CN106780809B - Punch card method based on water purifier - Google Patents
Punch card method based on water purifier Download PDFInfo
- Publication number
- CN106780809B CN106780809B CN201611112071.6A CN201611112071A CN106780809B CN 106780809 B CN106780809 B CN 106780809B CN 201611112071 A CN201611112071 A CN 201611112071A CN 106780809 B CN106780809 B CN 106780809B
- Authority
- CN
- China
- Prior art keywords
- point
- image
- user
- video camera
- feature
- 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
- G07—CHECKING-DEVICES
- G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
- G07C1/00—Registering, indicating or recording the time of events or elapsed time, e.g. time-recorders for work people
- G07C1/10—Registering, indicating or recording the time of events or elapsed time, e.g. time-recorders for work people together with the recording, indicating or registering of other data, e.g. of signs of identity
-
- 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
Abstract
The invention discloses a kind of punch card method based on water purifier, including controller, memory, infrared temperature sensor, the first video camera and for the second video camera of infrared thermal imaging being set on water purifier；APP software on the office computer of user, controller are electrically connected with infrared temperature sensor, the first video camera, the second video camera, memory and server respectively；When user is close to water purifier, the human body signal of infrared temperature sensor detection is obtained；First video camera and the second video camera acquire user images；The database of set of characteristic points and set of keypoints including all users is equipped in memory, controller carries out key point identification and the identification of matching characteristic point and matching, final to identify user and carry out attendance processing.The characteristics of present invention has discrimination height, strong applicability, at low cost, improves administrative convenience.
Description
Technical field
The present invention relates to intelligent identification technology field, more particularly, to it is a kind of be not easy to practise fraud, discrimination is high, base at low cost
In the punch card method of water purifier.
Background technique
Management system of intelligently checking card is the management of the clock in and out record equicorrelated case of the employee of a set of management company
System is the product that software combines with hardware of checking card of checking card, and generally HR department uses, and grasps and the employee for managing enterprise goes out
Diligent dynamic.
Common intelligence punch card system includes fingerprint punch card system and recognition of face punch card system, but there is following lack
Point:
Intelligent human-face identification punch card system, which is directed to similar two people of appearance, effectively to be identified；Recognition of face
Success rate by more multifactor limitation, as: will lead to identification error if when pattern of body form change causes shape of face to change, more restyle the hair and wear
Cap may also lead to recognition failures, needs manpower to carry out data replacement if replacing customer identification information, increases manpower
Cost；
Fingerprint recognition requires finger cleaning, has water stain, greasy dirt all can cause fingerprint that can not identify that fingerprint recognition refers to typing
The fingerprint integrity degree of line has higher requirements, and fingerprint recognition replaceability is higher, has many fingerprint recognition sets can be in the market
Instead of checking card；The higher cost that the existing higher iris recognition of resolution is checked card, is unable to get and is widely popularized.
Summary of the invention
Goal of the invention of the invention is the deficiency at high cost in order to overcome punch card method in the prior art to be easy to practise fraud,
Provide it is a kind of be not easy to practise fraud, discrimination is high, the punch card method at low cost based on water purifier.
To achieve the goals above, the invention adopts the following technical scheme:
A kind of punch card method based on water purifier, including controller, memory, the infrared temperature sensing being set on water purifier
Device, the first video camera and the second video camera for infrared thermal imaging；APP software on the office computer of user, control
Device is electrically connected with infrared temperature sensor, the first video camera, the second video camera, memory and server respectively；Including walking as follows
It is rapid:
Working moment and next moment are equipped in (1-1) memory, when user comes work position, opening computer daily
When, APP software reminds user to go at water purifier to check card；When user, which comes off duty, clicks the key for closing computer, APP software, which is reminded, to be used
Family is gone at water purifier to check card；
(1-2) user every time close to water purifier when, controller obtain infrared temperature sensor detection human body signal；Control
Device controls the first video camera and the second video camera is started to work, and the first video camera and the second video camera acquire user images；
Be equipped in (1-3) memory includes the set of characteristic points of all registration users and the database of set of keypoints, control
Device processed obtains each characteristic point of user from the image that the first video camera is shot, will be in each characteristic point of user and database
The set of characteristic points of all users be compared, select correct matched characteristic point；
Controller obtains each key point of user from the image that the second video camera is shot, by each key point of user
It is compared with the set of keypoints of all users in database, selectes correct matched key point；
(1-4) utilizes formulaCalculate discrimination γ1, wherein n is accumulative correct matched characteristic point and key
The feature sum of point, N are the characteristic point of setting and the sum of key point, and K is the characteristic of each characteristic point and each key point；
Work as γ1> γ, controller make the judgement of successful match, controller find in database with γ1Corresponding user name
Claim, user's name is passed into server, server stores current time, discrimination γ1And user's name；γ is standard identification
Rate；
(1-5) server using each user, check card the time as working for the first time in one day by the identified time, will be every
A user checks card the time last time identified time in one day as coming off duty, by working check card the time, come off duty check card when
Between respectively with working the moment and the next moment compare, calculate whether user is late daily, leaves early and works overtime and store clothes
It is engaged in device.
The function of realizing user identity identification the present invention is based on intelligent water purifier and check card, when user is close to water purifier
When, obtain the human body signal of infrared temperature sensor detection；First video camera and the second video camera acquire user images；Memory
In be equipped with include it is all registration users set of characteristic points and set of keypoints database, controller carry out key point identification and
The identification of matching characteristic point and matching, it is final to identify user and automatic attendance.
The present invention combines identifying system with water purifier, is the market survey based on duration one month, sends out after study
It is existing, the commercial affairs crowd such as employee, white collar be averaged the daily working time it is interior be 2 minutes numbers in the doorway residence time be 6 times, checking card
It is 2 times that the residence time, which is 2 minutes numbers, before machine, and it is 8 times that the residence time, which is 16 minutes numbers, by water purifier, is stopped on station
Time is that 6 hours numbers are 10 times, is found by investigational data, removes the off-the-job before station daily, before water purifier
Dwell times and residence time all account for higher proportion, and the present invention combines intelligent identifying system with water purifier, can effectively increase
Add resolution, reduces the cost of manual maintenance；
The present invention improves the convenience of management work, reduces the cost of manual maintenance, and the user experience is improved spends；Have
Higher resolution can achieve percent 99 recognition accuracy in application range；The present invention presses etc. without user
Deliberately identification operation can realize intelligent recognition in user's water receiving every morning, and it is more convenient to identify；With high resolution
Iris recognition is compared, water purifier intelligent identifying system have the advantages that it is at low cost, have a wide range of application, the company that is more convenient for etc. commercial affairs field
It closes and uses.
Preferably, working as γ1≤ γ, controller make the judgement that the user is nonregistered user；
What controller was shot by each characteristic point obtained from the image that the first video camera is shot and from the second video camera
The each key point obtained in image is sent to server, and server generates the number of a nonregistered user, and by non-note
The number and current time, each characteristic point and each key point associated storage of volume user.
Preferably, range locating for each key point is user face up to hair line, down toward chin minimum point, left and right
To ear edge point；Including 7 regions, 7 regions are respectively forehead region, left eye region, right eye region, nasal area, a left side
Face region, right face region and nose chin area；Crucial point symmetry in left eye region, right eye region is chosen, left face region, the right side
Crucial point symmetry in face region is chosen.
Preferably, each characteristic point is located at face trigonum, characteristic point is 30.
Preferably, the controller obtains each characteristic point of user from the image that the first video camera is shot, will use
Each characteristic point at family is compared with the set of characteristic points of all users in database, selectes correct matched characteristic point packet
Include following steps:
The image I (x, y) that (5-1) shoots the first video camera, using formula G (i)=| [f (i-1, j-1)+f (i-1,
J)+f (i-1, j+1)]-[f (i+1, j-1)+f (i+1, j)+f (i+1, j+1)] | and G (j)=| [f (i-1, j+1)+f (i, j+1)
+ f (i+1, j+1)]-[f (i-1, j-1)+f (i, j-1)+f (i+1, j-1)] | calculate each pixel (i, j) in image I (x, y)
Neighborhood convolution G (i), G (j), set P (i, j)=max [G (i), G (j)], select P (i, j) be image border point；
The image I (x, y) that (5-2) shoots the first video camera, using formula L (x, y, σ)=g (x, y, σ) × I (x,
Y) scale space images L (x, y, σ) being constructed, g (x, y, σ) is scale Gauss variable function,(x, y) is space coordinate, and σ is Image Smoothness；
(5-3) utilizes formula
D (x, y, σ)=(g (x, y, k σ)-g (x, y, σ)) × I (x, y)=L (x, y, k σ)-upper (x, y, σ) calculates Gaussian difference
Divide scale space D (x, y, σ)；K is the constant of adjacent scale space multiple；
For each pixel in image I (x, y), the sub- octave image that s layers length and width halve respectively is successively established,
In, the first straton octave image is original image；
The D (x, y, σ) of D (x, y, σ) pixel adjacent thereto of each pixel is compared by (5-4), if described
When the D (x, y, σ) of pixel is maximum or minimum value in this layer and bilevel every field, take the pixel for spy
Sign point；
(5-5) obtains the dog figure being made of each selected characteristic point, schemes to carry out low-pass filtering to dog；Remove dog figure
Each point except middle marginal point, obtains two-dimentional point diagram；
(5-6) utilizes formula
With θ (x, y)=arc
Tan ((L (x, y+1)-L (x, y-1))/(L (x+1, y)-L (x-1, y))) calculates the modulus value m (x, y) and angle, θ of each characteristic point
(x, y) sets the number of plies of sub- octave image of the scale of each characteristic point as where it；Set modulus value, the angle of each characteristic point
Degree and scale are characterized feature 1, feature 2 and feature 3 a little；The scale of L (x+1, y) characteristic point (x+1, y)；
(5-7) is by the 3 of each characteristic point of set of characteristic points all in 3 features of each characteristic point A1 and database
A feature is compared respectively, and characteristic point B1 most close with A1 and time similar characteristic point C1 are found out in set of characteristic points；
The difference of the feature 1 of characteristic point A1 and B1 is set as a11, set the difference of the feature 1 of characteristic point A1 and C1 as
b11；
The difference of the feature 2 of characteristic point A1 and B1 is set as a12, set the difference of the feature 1 of characteristic point A1 and C1 as
b12；
The difference of the feature 32 of characteristic point A1 and B1 is set as a13, set the difference of the feature 1 of characteristic point A1 and C1 as
b13；
WhenAndAndRatio is the rate threshold of setting；
Then selecting characteristic point B1 is correct match point.
Preferably, controller obtains each key point of user from the image that the second video camera is shot, by user's
Each key point is compared with the set of keypoints of all users in database, and selecting correct matched key point includes such as
Lower step:
(6-1) sets the gray value of (i, j) point in the image that f (i, j) is the shooting of second video camera, is with (i, j) point
The heart takes N ' × N ' window in the picture, sets the point set of pixel composition in window as A ', utilizes formulaIt is filtered, the image g (i, j) after being denoised；
(6-2) is slided on the image with N ' × N ' window, and the gray value of all pixels in window is pressed and rises sequential arrangement,
Take gray value of the gray value for being arranged in middle as window center pixel；
(6-3) utilizes formulaEdge inspection is carried out to image f (x, y)
It surveys, obtains marginal point h (x, y)；
The image f (x, y) that (6-4) shoots the second video camera, using formula L ' (x, y, σ)=g (x, y, σ) × f (x,
Y) scale space images L ' (x, y, σ) being constructed, g (x, y, σ) is scale Gauss variable function,(x, y) is space coordinate, and σ is Image Smoothness；
(6-5) utilizes formula
D ' (x, y, σ)=(g (x, y, k σ)-g (x, y, σ)) × f (x, y)=L ' (x, y, k σ)-L ' (x, y, σ) calculates Gauss
Difference scale space D ' (x, y, σ)；
For each pixel in image f (x, y), the sub- octave image that s layers of length and width halve respectively is successively established, wherein
First straton octave image is original image；
The D ' (x, y, σ) of D ' (x, y, σ) pixel adjacent thereto of each pixel is compared by (6-6), if institute
When the D ' (x, y, σ) for stating pixel is maximum or minimum value in this layer and bilevel each neighborhood, the pixel is taken
For key point；
(6-7) obtains the dog figure being made of each selected key point, schemes to carry out low-pass filtering to dog；Remove dog figure
Each point except middle marginal point, obtains two-dimentional point diagram；
(6-8) utilizes formula
With θ (x, y)=arc
Tan2 ((L (x, y+1)-L (x, y-1))/(L (x+1, y)-L (x-1, y))) calculates the modulus value m (x, y) and angle of each key point
θ (x, y), L (x+1, y) are the scale of key point (x+1, y)；The modulus value, angle and scale of each key point are set as key point
Feature 1, feature 2 and feature 3；
(6-9) is by the 3 of each characteristic point of set of keypoints all in 3 features of each key point A2 and database
A feature is compared respectively, and key point B2 most close with A and time similar key point C2 are found out in set of keypoints；
The difference of the feature 1 of key point A2 and B2 is set as a21, set the difference of the feature 1 of key point A2 and C2 as
b21；
The difference of the feature 2 of key point A2 and B2 is set as a22, set the difference of the feature 1 of key point A2 and C2 as
b22；
The difference of the feature 32 of key point A2 and B2 is set as a23, set the difference of the feature 1 of key point A2 and C2 as
b23；
WhenAndAndRatio is the rate threshold of setting；
Then selecting key point B2 is correct match point.
Preferably, I (x, y) is handled as follows before step (5-1):
Utilize formula R=1.164 (Y-16)+1.596 (Cr-128)
G=1.164 (Y-16) -0.813 (Cr-128) -0.392 (Cb-128)
B=1.164 (Y-16)+2.017 (Cb-128)
The I (x, y) of YCrCb format is converted into rgb color image；
Black white image is converted by rgb color image using formula Gray=0.229R+0.587G+0.11B；Wherein, R is
Red component, G are green component, and B is blue component.
Therefore, high, strong applicability that the invention has the following beneficial effects: discriminations, it is at low cost, improve administrative convenience
Property.
Detailed description of the invention
Fig. 1 is a kind of flow chart of the invention；
Fig. 2 is a kind of set of keypoints figure of the invention；
Fig. 3 is a kind of set of characteristic points figure of the invention.
Specific embodiment
The present invention will be further described with reference to the accompanying drawings and detailed description.
Embodiment as shown in Figure 1 is a kind of punch card method based on water purifier, including be set to water purifier on controller,
Memory, infrared temperature sensor, the first video camera and the second video camera for infrared thermal imaging；Controller respectively with it is infrared
Temperature sensor, the first video camera, the second video camera, memory and server electrical connection；Include the following steps:
Step 100, it checks card prompting
Working moment and next moment are equipped in memory, when user comes work position daily, opens computer, APP
Software reminds user to go at water purifier to check card；When user, which comes off duty, clicks the key for closing computer, APP software reminds user to remove
It checks card at hydrophone；
Step 200, human testing and Image Acquisition
User every time close to water purifier when, controller obtain infrared temperature sensor detection human body signal；Controller control
It makes the first video camera and the second video camera is started to work, the first video camera and the second video camera acquire user images；
Step 300, the identification and matching of characteristic point and key point
In memory be equipped with include it is all registration users set of characteristic points and set of keypoints database, controller from
Each characteristic point that user is obtained in the image of first video camera shooting, will be all in each characteristic point of user and database
The set of characteristic points of user are compared, and select correct matched characteristic point；
Specific step is as follows:
Step 310, controller obtains each characteristic point of user from the image that the first video camera is shot,
The image I (x, y) of first video camera shooting is handled as follows:
Utilize formula R=1.164 (Y-16)+1.596 (Cr-128)
G=1.164 (Y-16) -0.813 (Cr-128) -0.392 (Cb-128)
B=1.164 (Y-16)+2.017 (Cb-128)
The I (x, y) of YCrCb format is converted into rgb color image；
Black white image is converted by rgb color image using formula Gray=0.229R+0.587G+0.11B；Wherein, R is
Red component, G are green component, and B is blue component；
Step 311, for the first video camera shooting image I (x, y), using formula G (i)=| [f (i-1, j-1)+f
(i-1, j)+f (i-1, j+1)]-[f (i+1, j-1)+f (i+1, j)+f (i+1, j+1)] | and G (j)=| [f (i-1, j+1)+f
(i, j+1)+f (i+1, j+1)]-[f (i-1, j-1)+f (i, j-1)+f (i+1, j-1)] | calculate each pixel in image I (x, y)
The neighborhood convolution G (i) of point (i, j), G (j) are set P (i, j)=max [G (i), G (j)], and selecting P (i, j) is image border point；
Step 312, for the image I (x, y) of the first video camera shooting, formula L (x, y, σ)=g (x, y, σ) × I is utilized
(x, y) constructs scale space images L (x, y, σ), and g (x, y, σ) is scale Gauss variable function,(x, y) is space coordinate, and σ is Image Smoothness；
Step 313, formula is utilized
D (x, y, σ)=(g (x, y, k σ)-g (x, y, σ)) × I (x, y)=L (x, y, k σ)-L (x, y, σ) calculates Gaussian difference
Divide scale space D (x, y, σ)；K is the constant of adjacent scale space multiple；
For each pixel in image I (x, y), the sub- octave image that s layers length and width halve respectively is successively established,
In, the first straton octave image is original image；
Step 314, the D (x, y, σ) of D (x, y, σ) pixel adjacent thereto of each pixel is compared, if institute
When the D (x, y, σ) for stating pixel is maximum in this layer and bilevel every field or minimum value, the pixel is taken to be
Characteristic point；
Step 315, the dog figure being made of each selected characteristic point is obtained, dog is schemed to carry out low-pass filtering；Remove dog
Each point in figure except marginal point, obtains two-dimentional point diagram；
Step 316, formula is utilized
With θ (x, y)=arc
Tan ((L (x, y+1)-L (x, y-1))/(L (x+1, y)-L (x-1, y))) calculates the modulus value m (x, y) and angle, θ of each characteristic point
(x, y) sets the number of plies of sub- octave image of the scale of each characteristic point as where it；Set modulus value, the angle of each characteristic point
Degree and scale are characterized feature 1, feature 2 and feature 3 a little；The scale of L (x+1, y) characteristic point (x+1, y)；
Step 317, by each characteristic point of set of characteristic points all in 3 features of each characteristic point A1 and database
3 features be compared respectively, characteristic point B1 most close with A1 and time similar characteristic point C1 are found out in set of characteristic points；
The difference of the feature 1 of characteristic point A1 and B1 is set as a11, set the difference of the feature 1 of characteristic point A1 and C1 as
b11；
The difference of the feature 2 of characteristic point A1 and B1 is set as a12, set the difference of the feature 1 of characteristic point A1 and C1 as
b12；
The difference of the feature 32 of characteristic point A1 and B1 is set as a13, set the difference of the feature 1 of characteristic point A1 and C1 as
b13；
WhenAndAndRatio is the rate threshold of setting；
Then selecting characteristic point B1 is correct match point；
Step 320, key point identification and matching
Controller obtains each key point of user from the image that the second video camera is shot, by each key point of user
It is compared with the set of keypoints of all users in database, selectes correct matched key point；
Specific step is as follows:
Step 321, setting f (i, j) is the gray value of (i, j) point in the image of the second video camera shooting, is with (i, j) point
Center takes N ' × N ' window in the picture, sets the point set of pixel composition in window as A ', utilizes formulaIt is filtered, the image g (i, j) after being denoised；
Step 322, it is slided on the image with N ' × N ' window, the gray value of all pixels in window is arranged by order is risen
Column, take gray value of the gray value for being arranged in middle as window center pixel；
Step 323, formula is utilizedSide is carried out to image f (x, y)
Edge detection, obtains marginal point h (x, y)；
Step 324, for the image f (x, y) of the second video camera shooting, formula L ' (x, y, σ)=g (x, y, σ) × f is utilized
(x, y) constructs scale space images L ' (x, y, σ), and g (x, y, σ) is scale Gauss variable function,(x, y) is space coordinate, and σ is Image Smoothness；
Step 325, formula is utilized
D ' (x, y, σ)=(g (x, y, k σ)-g (x, y, σ)) × f (x, y)=L ' (x, y, k σ)-L ' (x, y, σ) calculates Gauss
Difference scale space D ' (x, y, σ)；
For each pixel in image f (x, y), the sub- octave image that s layers of length and width halve respectively is successively established, wherein
First straton octave image is original image；
Step 326, the D ' (x, y, σ) of D ' (x, y, σ) pixel adjacent thereto of each pixel is compared, if
When the D ' (x, y, σ) of the pixel is maximum or minimum value in this layer and bilevel each neighborhood, the pixel is taken
Point is key point；
Step 327, the dog figure being made of each selected key point is obtained, dog is schemed to carry out low-pass filtering；Remove dog
Each point in figure except marginal point, obtains two-dimentional point diagram；
Step 328, formula is utilized
With θ (x, y)=arc
Tan2 ((L (x, y+1)-L (x, y-1))/(L (x+1, y)-L (x-1, y))) calculates the modulus value m (x, y) and angle of each key point
θ (x, y), L (x+1, y) are the scale of key point (x+1, y)；The modulus value, angle and scale of each key point are set as key point
Feature 1, feature 2 and feature 3；
Step 329, by each characteristic point of set of keypoints all in 3 features of each key point A2 and database
3 features be compared respectively, key point B2 most close with A and time similar key point C2 are found out in set of keypoints；
The difference of the feature 1 of key point A2 and B2 is set as a21, set the difference of the feature 1 of key point A2 and C2 as
b21；
The difference of the feature 2 of key point A2 and B2 is set as a22, set the difference of the feature 1 of key point A2 and C2 as
b22；
The difference of the feature 32 of key point A2 and B2 is set as a23, set the difference of the feature 1 of key point A2 and C2 as
b23；
WhenAndAndRatio is the rate threshold of setting；
Then selecting key point B2 is correct match point.
Step 400, user is identified
Utilize formulaCalculate discrimination γ1, wherein n is the spy of accumulative correct matched characteristic point and key point
Sign sum, N are the characteristic point of setting and the sum of key point, and K is the characteristic of each characteristic point and each key point；γ=
90%；
Work as γ1> γ, controller make the judgement of successful match, controller find in database with γ1Corresponding user name
Claim, user's name is passed into server, server stores current time, discrimination γ1And user's name；
Step 500, attendance is handled
Server using each user, check card the time as working for the first time in one day by the identified time, by each user
The last time identified time checks card the time as coming off duty in one day, and working is checked card time, time difference of checking card of coming off duty
It is compared with working moment and next moment, calculates whether user is late daily, leaves early and works overtime and store in server.
As shown in Fig. 2, range locating for each face key point is user face up to hair line, it is minimum down toward chin
Point, left and right take ear edge point, and four direction is face maximum frame, carry out data sampling in facial maximum frame region, remove
Go characteristic point region take 86 key points (hair line to characteristic area brow portion take 22 key points (by 5 points of upper volume hair line, on
Combine cross to take two o'clock in the middle part of each 5 foreheads of 5 points of eyebrow marginal portion, left and right hair line), left 26 key point of cheek part
(left cheek is by taking 2 points of upper canthus following distance in the middle part of 16 points of ear boundary line, 8 points of side face limit, cheek), right cheek portion
Point taking 26 key points, (right cheek is by taking upper canthus following distance 2 in the middle part of 16 points of ear boundary line, 8 points of side face limit, cheek
Point), chin portion take 12 key points (chin portion by 6 points of chin boundary, 2 points of lip boundary, chengjiang and surrounding 3 points).
As shown in figure 3, human face characteristic point be located at the trigonum that the eyebrow intermediate point of face trigonum two and chengjiang are constituted and
Two boundary of shoulder is to face borderline region, wherein 16 characteristic points of eyes, 4 characteristic points of mouth, 4 characteristic points of nose, forehead
30 points of compositions of 4 characteristic points and face's shoulder.Ratio is 0.4.
It should be understood that this embodiment is only used to illustrate the invention but not to limit the scope of the invention.In addition, it should also be understood that,
After having read the content of the invention lectured, those skilled in the art can make various modifications or changes to the present invention, these etc.
Valence form is also fallen within the scope of the appended claims of the present application.
Claims (7)
1. a kind of punch card method based on water purifier, characterized in that including the controller, memory, infrared being set on water purifier
Temperature sensor, the first video camera and the second video camera for infrared thermal imaging；APP on the office computer of user is soft
Part, controller are electrically connected with infrared temperature sensor, the first video camera, the second video camera, memory and server respectively；Including
Following steps:
Working moment and next moment are equipped in (1-1) memory, when user comes work position daily, opens computer, APP
Software reminds user to go at water purifier to check card；When user, which comes off duty, clicks the key for closing computer, APP software reminds user to remove
It checks card at hydrophone；
(1-2) user every time close to water purifier when, controller obtain infrared temperature sensor detection human body signal；Controller control
It makes the first video camera and the second video camera is started to work, the first video camera and the second video camera acquire user images；
Be equipped in (1-3) memory includes the set of characteristic points of all registration users and the database of set of keypoints, controller
Each characteristic point that user is obtained from the image that the first video camera is shot, by the institute in each characteristic point of user and database
There are the set of characteristic points of user to be compared, selectes correct matched characteristic point；
Controller obtains each key point of user from the image that the second video camera is shot, by each key point and number of user
It is compared according to the set of keypoints of all users in library, selectes correct matched key point；
(1-4) utilizes formulaCalculate discrimination γ1, wherein n is accumulative correct matched characteristic point and key point
Feature sum, N are the characteristic point of setting and the sum of key point, and K is the characteristic of each characteristic point and each key point；
Work as γ1> γ, controller make the judgement of successful match, controller find in database with γ1Corresponding user's name,
User's name is passed into server, server stores current time, discrimination γ1And user's name；γ is standard discrimination；
(1-5) server using each user, check card the time as working for the first time in one day by the identified time, by each use
Family last time identified time in one day checks card the time as coming off duty, by working check card the time, come off duty the time point of checking card
It is not compared with working moment and next moment, calculates whether user is late daily, leaves early and works overtime and store server
In.
2. the punch card method according to claim 1 based on water purifier, characterized in that work as γ1≤ γ, controller make institute
State the judgement that user is nonregistered user；
Controller by each characteristic point obtained from the image that the first video camera is shot and from the second video camera shoot image
Each key point of middle acquisition is sent to server, and server generates the number of a nonregistered user, and by non-registered use
The number and current time, each characteristic point and each key point associated storage at family.
3. the punch card method according to claim 1 based on water purifier, characterized in that range locating for each key point is
User face is up to hair line, down toward chin minimum point, left and right to ear edge point；Including 7 regions, 7 regions are respectively
Forehead region, left eye region, right eye region, nasal area, left face region, right face region and nose chin area；Left eye region,
Crucial point symmetry in right eye region is chosen, and the crucial point symmetry in left face region, right face region is chosen.
4. the punch card method according to claim 1 based on water purifier, characterized in that each characteristic point is located at face triangle
Area, characteristic point are 30.
5. the punch card method according to claim 1 based on water purifier, characterized in that the controller is from the first video camera
Each characteristic point that user is obtained in the image of shooting, by the feature of all users in each characteristic point of user and database
Point set is compared, and selected correct matched characteristic point includes the following steps:
The image I (x, y) that (5-1) shoots the first video camera, using formula G (i)=| [f (i-1, j-1)+f (i-1, j)+f
(i-1, j+1)]-[f (i+1, j-1)+f (i+1, j)+f (i+1, j+1)] | and G (j)=| [f (i-1, j+1)+f (i, j+1)+f (i
+ 1, j+1)]-[f (i-1, j-1)+f (i, j-1)+f (i+1, j-1)] | calculate the neighbour of each pixel (i, j) in image I (x, y)
Domain convolution G (i), G (j) are set P (i, j)=max [G (i), G (j)], and selecting P (i, j) is image border point；
The image I (x, y) that (5-2) shoots the first video camera utilizes formula L (x, y, σ)=g (x, y, σ) × I (x, y) structure
It building scale space images L (x, y, σ), g (x, y, σ) is scale Gauss variable function,
(x, y) is space coordinate, and σ is Image Smoothness；
(5-3) utilizes formula
D (x, y, σ)=(g (x, y, k σ)-g (x, y, σ)) × I (x, y)=L (x, y, k σ)-L (x, y, σ) calculates difference of Gaussian ruler
It spends space D (x, y, σ)；K is the constant of adjacent scale space multiple；
For each pixel in image I (x, y), the sub- octave image that s layers length and width halve respectively is successively established, wherein the
One straton octave image is original image；
The D (x, y, σ) of D (x, y, σ) pixel adjacent thereto of each pixel is compared by (5-4), if the pixel
When the D (x, y, σ) of point is maximum or minimum value in this layer and bilevel every field, the pixel is taken to be characterized
Point；
(5-5) obtains the dog figure being made of each selected characteristic point, schemes to carry out low-pass filtering to dog；Remove side in dog figure
Each point except edge point, obtains two-dimentional point diagram；
(5-6) utilizes formula
With θ (x, y)=arc tan ((L
(x, y+1)-L (x, y-1))/(L (x+1, y)-L (x-1, y))) the modulus value m (x, y) and angle, θ (x, y) of each characteristic point are calculated,
Set the number of plies of sub- octave image of the scale of each characteristic point as where it；Set the modulus value, angle and ruler of each characteristic point
Degree is characterized feature 1, feature 2 and feature 3 a little；The scale of L (x+1, y) characteristic point (x+1, y)；
(5-7) is by 3 spies of each characteristic point of set of characteristic points all in 3 features of each characteristic point A1 and database
Sign is compared respectively, and characteristic point B1 most close with A1 and time similar characteristic point C1 are found out in set of characteristic points；
The difference of the feature 1 of characteristic point A1 and B1 is set as a11, sets the difference of the feature 1 of characteristic point A1 and C1 as b11；
The difference of the feature 2 of characteristic point A1 and B1 is set as a12, sets the difference of the feature 1 of characteristic point A1 and C1 as b12；
The difference of the feature 32 of characteristic point A1 and B1 is set as a13, sets the difference of the feature 1 of characteristic point A1 and C1 as b13；
WhenAndAndRatio is the rate threshold of setting；
Then selecting characteristic point B1 is correct match point.
6. the punch card method according to claim 1 based on water purifier, characterized in that controller is shot from the second video camera
Image in obtain each key point of user, by the crucial point set of all users in each key point of user and database
Conjunction is compared, and selected correct matched key point includes the following steps:
(6-1) set f (i, j) be the second video camera shooting image in (i, j) point gray value, with (i, j) point centered on
N ' × N ' window is taken in image, is set the point set of pixel composition in window as A ', is utilized formulaIt is filtered, the image g (i, j) after being denoised；
(6-2) is slided on the image with N ' × N ' window, and the gray value of all pixels in window is pressed and rises sequential arrangement, the row of taking
It is listed in gray value of the gray value of middle as window center pixel；
(6-3) utilizes formulaEdge detection is carried out to image f (x, y), is obtained
To marginal point h (x, y)；
The image f (x, y) that (6-4) shoots the second video camera utilizes formula L ' (x, y, σ)=g (x, y, σ) × f (x, y) structure
It building scale space images L ' (x, y, σ), g (x, y, σ) is scale Gauss variable function,
(x, y) is space coordinate, and σ is Image Smoothness；
(6-5) utilizes formula
D ' (x, y, σ)=(g (x, y, k σ)-g (x, y, σ)) × f (x, y)=L ' (x, y, k σ)-L ' (x, y, σ) calculates difference of Gaussian
Scale space D ' (x, y, σ)；
For each pixel in image f (x, y), the sub- octave image that s layers of length and width halve respectively is successively established, wherein first
Straton octave image is original image；
The D ' (x, y, σ) of D ' (x, y, σ) pixel adjacent thereto of each pixel is compared by (6-6), if the picture
When the D ' (x, y, σ) of vegetarian refreshments is maximum or minimum value in this layer and bilevel each neighborhood, the pixel is taken to close
Key point；
(6-7) obtains the dog figure being made of each selected key point, schemes to carry out low-pass filtering to dog；Remove side in dog figure
Each point except edge point, obtains two-dimentional point diagram；
(6-8) utilizes formula
With θ (x, y)=arc tan2
((L (x, y+1)-L (x, y-1))/(L (x+1, y)-L (x-1, y))) calculate each key point modulus value m (x, y) and angle, θ (x,
Y), L (x+1, y) is the scale of key point (x+1, y)；The modulus value, angle and scale of each key point are set as the spy of key point
Sign 1, feature 2 and feature 3；
(6-9) is by 3 spies of each characteristic point of set of keypoints all in 3 features of each key point A2 and database
Sign is compared respectively, and key point B2 most close with A and time similar key point C2 are found out in set of keypoints；
The difference of the feature 1 of key point A2 and B2 is set as a21, sets the difference of the feature 1 of key point A2 and C2 as b21；
The difference of the feature 2 of key point A2 and B2 is set as a22, sets the difference of the feature 1 of key point A2 and C2 as b22；
The difference of the feature 32 of key point A2 and B2 is set as a23, sets the difference of the feature 1 of key point A2 and C2 as b23；
WhenAndAndRatio is the rate threshold of setting；
Then selecting key point B2 is correct match point.
7. the punch card method according to claim 5 based on water purifier, characterized in that before the step (5-1) to I (x,
Y) it is handled as follows:
Utilize formula R=1.164 (Y-16)+1.596 (Cr-128)
G=1.164 (Y-16) -0.813 (Cr-128) -0.392 (Cb-128)
B=1.164 (Y-16)+2.017 (Cb-128)
The I (x, y) of YCrCb format is converted into rgb color image；
Black white image is converted by rgb color image using formula Gray=0.229R+0.587G+0.11B；Wherein, R is red
Component, G are green component, and B is blue component.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611112071.6A CN106780809B (en) | 2016-12-06 | 2016-12-06 | Punch card method based on water purifier |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611112071.6A CN106780809B (en) | 2016-12-06 | 2016-12-06 | Punch card method based on water purifier |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106780809A CN106780809A (en) | 2017-05-31 |
| CN106780809B true CN106780809B (en) | 2019-02-15 |
Family
ID=58878462
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611112071.6A Active CN106780809B (en) | 2016-12-06 | 2016-12-06 | Punch card method based on water purifier |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106780809B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110969713A (en) * | 2018-09-30 | 2020-04-07 | 上海小蚁科技有限公司 | Attendance statistics method, device and system and readable storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103984941A (en) * | 2014-06-10 | 2014-08-13 | 深圳市赛为智能股份有限公司 | Face recognition checking-in method and device thereof |
| CN104851140A (en) * | 2014-12-12 | 2015-08-19 | 重庆凯泽科技有限公司 | Face recognition-based attendance access control system |
| CN105718925A (en) * | 2016-04-14 | 2016-06-29 | 苏州优化智能科技有限公司 | Real person living body authentication terminal equipment based on near infrared and facial micro expression |
| CN205384664U (en) * | 2016-03-08 | 2016-07-13 | 广东履安实业有限公司 | Developments face identification entrance guard attendance voice broadcast system |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9695981B2 (en) * | 2012-04-20 | 2017-07-04 | Honeywell International Inc. | Image recognition for personal protective equipment compliance enforcement in work areas |
| GB201402856D0 (en) * | 2014-02-18 | 2014-04-02 | Right Track Recruitment Uk Ltd | System and method for recordal of personnel attendance |
-
2016
- 2016-12-06 CN CN201611112071.6A patent/CN106780809B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103984941A (en) * | 2014-06-10 | 2014-08-13 | 深圳市赛为智能股份有限公司 | Face recognition checking-in method and device thereof |
| CN104851140A (en) * | 2014-12-12 | 2015-08-19 | 重庆凯泽科技有限公司 | Face recognition-based attendance access control system |
| CN205384664U (en) * | 2016-03-08 | 2016-07-13 | 广东履安实业有限公司 | Developments face identification entrance guard attendance voice broadcast system |
| CN105718925A (en) * | 2016-04-14 | 2016-06-29 | 苏州优化智能科技有限公司 | Real person living body authentication terminal equipment based on near infrared and facial micro expression |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106780809A (en) | 2017-05-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11321963B2 (en) | Face liveness detection based on neural network model | |
| CN105359162B (en) | For the pattern mask of the selection and processing related with face in image | |
| CN103353933B (en) | Image recognition apparatus and control method thereof | |
| Hsu et al. | Face detection in color images | |
| CN106056064B (en) | A kind of face identification method and face identification device | |
| US8326001B2 (en) | Low threshold face recognition | |
| JP4377472B2 (en) | Face image processing device | |
| CN105139404B (en) | A kind of the license camera and shooting quality detection method of detectable shooting quality | |
| CN107909057A (en) | Image processing method, device, electronic equipment and computer-readable recording medium | |
| JP6822482B2 (en) | Line-of-sight estimation device, line-of-sight estimation method, and program recording medium | |
| CN107730446A (en) | Image processing method, device, computer equipment and computer-readable recording medium | |
| US7460705B2 (en) | Head-top detecting method, head-top detecting system and a head-top detecting program for a human face | |
| CN104317385A (en) | Gesture identification method and system | |
| US20220043895A1 (en) | Biometric authentication system, biometric authentication method, and program | |
| CN107545536A (en) | The image processing method and image processing system of a kind of intelligent terminal | |
| US11315360B2 (en) | Live facial recognition system and method | |
| CN108416291A (en) | Face datection recognition methods, device and system | |
| US20030007685A1 (en) | Methods and system for illuminant-compensation | |
| JP6956986B1 (en) | Judgment method, judgment device, and judgment program | |
| CN110838119A (en) | Human face image quality evaluation method, computer device and computer readable storage medium | |
| CN111191521B (en) | Face living body detection method and device, computer equipment and storage medium | |
| CN106780809B (en) | Punch card method based on water purifier | |
| JPWO2020213166A1 (en) | Image processing device, image processing method, and image processing program | |
| Long et al. | Near infrared face image quality assessment system of video sequences | |
| CN106780810B (en) | Work attendance method based on water purifier |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |