CN201910456973 A CN 201910456973A CN 110210373 A CN110210373 A CN 110210373A
- Authority
- CN
- China
- Prior art keywords
- matrix
- neural network
- sample
- trailing
- pnn neural
- 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.)
- Pending
Links
- 230000003542 behavioural effect Effects 0.000 title claims abstract description 34
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000013528 artificial neural network Methods 0.000 claims abstract description 68
- 238000012360 testing method Methods 0.000 claims abstract description 30
- 238000012549 training Methods 0.000 claims abstract description 26
- 230000005693 optoelectronics Effects 0.000 claims abstract description 20
- 239000011159 matrix material Substances 0.000 claims description 105
- 238000001514 detection method Methods 0.000 claims description 15
- 230000002860 competitive effect Effects 0.000 claims description 12
- 238000001914 filtration Methods 0.000 claims description 4
- 238000002474 experimental method Methods 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 230000035945 sensitivity Effects 0.000 claims description 3
- 230000017105 transposition Effects 0.000 claims description 3
- 210000004218 nerve net Anatomy 0.000 claims 1
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 230000006399 behavior Effects 0.000 abstract 1
- 230000006870 function Effects 0.000 description 14
- 238000007689 inspection Methods 0.000 description 3
- 238000005303 weighing Methods 0.000 description 2
- 238000003491 array Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
A method of trailing behavioral value.It includes that the detector being made of optoelectronic switch and Measuring light screen is constructed in gate equipment;PNN neural network is constructed, and by experimental data training PNN neural network, until obtaining the PNN neural network for meeting testing requirements;Using detector and PNN neural network to passenger's real-time perfoming trailing behavioral value by gate equipment.The method provided by the invention for trailing behavioral value, which has the following beneficial effects:, to be compared with the traditional method, the method of the present invention accuracy is high, real-time is good, it can accurately detect various trailing behaviors, it can prevent the security risk for illegally passing through gate equipment, effectively accordingly to ensure the access control of high security.
Description
Technical field
The invention belongs to technical field of civil aviation, more particularly to a kind of method for trailing behavioral value.
Background technique
In Civil Aviation System, existing trailing detection technique is mainly used in two class scenes: (1) for leading in gate class equipment
The control of administrative staff, the general quantity by personnel in sense channel, determines trailing when number is greater than 1 people.Personnel's number
The detection of amount can be obtained by sensors such as weighing sensor, infrared array sensor, video camera and opposite type sensor arrays
It takes.(2) personnel in ATM compartment, indoor population surveillance and safety door trail identification, generally by video camera and laser radar come
It identifies the position of people and trails behavior.Trailing detection method based on weighing sensor, adaptability is not high, is influenced by personnel's weight
It is larger.Method based on video camera, is affected by ambient lighting, and reliability is not high.The method of laser radar, higher cost.
Trailing detection method based on infrared array sensor, since perception data is very little, detection accuracy is lower.
Summary of the invention
To solve the above-mentioned problems, the purpose of the present invention is to provide a kind of methods for trailing behavioral value.
In order to achieve the above object, the method provided by the invention for trailing behavioral value includes the following step carried out in order
It is rapid:
Step 1) constructs the detector being made of optoelectronic switch and Measuring light screen in gate equipment;
Step 2) constructs PNN neural network, and by experimental data training PNN neural network, meets detection until obtaining
It is required that PNN neural network;
The PNN neural network pair for meeting testing requirements that step 3) is obtained using the detector and step 2) of step 1) building
Behavioral value is trailed by passenger's real-time perfoming of gate equipment.
The gate equipment includes: early gate, late gate and sense channel;Early gate and late gate are separately positioned on inspection
It surveys on the inside of the front and rear part in channel;The detector includes six groups of optoelectronic switches and one group of Measuring light screen;Wherein, it is located at early gate
Two groups of optoelectronic switches are respectively set on sense channel inner sidewall between front side, early gate and late gate and on rear side of late gate;
Measuring light screen is arranged in the sense channel inner sidewall lower part between early gate and late gate.
Every group of optoelectronic switch is made of the transmitter and receiver being separately positioned on sense channel two sidewalls.
The Measuring light screen is made of the transmitting light curtain and reception light curtain being separately positioned on sense channel two sidewalls.
In step 2), the building PNN neural network, and by experimental data training PNN neural network, until obtaining
The PNN neural networks of testing requirements must be met, and specific step is as follows:
Step 2.1) constructs the PNN neural network being made of input layer, hidden layer and output layer;
Step 2.2) carries out repeatedly typical current experiment on self check door model machine, acquires passenger using Measuring light screen
Leg shelters from the multiframe data of light as sample input matrix, then every frame data analyze every frame as a sample
Moving target number in data forms sample output matrix by these moving target numbers;By sample input matrix and sample
Output matrix forms sample set;
Step 2.3) selects training set of a part as PNN neural network, another part at random from above-mentioned sample set
As the test set of PNN neural network, sample set is grouped with this;
Step 2.4) by training set sample input matrix and sample output matrix be separately input to PNN neural network
In hidden layer and output layer, respectively as the weight of PNN neural network hidden layer and output layer, thus to PNN neural network into
Row training;
Data in test set in sample input matrix are input to frame by frame and above-mentioned have been subjected to trained PNN mind by step 2.5)
Through in network, by forming the moving target matrix number at each moment in pickup area after the PNN neural network recognization, and by the fortune
Moving-target matrix number is compared with the sample output matrix in test set, analyzes accuracy rate;
If the accuracy rate of PNN neural network is not able to satisfy detection after step 2.6) collects test PNN neural network after tested
It is required that being then grouped again to sample set, by the popularization of test set, until the PNN neural network meets testing requirements.
In step 2.1), specific step is as follows for the building PNN neural network:
2.2.1 input layer
The input layer of PNN neural network is by certain data line in the current data matrix of passenger through being formed after transposition,
Referred to as " input matrix ";
2.2.2 hidden layer
Hidden layer is used to calculate the degree of approximation in above-mentioned input matrix and training set between sample input matrix, using diameter
To basic function as the Probability estimate function for calculating similarity between input matrix and sample input matrix, and using calculating Euclidean
The mode of distance calculates the deviation in input matrix and training set between each sample input matrix;
Assuming that certain sample input matrix IW in training set1.1Are as follows:
Input matrix p are as follows:
P=(p1 p2 ... pn) (2)
The then Euclidean distance between sample input matrix and input matrix are as follows:
Again via the operation of radial basis function, the output matrix a of available hidden layer1Are as follows:
Wherein b is bias matrix, for adjusting the sensitivity of probability variation;Bias matrix b takes:
B=(1 1 ... 1)T (5)
2.2.3 output layer
Output layer is a competitive function, in the output matrix a of hidden layer1It needs to make before input competitive function as follows
Processing:
Assuming that certain sample output matrix is Y in training set:
Y=(y1 y2 ... ym)T y∈N+ (6)
So by sample output matrix Y shape at sparse matrix LW2.1It can indicate are as follows:
Sparse matrix LW2.1The characteristics of be one and only one element of every row be 1, remaining element is 0;By sparse square
Battle array LW2.1With the output matrix a of hidden layer1It can obtain inputting the matrix n of competitive function after multiplication2:
And the output a of PNN neural network2It also is exactly matrix n2The maximum probability label obtained after competitive function operation exists
Corresponding element in sample output matrix:
a2=Y { label [max (n2)]} (9)。
In step 3), what the detector and step 2) of the utilization step 1) building obtained meets testing requirements
PNN neural network trails behavioral value to passenger's real-time perfoming by gate equipment, and specific step is as follows:
Step 3.1) after passenger completes authentication, open, and passenger is normal through sense channel, at this point, not by early gate
It is possible to trail the passenger before early gate is closed by gate sense channel by the personnel of certification, passes through detection in passenger
During channel, using on detector six groups of optoelectronic switches and one group of Measuring light screen passenger's passage situation is detected, so
The measurement data of Measuring light screen and all optoelectronic switches is constituted into input data vector afterwards;
Step 3.2) is filtered above-mentioned input data vector by the way of median filtering, is individually made an uproar with removal
Sound point;
Data after denoising are input in the above-mentioned PNN neural network for meeting testing requirements and handle by step 3.3),
Then moving target number is exported;
Step 3.4) is within the judgement period of setting, if the moving target number of output is greater than 3, determines in sense channel 5
Suspicious trailing behavior has occurred, further progress is needed to judge;
Step 3.5) when detected within two adjacent judgement periods suspicious trailing behavior has occurred when, can determine that inspection
It surveys in channel and trailing behavior has occurred, at this moment close late gate and alarm signal is issued by gate equipment, under otherwise continuing
Primary trailing behavioral value.
The judgement cycle set is 1 second.
The method provided by the invention for trailing behavioral value has the following beneficial effects:
It is compared with the traditional method, the method for the present invention accuracy is high, real-time is good, can accurately detect various trailing rows
For the security risk for illegally passing through gate equipment can be prevented, effectively accordingly to ensure the access control of high security.
Detailed description of the invention
Fig. 1 is that gate equipment employed in the method provided by the invention for trailing behavioral value and panel detector structure are three-dimensional
Figure;
Fig. 2 is that detector employed in the method provided by the invention for trailing behavioral value is laid out top view;
Fig. 3 is that detector employed in the method provided by the invention for trailing behavioral value is laid out side view;
Fig. 4 is Measuring light screen occlusion area schematic diagram in the method provided by the invention for trailing behavioral value;
Fig. 5 is detection method flow chart in the method provided by the invention for trailing behavioral value;
Fig. 6 is PNN neural network structure figure.
Specific embodiment
The method provided by the invention for trailing behavioral value is described in detail in the following with reference to the drawings and specific embodiments.
As shown in Figure 1, the method provided by the invention for trailing behavioral value includes the following steps carried out in order:
Step 1) constructs the detection being made of optoelectronic switch 3 and Measuring light screen 4 as shown in Fig. 1-Fig. 3 in gate equipment
Device;
The gate equipment includes: early gate 1, late gate 2 and sense channel 5;Early gate 1 and late gate 2 are set respectively
It sets on the inside of the front and rear part of sense channel 5.
The detector includes six groups of optoelectronic switches 3 and one group of Measuring light screen 4;Wherein, it is located on front side of early gate 1, is preceding
Two groups of optoelectronic switches 3 are respectively set between gate 1 and late gate 2 and on 5 inner sidewall of sense channel of 2 rear side of late gate;It surveys
Amount light curtain 4 is arranged in the 5 inner sidewall lower part of sense channel between early gate 1 and late gate 2.Every group of optoelectronic switch 3 is by dividing
Transmitter 3.1 on 5 two sidewalls of sense channel is not set and receiver 3.2 forms.The Measuring light screen 4 is by being respectively set
Transmitting light curtain 4.1 and reception light curtain 4.2 on sense channel two sidewalls form.
Transmitting light curtain 4.1 and reception light curtain 4.2 in the Measuring light screen 4 is by several evenly distributed photoelectric tubes
Array composition;Practical studies discovery can be captured relatively accurately when at ground 10 to 60cm when Measuring light screen 4 is arranged
To passenger leg information.As shown in figure 4, leg will be sheltered from by transmitting light curtain when passenger enters in sense channel 5
4.1 light issued, " occlusion area " will be recorded by this moment receiving light curtain 4.2.
Step 2) constructs PNN neural network, and by experimental data training PNN neural network, meets detection until obtaining
It is required that PNN neural network;
Specific step is as follows:
Step 2.1) constructs the PNN neural network as shown in FIG. 6 being made of input layer, hidden layer and output layer;
Step 2.2) carries out repeatedly typical current experiment on self check door model machine, acquires passenger using Measuring light screen
Leg shelters from the multiframe data of light as sample input matrix, then every frame data analyze every frame as a sample
Moving target number in data forms sample output matrix by these moving target numbers;By sample input matrix and sample
Output matrix forms sample set;
Step 2.3) selects training set of a part as PNN neural network, another part at random from above-mentioned sample set
As the test set of PNN neural network, sample set is grouped with this;
Step 2.4) by training set sample input matrix and sample output matrix be separately input to PNN neural network
In hidden layer and output layer, respectively as the weight of PNN neural network hidden layer and output layer, thus to PNN neural network into
Row training;
Data in test set in sample input matrix are input to frame by frame and above-mentioned have been subjected to trained PNN mind by step 2.5)
Through in network, by forming the moving target matrix number at each moment in pickup area after the PNN neural network recognization, and by the fortune
Moving-target matrix number is compared with the sample output matrix in test set, analyzes accuracy rate;
If the accuracy rate of PNN neural network is not able to satisfy detection after step 2.6) collects test PNN neural network after tested
It is required that being then grouped again to sample set, by the popularization of test set, until the PNN neural network meets testing requirements.
In step 2.1), the method for the building PNN neural network is as follows:
2.2.1 input layer
The input layer of PNN neural network is by certain data line in the current data matrix of passenger through being formed after transposition,
Referred to as " input matrix ".Since the points of Measuring light screen 4 and optoelectronic switch 3 are 72, so the characteristic of sample is 72, input layer
Neuron number be 72.
2.2.2 hidden layer
Hidden layer is used to calculate degree of approximation in above-mentioned input matrix and training set between sample input matrix, using appointing
The Probability estimate function of meaning calculates " degree of approximation " between input matrix and sample input matrix.The present invention is using radial base letter
Number uses as the Probability estimate function for calculating similarity between input matrix and sample input matrix and calculates Euclidean distance
Mode calculates the deviation in input matrix and training set between each sample input matrix.
Assuming that certain sample input matrix IW in training set1.1Are as follows:
Input matrix p are as follows:
P=(p1 p2 ... pn) (2)
The then Euclidean distance between sample input matrix and input matrix are as follows:
Again via the operation of radial basis function, the output matrix a of available hidden layer1Are as follows:
Wherein b is bias matrix, for adjusting the sensitivity of probability variation.Bias matrix b takes in the present invention:
B=(1 1 ... 1)T (5)
2.2.3 output layer
Output layer is a competitive function, in the output matrix a of hidden layer1It needs to make before input competitive function as follows
Processing:
Assuming that certain sample output matrix is Y in training set:
Y=(y1 y2 ... ym)T y∈N+ (6)
So by sample output matrix Y shape at sparse matrix LW2.1It can indicate are as follows:
Sparse matrix LW2.1The characteristics of be one and only one element of every row be 1, remaining element is 0.By sparse square
Battle array LW2.1With the output matrix a of hidden layer1It can obtain inputting the matrix n of competitive function after multiplication2:
And the output a of PNN neural network2It also is exactly matrix n2The maximum probability label obtained after competitive function operation exists
Corresponding element in sample output matrix:
a2=Y { label [max (n2)]} (9)
The PNN neural network pair for meeting testing requirements that step 3) is obtained using the detector and step 2) of step 1) building
Behavioral value is trailed by passenger's real-time perfoming of gate equipment.
As shown in figure 5, the real-time perfoming trails behavioral value, the specific method is as follows:
Step 3.1) when passenger complete authentication after, early gate 1 open, passenger normal through sense channel 5, at this point,
Unauthenticated personnel are possible to trail the passenger before early gate 1 is closed by gate sense channel 5, pass through in passenger
During sense channel 5, using on detector six groups of optoelectronic switches 3 and one group of Measuring light screen 4 to passenger's passage situation carry out
Then the measurement data of Measuring light screen 4 and all optoelectronic switches 3 is constituted input data vector by detection;
Step 3.2) is filtered above-mentioned input data vector by the way of median filtering, is individually made an uproar with removal
Sound point;
Data after denoising are input in the above-mentioned PNN neural network for meeting testing requirements and handle by step 3.3),
Then moving target number is exported;
Step 3.4) is within the judgement period of setting, if the moving target number of output is greater than 3, determines in sense channel 5
Suspicious trailing behavior has occurred, further progress is needed to judge;
Step 3.5) when detected within two adjacent judgement periods suspicious trailing behavior has occurred when, can determine that inspection
It surveys in channel 5 and trailing behavior has occurred, at this moment close late gate 2 and alarm signal is issued by gate equipment, otherwise continue
Trailing behavioral value next time.
In the present invention, the judgement cycle set is 1 second.
Wherein, filtering de-noising is mainly interference of the noise to recognition result eliminated in input data vector, judges to detect
It is the moving target number and duration that PNN neural network recognization goes out that the foundation of trailing behavior whether occurs in channel 5, if
Meet moving target number greater than 3 and the duration is more than then to be judged within two seconds that trailing behavior has occurred.
Claims (8)
1. a kind of method for trailing behavioral value, it is characterised in that: the method for the described trailing behavioral value include in order into
Capable the following steps:
Step 1) constructs the detector being made of optoelectronic switch (3) and Measuring light screen (4) in gate equipment;
Step 2) constructs PNN neural network, and by experimental data training PNN neural network, meets testing requirements until obtaining
PNN neural network;
The PNN neural network for meeting testing requirements that step 3) is obtained using the detector and step 2) of step 1) building is to passing through
Passenger's real-time perfoming of gate equipment trails behavioral value.
2. the method according to claim 1 for trailing behavioral value, it is characterised in that: before the gate equipment includes:
Gate (1), late gate (2) and sense channel (5);Before early gate (1) and late gate (2) are separately positioned on sense channel (5)
Posterior medial;The detector includes six groups of optoelectronic switches (3) and one group of Measuring light screen (4);Wherein, it is located at early gate (1)
Two are respectively set on sense channel (5) inner sidewall between front side, early gate (1) and late gate (2) and on rear side of late gate (2)
Group optoelectronic switch (3);Measuring light screen (4) is arranged on the inside of the sense channel (5) between early gate (1) and late gate (2)
Wall lower part.
3. the method according to claim 2 for trailing behavioral value, it is characterised in that: every group of optoelectronic switch (3)
It is made of the transmitter (3.1) and receiver (3.2) that are separately positioned on sense channel (5) two sidewalls.
4. the method according to claim 2 for trailing behavioral value, it is characterised in that: the Measuring light screen (4) is by dividing
Transmitting light curtain (4.1) on sense channel two sidewalls is not set and receives light curtain (4.2) composition.
5. the method according to claim 1 for trailing behavioral value, it is characterised in that: in step 2), the building
PNN neural network, and by experimental data training PNN neural network, until obtaining the PNN neural network for meeting testing requirements
Specific step is as follows:
Step 2.1) constructs the PNN neural network being made of input layer, hidden layer and output layer;
Step 2.2) carries out repeatedly typical current experiment on self check door model machine, acquires passenger leg using Measuring light screen
The multiframe data of light are sheltered from as sample input matrix, then every frame data analyze every frame data as a sample
In moving target number, form sample output matrix by these moving target numbers;It is exported by sample input matrix and sample
Matrix forms sample set;
Step 2.3) selects training set of a part as PNN neural network, another part conduct at random from above-mentioned sample set
The test set of PNN neural network, is grouped sample set with this;
Step 2.4) by training set sample input matrix and sample output matrix be separately input to the implicit of PNN neural network
In layer and output layer, respectively as the weight of PNN neural network hidden layer and output layer, thus PNN neural network is instructed
Practice;
Data in test set in sample input matrix are input to frame by frame and above-mentioned have been subjected to trained PNN nerve net by step 2.5)
In network, by forming the moving target matrix number at each moment in pickup area after the PNN neural network recognization, and by the movement mesh
It marks a matrix number to be compared with the sample output matrix in test set, analyzes accuracy rate;
If the accuracy rate of PNN neural network is not able to satisfy detection and wants after step 2.6) collects test PNN neural network after tested
It asks, then sample set is grouped again, by the popularization of test set, until the PNN neural network meets testing requirements.
6. the method according to claim 5 for trailing behavioral value, it is characterised in that: in step 2.1), the structure
Building PNN neural network, specific step is as follows:
2.2.1 input layer
The input layer of PNN neural network is by certain data line after transposition through forming in the current data matrix of passenger, referred to as
" input matrix ";
2.2.2 hidden layer
Hidden layer is used to calculate the degree of approximation in above-mentioned input matrix and training set between sample input matrix, using radial base
Function uses as the Probability estimate function for calculating similarity between input matrix and sample input matrix and calculates Euclidean distance
Mode calculate the deviation in input matrix and training set between each sample input matrix;
Assuming that certain sample input matrix IW in training set1.1Are as follows:
Input matrix p are as follows:
P=(p1 p2 ... pn) (2)
The then Euclidean distance between sample input matrix and input matrix are as follows:
Again via the operation of radial basis function, the output matrix a of available hidden layer1Are as follows:
Wherein b is bias matrix, for adjusting the sensitivity of probability variation;Bias matrix b takes:
B=(1 1 ... 1)T (5)
2.2.3 output layer
Output layer is a competitive function, in the output matrix a of hidden layer1Need to make following processing before input competitive function:
Assuming that certain sample output matrix is Y in training set:
Y=(y1 y2 ... ym)Ty∈N+ (6)
So by sample output matrix Y shape at sparse matrix LW2.1It can indicate are as follows:
Sparse matrix LW2.1The characteristics of be one and only one element of every row be 1, remaining element is 0;By sparse matrix
LW2.1With the output matrix a of hidden layer1It can obtain inputting the matrix n of competitive function after multiplication2:
And the output a of PNN neural network2It also is exactly matrix n2The maximum probability label obtained after competitive function operation is in sample
Corresponding element in output matrix:
a2=Y { label [max (n2)]} (9)。
7. the method according to claim 1 for trailing behavioral value, it is characterised in that: in step 3), the utilization
The PNN neural network for meeting testing requirements that the detector and step 2) of step 1) building obtain is to the passenger for passing through gate equipment
Real-time perfoming trails behavioral value, and specific step is as follows:
Step 3.1) when passenger complete authentication after, early gate (1) open, passenger normal through sense channel (5), at this point,
Unauthenticated personnel are possible to trail the passenger before early gate (1) is closed by gate sense channel (5), in passenger
During sense channel (5), using on detector six groups of optoelectronic switches (3) and one group of Measuring light screen (4) it is logical to passenger
Market condition is detected, and the measurement data of Measuring light screen (4) and all optoelectronic switches (3) is then constituted input data vector;
Step 3.2) is filtered above-mentioned input data vector by the way of median filtering, to remove single noise
Point;
Data after denoising are input in the above-mentioned PNN neural network for meeting testing requirements and handle by step 3.3), then
Export moving target number;
Step 3.4) is within the judgement period of setting, if the moving target number of output is greater than 3, determines hair in sense channel (5)
Suspicious trailing behavior has been given birth to, further progress is needed to judge;
Step 3.5) when detected within two adjacent judgement periods suspicious trailing behavior has occurred when, can determine that detection is logical
Trailing behavior has occurred in road (5), at this moment close late gate (2) and alarm signal is issued by gate equipment, otherwise continues
Trailing behavioral value next time.
8. the method according to claim 7 for trailing behavioral value, it is characterised in that: the judgement cycle set is 1
Second.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910456973.9A CN110210373A (en) | 2019-05-29 | 2019-05-29 | A method of trailing behavioral value |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910456973.9A CN110210373A (en) | 2019-05-29 | 2019-05-29 | A method of trailing behavioral value |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110210373A true CN110210373A (en) | 2019-09-06 |
Family
ID=67789441
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910456973.9A Pending CN110210373A (en) | 2019-05-29 | 2019-05-29 | A method of trailing behavioral value |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110210373A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113240865A (en) * | 2021-04-21 | 2021-08-10 | 安徽源泽智能科技有限公司 | Self-service silver-colored pavilion is prevented following peep-proof integration mechanism and system thereof |
CN114333123A (en) * | 2021-12-13 | 2022-04-12 | 南京熊猫电子股份有限公司 | Gate passage detection method, device and medium based on laser ranging element group |
CN117275128A (en) * | 2023-11-22 | 2023-12-22 | 浙江宇视科技有限公司 | Vehicle intelligent management method and device for vehicle-road cooperation, electronic equipment and medium |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN205031391U (en) * | 2015-10-13 | 2016-02-17 | 河北工业大学 | Road conditions recognition device of power type artificial limb |
CN206292886U (en) * | 2016-11-10 | 2017-06-30 | 康保环亚安全防范技术(北京)有限公司 | A kind of anti-trailing self check equipment |
CN107019501A (en) * | 2017-05-05 | 2017-08-08 | 山东师范大学 | Detection method and system are fallen down based on genetic algorithm and the long-range of probabilistic neural network |
CN107146302A (en) * | 2017-04-21 | 2017-09-08 | 南京理工大学 | A kind of gate based on infrared detection key point passes through recognition methods |
EP3269608A1 (en) * | 2016-06-20 | 2018-01-17 | Trakm8 Ltd | Detection of tailgating situations |
CN108241177A (en) * | 2016-12-26 | 2018-07-03 | 航天信息股份有限公司 | The anti-trailing detecting system and detection method of single transit passage |
CN109300204A (en) * | 2018-08-20 | 2019-02-01 | 新疆民航凯亚信息网络有限责任公司 | Self-service double-gate clearance safety check apparatus and its clearance safety inspection method |
CN109657581A (en) * | 2018-12-07 | 2019-04-19 | 南京高美吉交通科技有限公司 | Urban track traffic gate passing control method based on binocular camera behavioral value |
-
2019
- 2019-05-29 CN CN201910456973.9A patent/CN110210373A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN205031391U (en) * | 2015-10-13 | 2016-02-17 | 河北工业大学 | Road conditions recognition device of power type artificial limb |
EP3269608A1 (en) * | 2016-06-20 | 2018-01-17 | Trakm8 Ltd | Detection of tailgating situations |
CN206292886U (en) * | 2016-11-10 | 2017-06-30 | 康保环亚安全防范技术(北京)有限公司 | A kind of anti-trailing self check equipment |
CN108241177A (en) * | 2016-12-26 | 2018-07-03 | 航天信息股份有限公司 | The anti-trailing detecting system and detection method of single transit passage |
CN107146302A (en) * | 2017-04-21 | 2017-09-08 | 南京理工大学 | A kind of gate based on infrared detection key point passes through recognition methods |
CN107019501A (en) * | 2017-05-05 | 2017-08-08 | 山东师范大学 | Detection method and system are fallen down based on genetic algorithm and the long-range of probabilistic neural network |
CN109300204A (en) * | 2018-08-20 | 2019-02-01 | 新疆民航凯亚信息网络有限责任公司 | Self-service double-gate clearance safety check apparatus and its clearance safety inspection method |
CN109657581A (en) * | 2018-12-07 | 2019-04-19 | 南京高美吉交通科技有限公司 | Urban track traffic gate passing control method based on binocular camera behavioral value |
Non-Patent Citations (3)
Title |
---|
刘发青等: ""基于步态模型的自助登机尾随检测算法", 《电子世界》 * |
董长虹: "《神经网络与应用》", 31 January 2005 * |
袁娜等: "利用平均影响值和概率神经网络的步态识别", 《哈尔滨工程大学学报》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113240865A (en) * | 2021-04-21 | 2021-08-10 | 安徽源泽智能科技有限公司 | Self-service silver-colored pavilion is prevented following peep-proof integration mechanism and system thereof |
CN114333123A (en) * | 2021-12-13 | 2022-04-12 | 南京熊猫电子股份有限公司 | Gate passage detection method, device and medium based on laser ranging element group |
CN117275128A (en) * | 2023-11-22 | 2023-12-22 | 浙江宇视科技有限公司 | Vehicle intelligent management method and device for vehicle-road cooperation, electronic equipment and medium |
CN117275128B (en) * | 2023-11-22 | 2024-02-20 | 浙江宇视科技有限公司 | Vehicle intelligent management method and device for vehicle-road cooperation, electronic equipment and medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110210373A (en) | A method of trailing behavioral value | |
CN110210563A (en) | The study of pattern pulse data space time information and recognition methods based on Spike cube SNN | |
CN101385050B (en) | Method of securing a physical access and access device implementing the method | |
CN102799893B (en) | Method for processing monitoring video in examination room | |
CN104751580B (en) | The recognition methods of distributing optical fiber sensing signal mode and system | |
CN101770644A (en) | Forest-fire remote video monitoring firework identification method | |
Pacifici et al. | Automatic change detection in very high resolution images with pulse-coupled neural networks | |
CN105575034B (en) | A kind of two waveband forest fire protection intelligent monitoring software image processing analysis method | |
CN104720796A (en) | Automatic detecting system and method for epileptic attack time period | |
CN106650584A (en) | Fire flame detection method and system | |
CN105357063A (en) | Cyberspace security situation real-time detection method | |
CN104751185B (en) | SAR image change detection based on average drifting genetic cluster | |
CN107506734A (en) | One kind of groups unexpected abnormality event detection and localization method | |
CN107315993A (en) | A kind of peephole system and its face identification method based on recognition of face | |
US7013245B2 (en) | Method and apparatus for detecting concealed weapons | |
CN110723621B (en) | Device and method for detecting smoking in elevator car based on deep neural network | |
CN106997475A (en) | A kind of insect image-recognizing method based on parallel-convolution neutral net | |
CN106503643A (en) | Tumble detection method for human body | |
CN110503021A (en) | Fire hazard smoke detecting method based on time compression track characteristic identification | |
CN109522876A (en) | Subway station building staircase selection prediction technique and system based on BP neural network | |
Elihos et al. | Comparison of image classification and object detection for passenger seat belt violation detection using NIR & RGB surveillance camera images | |
CN108334902A (en) | A kind of track train equipment room smog fireproof monitoring method based on deep learning | |
CN112883802A (en) | Method for identifying destructive event of pipeline optical fiber vibration safety early warning system | |
CN108921826A (en) | The transmission line of electricity that super-pixel segmentation is combined with deep learning invades object detecting method | |
CN104077571B (en) | A kind of crowd's anomaly detection method that model is serialized using single class |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190906 |
|
RJ01 | Rejection of invention patent application after publication |