CN105701457B - Direct current electromagnetic relay device and its control method based on recognition of face control - Google Patents
Direct current electromagnetic relay device and its control method based on recognition of face control Download PDFInfo
- Publication number
- CN105701457B CN105701457B CN201610009547.7A CN201610009547A CN105701457B CN 105701457 B CN105701457 B CN 105701457B CN 201610009547 A CN201610009547 A CN 201610009547A CN 105701457 B CN105701457 B CN 105701457B
- Authority
- CN
- China
- Prior art keywords
- face
- matrix
- classifier
- direct current
- electromagnetic relay
- 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.)
- Active
Links
Classifications
-
- 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
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
- G06V40/166—Detection; Localisation; Normalisation using acquisition arrangements
-
- 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
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/168—Feature extraction; Face representation
- G06V40/171—Local features and components; Facial parts ; Occluding parts, e.g. glasses; Geometrical relationships
-
- 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
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/172—Classification, e.g. identification
-
- 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
- G06V40/50—Maintenance of biometric data or enrolment thereof
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01H—ELECTRIC SWITCHES; RELAYS; SELECTORS; EMERGENCY PROTECTIVE DEVICES
- H01H47/00—Circuit arrangements not adapted to a particular application of the relay and designed to obtain desired operating characteristics or to provide energising current
- H01H47/002—Monitoring or fail-safe circuits
- H01H47/004—Monitoring or fail-safe circuits using plural redundant serial connected relay operated contacts in controlled circuit
- H01H47/005—Safety control circuits therefor, e.g. chain of relays mutually monitoring each other
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- General Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of direct current electromagnetic relay devices based on recognition of face control, including face recognition module, power supply switch controller, voice announcer, direct current electromagnetic relay, liquid crystal display, face recognition module respectively with liquid crystal display, power switch control module and voice announcer connection, the connection of the low-voltage control circuit of power switch control module and direct current electromagnetic relay, it can achieve the purpose that the on-off for controlling direct current electromagnetic relay circuit by the method for identifying authorized user's real-time face image, it can be simple, quickly, safety, reliably, the control electrical equipment of intelligence.The invention also discloses the control methods of the direct current electromagnetic relay device based on recognition of face control, specific steps are as follows: step 1, collected facial image is pre-processed, step 2, after the eigenface vector pretreatment exported to strong classifier, Angle Contex matrix is recycled to carry out similar face matching to feature vector.
Description
Technical field
The invention belongs to electric control system technical fields, are related to a kind of direct solenoid relay based on recognition of face control
Device device, the invention further relates to the control methods of the direct current electromagnetic relay device.
Background technique
As information technology is permeated to service for life various aspects, industry internet, cloud computing, the new skill of big data and internet
Continuous fusion application between art, along with the rise of Internet of Things promotes industry move towards intelligence, power consumer is to electrically setting
Standby demand also gradually tends to intelligentized side's exhibition.Traditional electromagnetic relay is usually applied in automatic control circuit, is used
Lesser electric current and voltage control have the electrical equipment of larger current and high voltage.If by traditional electromagnetic relay switch
For control electrical aspect special equipment or high-risk equipment, it is easy to the illegal operation for causing layman causes to electricity
The damage of gas equipment, it could even be possible to entire electrical system is made to paralyse.Traditional electromagnetic relay switch is set for controlling special type
Problem unsafe and unreliable existing for standby or high-risk equipment.
Summary of the invention
The purpose of the present invention is to provide a kind of direct current electromagnetic relay devices based on recognition of face control, can pass through
The method of identification authorized user's real-time face image achievees the purpose that the on-off for controlling direct current electromagnetic relay circuit, and can
Simply, control electrical equipment quickly, safe and reliable, intelligent.
The technical solution adopted by the present invention is that a kind of direct current electromagnetic relay device based on recognition of face control, including
Face recognition module, power supply switch controller, voice announcer, direct current electromagnetic relay, liquid crystal display, face recognition module
It is connect respectively with liquid crystal display, power switch control module and voice announcer, power switch control module and direct solenoid
The low-voltage control circuit of relay connects, and can show in real time on a liquid crystal display when face recognition module recognizes user's face
The facial image of user is shown, if power switch control module can export 24V direct current after face recognition module identifies successfully
The low-voltage control circuit of pressure supply direct current electromagnetic relay.
It is another object of the invention to provide the controls of the above-mentioned direct current electromagnetic relay device based on recognition of face control
Method.
Another technical solution of the invention is the controlling party of the direct current electromagnetic relay device based on recognition of face control
Method is specifically implemented according to the following steps:
Step 1, collected facial image is pre-processed, the rectangle for calculating image using Haar wavelet basis function is special
Simultaneously structural classification device is levied, face is detected,
Step 2, after eigenface vector R (D) pretreatment exported to strong classifier, Angle Contex is recycled
Matrix carries out similar face matching to feature vector R (D), improves recognition of face precision, establishes the face number of authorized user in advance
According to library, the authorization human face data for calculating it and establishing then is carried out using Angle Contex matrix to the face characteristic extracted
The similarity in library judges whether it is authorized user's face.
The features of the present invention also characterized in that
Step 1 comprises the concrete steps that:
Step 1.1, rectangular characteristic value is the sum of all pixels gray value and all pixels in white area in black region
The difference of the sum of gray value, using a kind of rectangular characteristic similar to Haar wavelet basis function, the rectangle in detection window is special
Value indicative, rectangular characteristic value can be calculated by the gray value of integral image pixel, and coordinate is (X, Y) pixel in integral image
Gray value is equal to the cumulative of corresponding position upper left side all pixels gray value on original image；
The calculation formula of integral image is as follows:
I (X, Y) is the image after integral, and i (x, y) is original image,
Step 1.2, the integral image I (X, Y) for acquiring pixel is formed the integrogram square of the figure as a sequence
Battle array, the set D of matrix exgenvalue is found according to integrogram matrix,
D={ a1,a2,a3,a4,........am} (1-2)
Step 1.3, in rectangular window, building multistage classifier obtains the classification function of optimal threshold, to matrix character
Value carries out best classification,
Step 1.4, eyes positioning is carried out using binarization of gray value and integral projection method in the human face region detected；With
Point on the basis of eyes coordinates makes two eyes be in same horizontal line, by scale transformation to facial image by translating, rotating
It is unified to carry out size, grayscale normalization is carried out to facial image, passes through histogram equalization, image smoothing, gray scale normalizing
Change etc. calculates the mean value and mean square deviation of facial contour inner region gray scale, carries out greyscale transformation to whole picture facial image, highlights face
Key facial features.
The specific steps of step 1.3 are as follows:
Step 1.3.1 constructs training sample set L:
L={ (D1,K1),(D2,K2),......(Dm,Km)} (1-3)
Wherein D indicates that the matrix characteristic vector of sample, K indicate that training result, m are to train total number of samples, m=1,2,
3,.......n；
The initial probability distribution of step 1.3.2, sample L are sample weights:
N indicates that number of training, the number of iterations of classifier are t=1, and 2,3,4.......T, T is any positive integer, meter
Calculate the weight of normalization sample:
Wherein j=1,2,3 ... .n；
Step 1.3.3 calls weak learning algorithm, one Weak Classifier H of training to the matrix characteristic vector D of each samplej
(Dm), and calculate the error rate ξ of the classifierj
Wherein θjIt is the threshold value of Weak Classifier, pjFor the biasing of classifier, the direction of majorization inequality, fjIt is characterized matrix
Characteristic value,
Step 1.3.4, selection have minimal error rate ξtCorresponding Weak Classifier Ht(Dm) be used as optimal classification device, i.e.,
ξt=min ξj (1-7)
Step 1.3.5 readjusts sample weights according to optimal classification device are as follows:
Wherein m-th of sample correct judgment then em=0 e on the contrarym=1. can increase classifier error sample weight in this way,
Reduce the weight for the correct sample of classifying, the selection of Weak Classifier will pay attention to last mistake classification error when next iteration
Sample,
βtThe coefficient of sample weights is readjusted for optimal classification device；
Step 1.3.6 generates strong classifier:
IfThen classifier output result is 1, then it is assumed that the detection window may be face area
Domain, R (D) are the eigenface vector of output；
Most of non-face regions can be excluded using the multistage classifier of construction, detect almost possible face area
Domain；If every first-level class device output is 1, then it is assumed that the detection window may be human face region, and the window that will test is input to
Next stage classifier continues to judge, otherwise, will test window as non-face region in the grade and exclude.
Step 2 specifically:
When extracting face characteristic, it is identical for belonging to a part of human face characteristic point distance, and the difference of face
It is different to be mainly manifested in face each section contour line curvature, the difference of angle, 8 degree of drawing as face characteristic average distance can be used
Point, 45 regions will be divided by the plane of coordinate origin of datum mark, generate Angle Contex matrix, calculating falls into each area
The number of other discrete points in domain,
It is specifically implemented according to the following steps:
Step 2.1, on the eigenface obtained by strong classifier, with nose, corners of the mouth midpoint, point, left eyebrow, a left side
Canthus, the left corners of the mouth, right eyebrow, right eye angle, the right corners of the mouth are that 9 character references points select each datum mark, are to draw with 8 degree of angles
Divide interval to choose 45 regions (bin), 9*45 Angle Contex matrix thus can be obtained,
Step 2.2, it is calculated using angular histogram intersection method and 9*45 obtained Angle Contex matrix to be matched
The similarity of face and target user's face, calculation formula are as follows:
P (s, v) is the intersection distance of two feature vectors, and M (R, R') is each division region (bin) in two histograms
The ratio of shared 9*45 overall area (bin) pixel,
Wherein, Rl,gIndicate the Angle Contex matrix of face to be matched, R'l,gIndicate the Angle of target user's face
Contex matrix, l=1,2 ..., 9 indicate 9 character references points；G=1,2,3,4,5,6 ..., 44,45 indicate with 8 degree
Angle is to divide interval to choose 45 regions；
Step 2.3, the value for calculating M (R, R') can be between 0~1, and taking M (R, R') maximum value is just the highest phase of matching degree
Like face,
Step 2.4, after similar face successful match, voice announcer 3 can broadcast the voice of " identifying successfully ", otherwise,
Voice announcer 3 can broadcast the voice of " identification mistake ",
Wherein human face similarity degree matches, on obtained eigenface image, with nose, corners of the mouth midpoint, point, Zuo Mei
Hair, left eye angle, the left corners of the mouth, right eyebrow, right eye angle, the right corners of the mouth are 9 character references points, in the feature obtained by strong classifier
Each datum mark is selected on facial image, is to divide interval to choose 45 regions (bin) with 8 degree of angles, thus be can be obtained
9*45 Angle Contex matrix, as far as possible multizone, which divide, improves matching precision.
The invention has the advantages that equipment used in the implementation present invention is less compared with existing system, structure letter
Single, low in cost, the feasibility of method and safety are higher.Face recognition technology and direct current electromagnetic relay can be made full use of
In conjunction with the effective special equipment controlled in electrical system or high-risk equipment avoid the improper operation of illegal user, for work
Industry, which implements intelligence, has long-range meaning, carries out identification using user to electrical equipment using face recognition technology, assigns
User identity permission remotely controls the control loop of direct current electromagnetic relay, to reach authorized user to special equipment or height
The operation control for equipment of endangering.
Detailed description of the invention
Fig. 1 is a kind of structural schematic diagram of direct current electromagnetic relay based on recognition of face control of the invention；
Fig. 2 is the power on-off control circuit in a kind of direct current electromagnetic relay based on recognition of face control of the invention
Figure；
Fig. 3 is the face recognition module identification in a kind of direct current electromagnetic relay based on recognition of face control of the invention
The flow chart of method.
In figure, 1. face recognition modules, 2. power supply switch controllers, 3. voice announcers, 4. direct current electromagnetic relays, 5.
Liquid crystal display.
Specific embodiment
With reference to the accompanying drawing with specific embodiment, the present invention will be described in detail.
The present invention provides a kind of direct current electromagnetic relay devices based on recognition of face control, as shown in Figure 1, including people
Face identification module 1, power supply switch controller 2, voice announcer 3, direct current electromagnetic relay 4, liquid crystal display 5.
Face recognition module 1 is connect with liquid crystal display 5, power switch control module 2 and voice announcer 3 respectively, electricity
Source switch control module 2 is connect with the low-voltage control circuit of direct current electromagnetic relay 4.When face recognition module 1 recognizes user
Real-time display can go out the facial image of user on liquid crystal display 5 when face, if face recognition module 1 identify successfully after electricity
Source switch control module 2 can export the low-voltage control circuit of 24V DC voltage supply direct current electromagnetic relay 4.
As shown in Fig. 2, recognition of face device acquires facial image using camera, recognition of face device 1 can be installed
Indoors, reach long-range control electromagnetic relay and acquisition clearly video image purpose.When user needs to open electrical equipment
When, direct row to before arriving indoor recognition of face device 1, camera on recognition of face device 1 to the facial information of active user into
Row detection and positioning, using user it will be clear that facial image on liquid crystal display 5.It will using video capture circuit
The vision signal of video camera acquisition is converted into digital signal, and carries out subsequent processing to the digital picture after acquisition.
The power on-off control circuit figure of apparatus of the present invention is as shown in Figure 2, and the course of work is that voice announcer 3 is broadcast
After the voice for reporting " identifying successfully ", the high level that recognition of face device 1 can be 1 to 2 output signal of power supply switch controller,
Triode Q119013 in power supply switch controller will be connected, so that the field-effect tube of enhanced P-channel
The grid of Q7 is in negative potential, the source electrode of field-effect tube Q7 voltage with -24V under the DC voltage effect of 24V, according to field
The working principle of effect pipe Q7, so that the end QX, and the low voltage control for exporting 24V DC voltage conducting direct current electromagnetic relay 4 is returned
The armature on road, relay 4 is inhaled by electromagnet, and for movable contact simultaneously and after stationary contact point contact, authorized user can remote opening electricity
Gas equipment.Otherwise, the voice of the casting of voice announcer 3 " identification mistake ", power supply switch controller 2 export low level, triode
In off state, the grid and source electrode both end voltage of field-effect tube Q7 is 0, and field-effect tube Q7 is in off state, direct current
The low-voltage control circuit of magnetic relay 4 will not be powered, and armature remains static, and the electrical equipment of peripheral circuit fails normal
It opens.
It is a kind of based on recognition of face control direct current electromagnetic relay device control method, as shown in figure 3, specifically according to
Following steps are implemented:
Step 1, collected facial image is pre-processed, the rectangle for calculating image using Haar wavelet basis function is special
Simultaneously structural classification device is levied, face is detected, specific step are as follows:
Step 1.1, rectangular characteristic value is the sum of all pixels gray value and all pixels in white area in black region
The difference of the sum of gray value, using a kind of rectangular characteristic similar to Haar wavelet basis function, the rectangle in detection window is special
Value indicative.Rectangular characteristic value can be calculated by the gray value of integral image pixel, and coordinate is (X, Y) pixel in integral image
Gray value is equal to the cumulative of corresponding position upper left side all pixels gray value on original image；
The calculation formula of integral image is as follows:
I (X, Y) is the image after integral, and i (x, y) is original image,
Step 1.2, the integral image I (X, Y) for acquiring pixel is formed the integrogram square of the figure as a sequence
Battle array, the set D of matrix exgenvalue is found according to integrogram matrix,
D={ a1,a2,a3,a4,........am} (1-2)
Step 1.3, in rectangular window, building multistage classifier obtains the classification function of optimal threshold, to matrix character
Value carries out best classification, and specific algorithm is as follows:
Step 1.3.1 constructs training sample set L:
L={ (D1,K1),(D2,K2),......(Dm,Km)} (1-3)
Wherein D indicates that the matrix characteristic vector of sample, K indicate that training result, m are to train total number of samples, m=1,2,
3,.......n；
The initial probability distribution of step 1.3.2, sample L are sample weights:
N indicates that number of training, the number of iterations of classifier are t=1, and 2,3,4.......T, T is any positive integer, meter
Calculate the weight of normalization sample:
Wherein j=1,2,3 ... .n；
Step 1.3.3 calls weak learning algorithm, one Weak Classifier H of training to the matrix characteristic vector D of each samplej
(Dm), and calculate the error rate ξ of the classifierj
Wherein θjIt is the threshold value of Weak Classifier, pjFor the biasing of classifier, the direction of majorization inequality, fjIt is characterized matrix
Characteristic value.
Step 1.3.4, selection have minimal error rate ξtCorresponding Weak Classifier Ht(Dm) be used as optimal classification device, i.e.,
ξt=min ξj (1-7)
Step 1.3.5 readjusts sample weights according to optimal classification device are as follows:
Wherein m-th of sample correct judgment then em=0 e on the contrarym=1. can increase classifier error sample weight in this way,
Reduce the weight for the correct sample of classifying, the selection of Weak Classifier will pay attention to last mistake classification error when next iteration
Sample.
βtThe coefficient of sample weights is readjusted for optimal classification device；
Step 1.3.6 generates strong classifier:
IfThen classifier output result is 1, then it is assumed that the detection window may be face area
Domain, R (D) are the eigenface vector of output；
Most of non-face regions can be excluded using the multistage classifier of construction, detect almost possible face area
Domain；If every first-level class device output is 1, then it is assumed that the detection window may be human face region, and the window that will test is input to
Next stage classifier continues to judge.Otherwise, it will test window as non-face region in the grade to exclude；
Step 1.4, eyes positioning is carried out using binarization of gray value and integral projection method in the human face region detected；With
Point on the basis of eyes coordinates makes two eyes be in same horizontal line, by scale transformation to facial image by translating, rotating
It is unified to carry out size, grayscale normalization is carried out to facial image, passes through histogram equalization, image smoothing, gray scale normalizing
Change etc. calculates the mean value and mean square deviation of facial contour inner region gray scale, carries out greyscale transformation to whole picture facial image, highlights face
Key facial features；
Step 2, after eigenface vector R (D) pretreatment exported to strong classifier, Angle Contex is recycled
Matrix carries out similar face matching to feature vector R (D), improves recognition of face precision.The face number of authorized user is established in advance
According to library, the authorization human face data for calculating it and establishing then is carried out using Angle Contex matrix to the face characteristic extracted
The similarity in library judges whether it is authorized user's face, specific step are as follows:
When extracting face characteristic, it is identical for belonging to a part of human face characteristic point distance, and the difference of face
It is different to be mainly manifested in face each section contour line curvature, the difference of angle, 8 degree of drawing as face characteristic average distance can be used
Point, 45 regions will be divided by the plane of coordinate origin of datum mark, generate Angle Contex matrix, calculating falls into each area
The number of other discrete points in domain,
Specific algorithm explanation:
Step 2.1, on the eigenface obtained by strong classifier, with nose, corners of the mouth midpoint, point, left eyebrow, a left side
Canthus, the left corners of the mouth, right eyebrow, right eye angle, the right corners of the mouth are that 9 character references points select each datum mark, are to draw with 8 degree of angles
Divide interval to choose 45 regions (bin), 9*45 Angle Contex matrix thus can be obtained；
Step 2.2, it is calculated using angular histogram intersection method and 9*45 obtained Angle Contex matrix to be matched
The similarity of face and target user's face, calculation formula are as follows:
P (s, v) is the intersection distance of two feature vectors, and M (R, R') is each division region (bin) in two histograms
The ratio of shared 9*45 overall area (bin) pixel；
Wherein, Rl,gIndicate the Angle Contex matrix of face to be matched, R'l,gIndicate the Angle of target user's face
Contex matrix, l=1,2 ..., 9 indicate 9 character references points；G=1,2,3,4,5,6 ..., 44,45 indicate with 8 degree
Angle is to divide interval to choose 45 regions；
Step 2.3, the value for calculating M (R, R') can be between 0~1, and taking M (R, R') maximum value is just the highest phase of matching degree
Like face；
Step 2.4, after similar face successful match, voice announcer 3 can broadcast the voice of " identifying successfully ", otherwise,
Voice announcer 3 can broadcast the voice of " identification mistake ".
Its feature of control method of the invention is mainly reflected in the following aspects:
1) face database is established, the face characteristic of authorized user is subjected to registration typing, in order to improve face database
Safety, the data saved in face database can only check and be changed by the user having permission.
2) face image acquisition is carried out to user using video camera, Face datection and positioning is carried out to the image of acquisition, and
Can be to given arbitrary image, the information such as position, size and state for determining whether that there are faces, and providing face.
3) digital signal is converted by the vision signal that video camera acquires using video capture circuit, and to the number after acquisition
Word image is pre-processed.Mainly for the noise in image, illumination is insufficient, size and angle are bad etc. after reasons lead to acquisition
The not high problem of picture quality is extracted and is selected to facial image feature using image enhancement means.
4) recognition of face is carried out, the feature vector in the face feature vector and face database of acquisition is carried out at calculating
Reason, finds out most similar vector, obtains the maximum value of calculated result, being considered as it is the face to be looked for.
5) result that recognition of face goes out is authorized user's face, and voice announcer can broadcast out " identifying successfully " voice letter
It ceases, the high level that face recognition module will be 1 to direct current electromagnetic relay control circuit output signal, in energy supply control module
Triode and field-effect tube conducting, output 24V DC voltage supply direct current electromagnetic relay control circuit, direct solenoid after
Armature is inhaled by electromagnet after electric appliance is powered, movable contact and meanwhile with stationary contact point contact, the electrical equipment of peripheral circuit can be by
Authorized user's starting.
If 6) recognition result is the face of the user of unauthorized, voice announcer can broadcast out " identification mistake " voice letter
Breath, the low level that face recognition module can be 0 to direct current electromagnetic relay control circuit output signal, triode are in cut-off shape
State, the grid and source electrode both end voltage of field-effect tube are 0, and field-effect tube is in off state, the control circuit of electromagnetic relay
Do not turn on, the electrical equipment of peripheral circuit can not authorized user started.
Compared with existing system, equipment used in the implementation present invention is less, and structure is simple, low in cost, method
Feasibility and safety are higher.The combination of face recognition technology and direct current electromagnetic relay can be made full use of, effective control electricity
Special equipment or high-risk equipment in gas system, avoid the improper operation of illegal user, have for industrial implementation intelligence
Long-range meaning carries out identification using user to electrical equipment using face recognition technology, assigns user identity permission and come far
The control loop of process control direct current electromagnetic relay, to reach operation control of the authorized user to special equipment or high-risk equipment
System.
Claims (2)
1. a kind of control method of the direct current electromagnetic relay device based on recognition of face control, is controlled using based on recognition of face
Direct current electromagnetic relay device, including face recognition module (1), power supply switch controller (2), voice announcer (3), direct current
Electromagnetic relay (4), liquid crystal display (5),
The face recognition module (1) respectively with liquid crystal display (5), power switch control module (2) and voice announcer
(3) it connects, the power switch control module (2) is connect with the low-voltage control circuit of direct current electromagnetic relay (4), works as face
Identification module (1) real-time display can go out the facial image of user on liquid crystal display (5) when recognizing user's face, if people
Power switch control module (2) can export 24V DC voltage supply direct current electromagnetic relay after face identification module (1) identifies successfully
(4) low-voltage control circuit,
It is characterized in that, being specifically implemented according to the following steps:
Step 1, collected facial image is pre-processed, the rectangular characteristic of image is calculated simultaneously using Haar wavelet basis function
Structural classification device, detects face,
Step 2, after eigenface vector R (D) pretreatment exported to strong classifier, Angle Contex matrix is recycled
Similar face matching is carried out to feature vector R (D), recognition of face precision is improved, establishes the face database of authorized user in advance,
Then the face characteristic extracted is carried out calculating it and the authorization face database of foundation using Angle Contex matrix
Similarity judges whether it is authorized user's face；
The step 1 comprises the concrete steps that:
Step 1.1, rectangular characteristic value is the sum of all pixels gray value and all pixels gray scale in white area in black region
The difference of the sum of value, using a kind of rectangular characteristic similar to Haar wavelet basis function, rectangular characteristic value in detection window,
Rectangular characteristic value can be calculated by the gray value of integral image pixel, and coordinate is the gray scale of (X, Y) pixel in integral image
Value is equal to the cumulative of corresponding position upper left side all pixels gray value on original image；
The calculation formula of integral image is as follows:
I (X, Y) is the image after integral, and i (x, y) is original image,
Step 1.2, the integral image I (X, Y) for acquiring pixel is formed the integrogram matrix of the figure, root as a sequence
The set D of matrix exgenvalue is found according to integrogram matrix,
D={ a1,a2,a3,a4,........am} (1-2)
Step 1.3, in rectangular window, building multistage classifier obtain the classification function of optimal threshold, to matrix exgenvalue into
The best classification of row,
Step 1.4, eyes positioning is carried out using binarization of gray value and integral projection method in the human face region detected；With eyes
Point on the basis of coordinate makes two eyes be in same horizontal line by translating, rotating, and is carried out by scale transformation to facial image
Size is unified, carries out grayscale normalization to facial image, passes through histogram equalization, image smoothing, gray scale normalization etc.
The mean value and mean square deviation for calculating facial contour inner region gray scale carry out greyscale transformation to whole picture facial image, it is main to highlight face
Facial characteristics；
The specific steps of the step 1.3 are as follows:
Step 1.3.1 constructs training sample set L:
L={ (D1,K1),(D2,K2),......(Dm,Km)} (1-3)
Wherein D indicates that the matrix characteristic vector of sample, K indicate that training result, m are to train total number of samples, m=1,2,
3,.......n；
The initial probability distribution of step 1.3.2, sample L are sample weights:
N indicates that number of training, the number of iterations of classifier are t=1, and 2,3,4.......T, T is any positive integer, and calculating is returned
One changes the weight of sample:
Wherein j=1,2,3 ... .n；
Step 1.3.3 calls weak learning algorithm, one Weak Classifier H of training to the matrix characteristic vector D of each samplej(Dm),
And calculate the error rate ξ of the classifierj
Wherein θjIt is the threshold value of Weak Classifier, pjFor the biasing of classifier, the direction of majorization inequality, fjIt is characterized the spy of matrix
Value indicative,
Step 1.3.4, selection have minimal error rate ξtCorresponding Weak Classifier Ht(Dm) be used as optimal classification device, i.e.,
ξt=min ξj (1-7)
Step 1.3.5 readjusts sample weights according to optimal classification device are as follows:
Wherein then em=0 is reduced by m-th of sample correct judgment otherwise em=1. can increase classifier error sample weight in this way
Classify the weight of correct sample, the selection of Weak Classifier will pay attention to the sample of last wrong classification error when next iteration
This,
βtThe coefficient of sample weights is readjusted for optimal classification device；
Step 1.3.6 generates strong classifier:
IfThen classifier output result is 1, then it is assumed that the detection window may be human face region, R (D)
For the eigenface vector of output；
Most of non-face regions can be excluded using the multistage classifier of construction, detect almost possible human face region；Such as
The every first-level class device output of fruit is 1, then it is assumed that the detection window may be human face region, and the window that will test is input to next stage
Classifier continues to judge, otherwise, will test window as non-face region in the grade and exclude.
2. the control method of the direct current electromagnetic relay device according to claim 1 based on recognition of face control, special
Sign is, the step 2 specifically:
When extracting face characteristic, it is identical for belonging to a part of human face characteristic point distance, and the difference master of face
It shows face each section contour line curvature, the difference of angle, uses 8 degree of divisions as face characteristic average distance, it will be with
Datum mark is that the plane of coordinate origin divides 45 regions, generates Angle Contex matrix, calculates its for falling into each region
The number of his discrete point,
It is specifically implemented according to the following steps:
Step 2.1, on the eigenface obtained by strong classifier, with nose, corners of the mouth midpoint, point, left eyebrow, left eye
Angle, the left corners of the mouth, right eyebrow, right eye angle, the right corners of the mouth are that 9 character references points select each datum mark, are to divide with 8 degree of angles
45 regions (bin) are chosen at interval, and 9*45 Angle Contex matrix thus can be obtained,
Step 2.2, face to be matched is calculated using angular histogram intersection method and 9*45 obtained Angle Contex matrix
With the similarity of target user's face, calculation formula is as follows:
P (s, v) is the intersection distance of two feature vectors, and M (R, R') is that each division region (bin) is shared in two histograms
The ratio of 9*45 overall area (bin) pixel,
Wherein, Rl,gIndicate the Angle Contex matrix of face to be matched, R'l,gIndicate the Angle of target user's face
Contex matrix, l=1,2 ..., 9 indicate 9 character references points；G=1,2,3,4,5,6 ..., 44,45 indicate with 8 degree
Angle is to divide interval to choose 45 regions；
Step 2.3, the value for calculating M (R, R') can be between 0~1, and taking M (R, R') maximum value is just the highest similar people of matching degree
Face,
Step 2.4, after similar face successful match, voice announcer 3 can broadcast the voice of " identifying successfully ", otherwise, voice
Broadcast device 3 can broadcast the voice of " identification mistake ",
Wherein human face similarity degree match, on obtained eigenface image, with nose, corners of the mouth midpoint, point, left eyebrow,
Left eye angle, the left corners of the mouth, right eyebrow, right eye angle, the right corners of the mouth are 9 character references points, in the feature people obtained by strong classifier
Each datum mark is selected in face image, is to divide interval to choose 45 regions (bin) with 8 degree of angles, 9* thus can be obtained
45 Angle Contex matrixes, multizone, which divides, improves matching precision.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610009547.7A CN105701457B (en) | 2016-01-08 | 2016-01-08 | Direct current electromagnetic relay device and its control method based on recognition of face control |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610009547.7A CN105701457B (en) | 2016-01-08 | 2016-01-08 | Direct current electromagnetic relay device and its control method based on recognition of face control |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105701457A CN105701457A (en) | 2016-06-22 |
| CN105701457B true CN105701457B (en) | 2019-09-24 |
Family
ID=56226248
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610009547.7A Active CN105701457B (en) | 2016-01-08 | 2016-01-08 | Direct current electromagnetic relay device and its control method based on recognition of face control |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105701457B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108319829B (en) * | 2017-01-11 | 2022-05-06 | 中兴通讯股份有限公司 | Voiceprint verification method and device |
| CN107358207A (en) * | 2017-07-14 | 2017-11-17 | 重庆大学 | A kind of method for correcting facial image |
| CN109117745B (en) * | 2018-07-23 | 2021-11-09 | 青岛理工大学 | Cloud face recognition and positioning method based on building information model |
| CN109544721A (en) * | 2018-12-04 | 2019-03-29 | 北京诺士诚国际工程项目管理有限公司 | A kind of long-range punch card method and system |
| CN109739101A (en) * | 2018-12-30 | 2019-05-10 | 武汉市新源科创科技有限公司 | Intelligent home control system based on recognition of face |
| CN109816642B (en) * | 2019-01-16 | 2020-03-24 | 尚晓斌 | Compatible instant treatment mechanism |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103902978A (en) * | 2014-04-01 | 2014-07-02 | 浙江大学 | Face detection and identification method |
| CN104574597A (en) * | 2014-12-30 | 2015-04-29 | 苏州福丰科技有限公司 | Face recognition door lock system with lighting function |
| CN104816707A (en) * | 2015-03-25 | 2015-08-05 | 张彬彬 | Intelligent identification type car burglar alarm |
-
2016
- 2016-01-08 CN CN201610009547.7A patent/CN105701457B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103902978A (en) * | 2014-04-01 | 2014-07-02 | 浙江大学 | Face detection and identification method |
| CN104574597A (en) * | 2014-12-30 | 2015-04-29 | 苏州福丰科技有限公司 | Face recognition door lock system with lighting function |
| CN104816707A (en) * | 2015-03-25 | 2015-08-05 | 张彬彬 | Intelligent identification type car burglar alarm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105701457A (en) | 2016-06-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105701457B (en) | Direct current electromagnetic relay device and its control method based on recognition of face control | |
| Islam et al. | Efficient detection and recognition of 3D ears | |
| CN107392112A (en) | A kind of facial expression recognizing method and its intelligent lock system of application | |
| CN109460699A (en) | A kind of pilot harness's wearing recognition methods based on deep learning | |
| CN111079625B (en) | Control method for automatically following rotation of camera along with face | |
| CN104794439A (en) | Real-time approximate frontal face image optimizing method and system based on several cameras | |
| CN103902979B (en) | A kind of face characteristic is extracted and sorting technique | |
| CN109858362A (en) | A kind of mobile terminal method for detecting human face based on inversion residual error structure and angle associated losses function | |
| CN105404867A (en) | Vision-based substation disconnecting switch status recognition method | |
| US20220318369A1 (en) | Face recognition system and method capable of updating registered face templates | |
| CN102831408A (en) | Human face recognition method | |
| CN115063836A (en) | Pedestrian tracking and re-identification method based on deep learning | |
| CN106407904B (en) | A kind of method and device in determining fringe region | |
| Nassu et al. | A computer vision system for monitoring disconnect switches in distribution substations | |
| Byeon et al. | A surveillance system using CNN for face recognition with object, human and face detection | |
| CN113435280A (en) | Testimony verification method | |
| Yun et al. | Fast group verification system for intelligent robot service | |
| CN104751144B (en) | A kind of front face fast appraisement method of facing video monitoring | |
| CN111368803A (en) | Face recognition method and system | |
| Wan et al. | Face detection method based on skin color and adaboost algorithm | |
| Dass et al. | Face detection and synthesis using markov random field models | |
| Bing et al. | Research of face detection based on adaboost and asm | |
| Chen et al. | Research on The Detection Method for Insulation Piercing Connectors and Bolts on The Transmission Line Based on SSD Algorithm | |
| CN111144240A (en) | Image processing method and related equipment | |
| Palanivel et al. | Real time face recognition system using autoassociative neural network models |
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 | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210816 Address after: 710065 No. 11301, block B, Huajing Business Plaza, No. 20, Fenghui South Road, Zhangba street, high tech Zone, Xi'an, Shaanxi Province Patentee after: XI'AN JIN POWER ELECTRICAL Co.,Ltd. Address before: 710048 No. 19 Jinhua South Road, Shaanxi, Xi'an Patentee before: XI'AN POLYTECHNIC University |