CN110040592A - Lift car carrying number detection method and system based on the analysis of two-way monitor video - Google Patents
Lift car carrying number detection method and system based on the analysis of two-way monitor video Download PDFInfo
- Publication number
- CN110040592A CN110040592A CN201910298652.0A CN201910298652A CN110040592A CN 110040592 A CN110040592 A CN 110040592A CN 201910298652 A CN201910298652 A CN 201910298652A CN 110040592 A CN110040592 A CN 110040592A
- Authority
- CN
- China
- Prior art keywords
- elevator
- video
- road
- algoritic module
- analysis
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B66—HOISTING; LIFTING; HAULING
- B66B—ELEVATORS; ESCALATORS OR MOVING WALKWAYS
- B66B5/00—Applications of checking, fault-correcting, or safety devices in elevators
- B66B5/0006—Monitoring devices or performance analysers
Landscapes
- Indicating And Signalling Devices For Elevators (AREA)
Abstract
The present invention relates to a kind of lift car carrying number detection methods and system based on the analysis of two-way monitor video.This method implementation are as follows: firstly, outer interior two-way (the I and II) video data transmission of monitoring elevator is to hard disk video recorder；Secondly, hard disk video recorder two-path video is given respectively algoritic module 1 and algoritic module 2 is analyzed；Algoritic module 1 analyzes the road I video, and disengaging carriage total number of persons is calculated according to elevator door opening-closing process duration；Algoritic module 2 analyzes the road II video, and the directivity function of personnel's disengaging is calculated according to foreground area variation before and after elevator door opening-closing；Finally, the calculated result of algoritic module 1 and 2 gives algoritic module 3, algoritic module 3 will pass in and out the product of carriage total number of persons and directivity function as the net number for entering carriage；The calculated result deposit database of algoritic module 3 for analysis or in the overload of the number of discovery, alarm by sending.
Description
Technical field
It is the present invention relates to the application of the intelligent video-image analysis in artificial intelligence technology, in particular to a kind of to be based on two-way
The lift car carrying number detection method and system of monitor video analysis.
Background technique
Elevator is popularized in modern high-rise building.Multiple lift control system must be able to obtain number in lift car
Information.This is because: first, elevator is most basic, most important performance requirement is safety, measures number in carriage, can prevent
It overloads and tramples accident；Second, multiple lift control system needs to obtain number information in carriage, is managed by intelligent dispatching algorithm more
Portion's elevator reduces the energy consumption of elevator to improve the comfort level of user.
Number detection technique is mainly gravity sensor and infrared sensor etc. in traditional lift car.In order to eliminate list
A sensor fault bring hidden danger, industry are desirable to using more advanced reliable carrying number detection technique, or are passed through while being answered
The high reliability of number detection is realized with multiclass detection device.
The method of view-based access control model is increasingly becoming lift car people of greatest concern after gravity sensor, infrared sensor
Number detection methods, because it has the advantages such as easy to implement, precision is high and at low cost.This method is usually pacified in elevator car roof
Video camera is filled, video data is obtained with the visual angle taken a crane shot vertically downward, and count the number of people using certain video analysis method.
Here there are mainly three types of the video analysis algorithms used:
First is that the method based on prospect connected region pixels statistics, core is to establish number and connected region in video image
Relational model between the pixel number of domain is (see paper: unusual checking in land-sea peak lift car based on computer vision
The Hangzhou [master thesis]: Zhejiang Polytechnical University, 2009.).The disadvantage is that being blocked mutually by human somatotype, human body, personnel take
The factors such as band article are affected, and number computational accuracy is relatively low under more human feelings conditions.
Second is that the method based on motion analysis, core is each personnel of detection and tracking, judges that they pass in and out elevator
(see paper: Jin Haiyan, Xiong Qingyu, Wang Kai wait to study instrument based on demographic method in the lift car of image procossing in direction
Instruments and meters journal, 2011,32 (6): 161-165.).The disadvantage is that error is very big when more people pass in and out simultaneously or a people goes out into a people.
Third is that based on identification head feature method, core be head feature extract and detection (see paper: Lu Huchuan,
Zhang Jixia, Zhang Ming repair technique study Journal of System Simulation of the based on Hough transform head detection and tracking, 2008,20 (8):
2127-2132.).The disadvantage is that being easy to happen head erroneous detection when personnel wear dark clothes or belongings, causes to count and miss
Difference.
The above conventional method is other than the shortcomings that having been pointed out, there are one common wretched insufficiency: it is required that video image
It must be that video camera is obtained with viewing angles vertically downward.However, in actual application, in order to protrude video monitoring system
The security protection purpose of system, most of property departments can be mounted on camera elevator car roof corner, and video camera is with certain
Tilt angle observes elevator interior and elevator exit, to obtain maximum monitoring range.If desired pass through analysis monitoring view
Frequency then must additionally dispose a set of video camera with observation visual angle vertically downward again, can thus lead according to number is detected
Cause system cost too high.People highly desirable deployment has the video camera of observation visual angle vertically downward, pacifies it in addition to playing
Outside anti-effect, moreover it is possible to detect lift car carrying number by certain video analysis algorithm.Seldom report is logical in document at present
The monitor video for crossing analysis tilt angle video camera carrys out the technology or device of carrying number in analysis detection lift car.
Serious shielding when there is the difficult point of the monitor video of oblique view angle video camera to be mainly multiple passengers is analyzed, with
And human appearance significant changes etc., the effective information for causing video analysis to obtain are very little.Video monitoring system generally all can be
It is disposed outside elevator to monitor elevator outlet and its nearby video camera (for security protection purpose) in corridor, then inside and outside lift car
Respectively there is an at least video camera that can monitor elevator and its inside and outside human behavior.Help to overcome using the method for Multi-source Information Fusion
The problem of single channel video effective information deficiency.
Summary of the invention
The purpose of the present invention is to provide a kind of lift car carrying number detection methods based on the analysis of two-way monitor video
And system, pass through monitoring outside the monitor video and lift car of the tilted viewing angle video camera shooting of comprehensive analysis lift car
The video data of the video camera of elevator outlet, to solve the problems, such as detection lift car carrying number, and the former has both safety monitoring
Two kinds of purposes are detected with for carrying number.
To achieve the above object, the technical scheme is that a kind of lift car based on the analysis of two-way monitor video
Carrying number detection method, includes the following steps:
Step S1, the video data of outer interior I, II two-way of elevator is monitored, and is transferred to hard disk video recorder；
Step S2, hard disk video recorder respectively gives two-path video algoritic module 1 and algoritic module 2 is analyzed；
Wherein, algoritic module 1 analyzes the road I video data, and disengaging is calculated according to elevator door opening-closing process duration
Carriage total number of persons；Algoritic module 2 analyzes the road II video data, and people is calculated according to foreground area variation before and after elevator door opening-closing
The directivity function of member's disengaging；
Step S3, algoritic module 1 and algoritic module 2 give calculated result to algoritic module 3, and algoritic module 3 is according to disengaging
The product of carriage total number of persons and directivity function obtains the net number into carriage.
In an embodiment of the present invention, the road I video data is acquired by the road the I video camera being deployed in outside lift car
It obtains, the camera supervised elevator outlet in the road I and its neighbouring aisle region；The road II video data is by being deployed in elevator
The inclined road the II video camera acquisition of vertex and visual angle obtains in carriage, the camera supervised elevator inner region in the road II.
In an embodiment of the present invention, in step S2, algoritic module 1 analyzes the road I video data, according to elevator door opening-closing mistake
The implementation that disengaging carriage total number of persons is calculated in the journey duration is as follows:
Step S101, pass through the parameter alpha and β of experimental calibration lift car；
Step S102, by obtaining elevator door opening-closing process duration T to the road I video analysis, and step S101 is obtained
The parameter alpha and β of the lift car arrived substitute into following formula:
| io |=α T+ β
By above formula, can be calculated into carriage personnel and the sum for leaving carriage personnel | io |.
In an embodiment of the present invention, in step S102, continued by obtaining elevator door opening-closing process to the road I video analysis
The specific implementation of time T is as follows:
If foreground pixel number is n in the ROI as caused by noise when elevator door is closed1, when elevator door fully opens in ROI before
Scene prime number is n2, the start/stop time that opening and closing event occurs for elevator is t1、t′1；Before the ROI that then an elevator opening and closing event generates
The waveform available functions n (t of scene prime number variation1,t′1) description:
n(t1,t′1)=n1+(n2-n1)[S(k(t-t1))-S(k(t-t′1))]
In formula, coefficient k is the steep for influencing waveform raising and lowering, and S (t) is Sigmoid function, i.e.,
If occurring to be followed successively by t at the time of elevator opening and closing event in a period1,t1,t2,t2,...,tN,t′N, then before ROI
Scene prime number change curve may be expressed as:
During elevator opens or closes, the waveform that ROI foreground pixel number changes over time exist one rise or under
The slope section of drop, there are maximum value and minimum value, capture prime number is equal to being averaged for maximum value and minimum value for the ordinate of slope section
The starting and ending moment of process is opened and closed at the time of value for elevator cab door, even
n1And n2It can be acquired by the analysis to the road I video data
Searching algorithm is recycled to acquire t1With t '1；
Further, T=t ' is utilized1-t1Elevator door opening-closing process duration T is calculated.
In an embodiment of the present invention, described to acquire t using searching algorithm1With t '1Implementation it is as follows:
In continuous monitoring data sequence (t, n), pixel number n must can be found equal to (n1+n2Two points of)/2, this two
In a point, the lesser point of abscissa is the starting that elevator cab door is opened and closed process, and the biggish point of abscissa is opened for elevator cab door
Close the end of process, i.e. t1Equal to lesser abscissa, t '1Equal to biggish abscissa.
In an embodiment of the present invention, in step S2, algoritic module 2 analyzes the road II video data, according to elevator door opening-closing
The implementation that the directivity function of personnel's disengaging is calculated in foreground area variation in front and back is as follows:
If the foreground area of the road II video image is respectively S (t), S (t') before and after elevator door opening-closing, if moment t foreground facets
Product S (t) is more than preset area threshold value S0, then it is assumed that someone in elevator；If foreground area S (t) is less than preset area threshold value S0, then
Think in elevator nobody；The state for whether being detained personnel in elevator is indicated with function I (t), then
Remember that in is the net number into lift car, is i.e. enters carriage number x (t') in an elevator opening and closing event procedure
And leave the difference of carriage number x (t):
In=x (t')-x (t)
Note is t and t' by the start/stop time that the road I video analysis obtains wherein elevator opening and closing event, in and | io | ratio
Value is directivity function I (t, t'), then
In=I (t, t') | io |
| io | it is into carriage personnel and the sum for leaving carriage personnel；
Due to x (t), x (t') meets following linear equation in two unknowns group
So
Up to directivity function
Thus the net number in into lift car can be calculated.
The present invention also provides a kind of lift car carrying number detection systems based on the analysis of two-way monitor video, comprising:
The road I video camera, is deployed in outside lift car, the outlet of acquisition elevator and its nearby video data in aisle region；
The road II video camera is deployed in vertex in lift car, and visual angle is tilted towards outside elevator, acquires personnel in elevator
Behavior and exception video data；
Hard disk video recorder, the hard disk video recorder are connected with the road I video camera, the road II video camera, the hard disk
Video recorder receives, stores the video data from the road I video camera, the road II video camera；
Algoritic module 1 is connected with the hard disk video recorder, analyzes the road the I video data in the hard disk video recorder, according to
Disengaging carriage total number of persons is calculated in elevator door opening-closing process duration；
Algoritic module 2 is connected with the hard disk video recorder, analyzes the road hard disk video recorder Zhong II video data, root
The directivity function of personnel's disengaging is calculated according to foreground area variation before and after elevator door opening-closing；
Algoritic module 3 is connected with the algoritic module 1, algoritic module 2, receives the algoritic module 1, algoritic module 2
The product for passing in and out carriage total number of persons and directivity function is used as into carriage by analysis as a result, simultaneously further progress convergence analysis
Net number；
Database is connected with the algoritic module 3, stores all previous output of the algoritic module 3 as a result, for data mining
Analysis；
Alarm is connected with the algoritic module 3, issues alarm signal when lift car carrying number transfinites.
Compared to the prior art, the invention has the following advantages:
(1) conventional method require in lift car video camera must according to have the requirement of observation visual angle vertically downward into
Row installation and deployment are difficult to analyze number in carriage if video camera has oblique view angle.The present invention is by introducing elevator
The video data of the outer another way video camera of carriage, and convergence analysis two-path video data, overcome this difficulty, have reached effective
Detect the purpose of carrying number；
(2) video camera of the oblique view angle in lift car is multiplexing functions, can play real time monitoring and peace
Anti- effect, and can play the role of detecting carrying number.Avoiding problems newly increase one for detection carrying number to have vertically
The video camera of downward observation visual angle, substantially reduces hardware cost.
Detailed description of the invention
Fig. 1 is the structural block diagram of the lift car carrying number detection system based on the analysis of two-way monitor video.
Fig. 2 is foreground pixel number change curve in elevator door opening-closing process ROI.
Fig. 3 is the foreground pixel number variation waveform of zero moment elevator opening and closing event.
Fig. 4 is certain elevator opening/closing time and the function (scatter plot) for entering or leaving elevator number.
Fig. 5 is the foreground area change curve of two-way monitor video image when a people enters elevator.
Fig. 6 is the foreground area change curve of two-way monitor video image when more people enter elevator.
Fig. 7 is the foreground area change curve of two-way monitor video image when a people leaves elevator.
Fig. 8 is the foreground area change curve of two-way monitor video image when more people leave elevator.
Fig. 9 is the foreground area change curve of two-way monitor video image when more people enter and leave elevator.
Specific embodiment
With reference to the accompanying drawing, technical solution of the present invention is specifically described.
In general, it is observed that the video camera of lift car has two-way in video monitoring system: be mounted in outside elevator all the way,
It monitors elevator outlet and its nearby the video camera of corridor, its monitor video is denoted as " road I video ", elevator door can be observed
The state change of opening and closing and crowd enter and leave the behavior of elevator；Another way is mounted in apex angle in lift car
Video camera, its monitor video is denoted as " road II video ", can easily differentiate in cage that whether there is or not the states of passenger.Individually
Any of the above camera supervised video all the way is analyzed, is all difficult to achieve the purpose that in carriage that carrying number detects, but pass through fusion point
Two-path video is analysed, then can achieve carrying number testing goal.
Thus the present invention provides a kind of lift car carrying number detection system based on the analysis of two-way monitor video, such as scheme
Shown in 1, core is video big data analysis platform.The platform is mainly made of hard disk video recorder and video analytics server.
Hard disk video recorder is used to receive, stores video data that two-way video camera is shot and be managed to video camera.Video point
Analysis server is for analyzing the video data from hard disk video recorder, it includes three algoritic modules: algoritic module 1 is used
In the analysis road I video data, algoritic module 2 is for analyzing the road II video data, and algoritic module 3 is for merging two-path video data
Analysis as a result, so that carriage carrying number be calculated.
The main working process of lift car carrying number detection system based on the analysis of two-way monitor video is: monitoring elevator
Inside and outside two-way (I and II) video data transmission to hard disk video recorder；Hard disk video recorder send algoritic module 1 two-path video simultaneously
It is analyzed with algoritic module 2；Algoritic module 1 analyze the road I video, according to elevator door opening-closing process duration be calculated into
Carriage total number of persons out；Algoritic module 2 analyzes the road II video, and personnel are calculated according to foreground area variation before and after elevator door opening-closing
The directivity function of disengaging；The calculated result of algoritic module 1 and 2 gives algoritic module 3, and algoritic module 3 will pass in and out carriage total number of persons
Product with directivity function is as the net number for entering carriage；The calculated result deposit database of algoritic module 3 is for data mining
Analysis issues alarm in the overload of the number of discovery.
The working principle of the system is described in detail below.
1, algoritic module 1
Disengaging sedan-chair is calculated according to elevator door opening-closing process duration for analyzing the road I video data in algoritic module 1
Compartment total number of persons.Details are as follows for its working principle:
Area-of-interest (Region ofInterest, ROI) is set by elevator region in the video image of the road I, then
It is easy to calculate the area (the white pixel number i.e. in prospect binary image) for belonging to foreground part in ROI.Generate white pixel
The curves that change over time of number, and it compare and can be found with elevator behavior, the variation of white pixel number and elevator door
There is certain corresponding relationships for opening and closing process:
It is that noise is introduced 1. white pixel number is minimum (theoretically equal to zero) when elevator door is closed；
2. white pixel number gradually rises to maximum value from minimum value in the slow opening procedure of elevator door；
3. white pixel number is maximum when elevator door fully opens, it is approximately equal to the area of elevator door, and maximum value keeps one
The section time；
4. white pixel number gradually falls to minimum value from maximum value during elevator door slowly closing.
It is the changing rule that elevator door is once opened and closed the corresponding foreground pixel number of process above.Through (random after a period of time
Value), elevator opening and closing process also will appear the variation of similar foreground pixel number next time.
As shown in Fig. 2, the variation " record " of the foreground pixel number state change of the opening and closing of elevator door in ROI.
Elevator, which is once opened and closed process, one " pulse " is generated on foreground pixel number curve, the gradient of the rising edge of pulse, failing edge
Depending on elevator door movement velocity.The interval time of adjacent pulse be it is random, be equal to elevator from last opening and closing event under
The time interval of primary opening and closing event.The shape difference of different pulses essentially consists in time width (the i.e. elevator door standard-sized sheet of pulse
Time), it is mainly influenced by entering or leave elevator number, this can be used as the foundation counted indirectly to personnel.
If the time width of three pulses is respectively T in Fig. 21、T2、T3, corresponding number increment is respectively io1、io2、
io3, due to T1< T3< T2, therefore | io1| < | io3| < | io2|。
If foreground pixel number is n in the ROI as caused by noise when elevator door is closed1, when elevator door fully opens in ROI before
Scene prime number is n2, the start/stop time that opening and closing event occurs for elevator is t1、t′1；Before the ROI that then an elevator opening and closing event generates
The waveform available functions n (t of scene prime number variation1,t′1) description:
n(t1,t′1)=n1+(n2-n1)[S(k(t-t1))-S(k(t-t′1))] (1)
In formula, coefficient k is the steep for influencing waveform raising and lowering, and S (t) is Sigmoid function, i.e.,
Assuming that it is t that elevator, which opens the moment,1=0, elevator is opened and closed the duration as T (T=t '1), then the waveform of n (t) has such as
Shape shown in Fig. 3.
If occurring to be followed successively by t at the time of elevator opening and closing event in a period1,t′1,t2,t′2,...,tN,t′N, then ROI
Foreground pixel number change curve may be expressed as:
The waveform of n (t) is as shown in Fig. 2, wherein the duration of jth time elevator opening and closing event is
Tj=t 'j-tj, j=1,2 ..., N (4)
After obtaining the result of video object segmentation, number is monitored by the corresponding ROI foreground pixel number of elevator opening and closing event
According to being easy to acquire n1And n2.During elevator opens or closes, the waveform that ROI foreground pixel number changes over time has one
The slope section risen or fallen, for the ordinate of slope section there are maximum value and minimum value, capture prime number is equal to maximum value and minimum
The starting and ending moment of process is opened and closed at the time of the average value of value for elevator cab door, even
Recycle searching algorithm (in continuous monitoring data sequence (t, n), must can find pixel number n equal to (n1+n2)/
2 two points, the two point in, the lesser point of abscissa be elevator cab door be opened and closed process starting, the biggish point of abscissa
The end of process, i.e. t are opened and closed for elevator cab door1Equal to lesser abscissa, t '1Equal to biggish abscissa) t can be acquired1
With t '1.Further, elevator door opening-closing process duration T is calculated using formula (4).
T is equal to the summation of following four time interval: 1. all entering electricity to owner since stepping into elevator first man
The time interval of door close button is pressed after ladder；2. passing in and out the time interval of elevator opening since elevator door to people；3. being opened from door
Begin to open to completely open time interval；4. from the time interval that door close button to door is begun to shut off is pressed.To some determination
Elevator for, only first time interval be variation, determined by the number of disengaging lift car；Remaining time interval is all
It is fixed and invariable, is determined by the characteristic of elevator itself, and it is unrelated with number.Note | io | it is into carriage personnel and to leave carriage
The sum of personnel, then it is believed that | io | between T have linear relationship:
| io |=α T+ β (6)
Therefore, the course of work of algoritic module 1 is: first passing through the parameter alpha and β of experimental calibration lift car, then passes through the road I
Video analysis obtains elevator door opening-closing process duration T, and substitutes into formula (6), thus be calculated into carriage personnel and from
Open the sum of carriage personnel | io |.
2, algoritic module 2
Algoritic module 2 is calculated for analyzing the road II video data according to foreground area variation before and after elevator door opening-closing
The directivity function of personnel's disengaging.Details are as follows for its working principle:
Based on the road II video, it can easily judge whether be loaded in cage with background subtraction and threshold decision method
The two states of passenger.If moment t foreground area S (t) is more than preset area threshold value S0, then it is assumed that someone in elevator；If prospect
Area S (t) is less than preset area threshold value S0, then it is assumed that in elevator nobody；Indicate whether be detained personnel in elevator with function I (t)
State, then
Remember that in is the net number into lift car, is i.e. enters carriage number x (t') in an elevator opening and closing event procedure
And leave the difference of carriage number x (t):
In=x (t')-x (t) (8)
Note is t and t' by the start/stop time that the road I video analysis obtains wherein elevator opening and closing event, in and | io | ratio
Value is directivity function I (t, t'), then
In=I (t, t') | io | (9)
| io | it is into carriage personnel and the sum for leaving carriage personnel；
Due to x (t), x (t') meets following linear equation in two unknowns group
So
Up to directivity function
In is calculated by joint type (9) and formula (12).
Therefore, the course of work of algoritic module 2 is: first calculating separately the prospect of the road II video image before and after elevator door opening-closing
Area S (t), S (t') recycle formula (12) to calculate directivity function I (t, t').
3, algoritic module 3
3 blending algorithm module 1 and 2 of algoritic module as a result, will pass in and out the product of carriage total number of persons and directivity function as
Into the net number of carriage.Details are as follows for its working principle:
The thinking of two-way video analysis fusion is: obtaining lift car disengaging personnel's total increment, benefit using the road I video analysis
Lift car is obtained with the road II video analysis and passes in and out personnel's directivity function, then using the product of the two as the net entrance of lift car
Number.
Specific step is as follows for the fusion of two-path video analysis:
1) off-line learning elevator characteristic.
A) elevator determined to certain portion arranges Ni=i (i=1,2 ..., 10) is personal successively enter, leave elevator, passes through I
Road video obtains the width TEi and TLi of two impulse waveforms that ROI foreground pixel number changes in the process, obtains point (TEi,
Ni),(TLi,Ni)。
B) it is total three times to repeat step a), according to the data (TE of acquisitioni,Ni),(TLi,Ni) and linear regression algorithm, it obtains
Into or leave linear function of the elevator number about pulse width | io |=α T+ β completes calibration.
2) road I video is analyzed.
By the road I video detection ROI foreground pixel number delta data, T is calculated by formula (4), formula (5).
B) T value is substituted into | io |=α T+ β is calculated entrance or leaves elevator number | io | and approximate rounding.
(3) by the road II video detection elevator be opened and closed event start/stop time t, t', by formula (12) calculate directivity function I (t,
t')。
(4) combine | io | and I (t, t'), the number in for entering elevator by formula (11) calculating is simultaneously approximate to be rounded.
Table 1 gives several examples that in is calculated by the fusion road I, II video analysis result.
Table 1 enters the example of elevator carrying number by merging two-path video Analysis result calculation
The following are a specific embodiments of the invention.
To a certain elevator, first off-line learning elevator characteristic obtains 60 data about number and corresponding time relationship altogether,
As shown in Figure 4.It is using the function that linear regression algorithm obtains matched curve
| io |=0.3472T-1.125 (13)
That is, α=0.3472, β=1.125.
Observation monitors the road I of the elevator, the discovery of the road II video, includes that a people enters elevator, more into elevator number increment
People enters elevator, a people leaves elevator, more people leave elevator and more people pass in and out five seed types such as elevator.Fig. 5, Fig. 6, Fig. 7, Fig. 8,
The foreground area variation that two-way monitor video image under various disengaging elevator number increment type cases is set forth in Fig. 9 is bent
Line.
The ROI foreground area change curve that can be seen that the road I video image from Fig. 5-Fig. 9 has following characteristics:
1. n (t) and n (t') are equal to zero point for various disengaging elevator number increment types.
2. the duration of curve ascent stage and descending branch is of substantially equal for various disengaging elevator number increment types.
3. for various disengaging elevator number increment types, the ordinate of middle flat section is all of substantially equal.
4. entering or leaving the duration that elevator number determines curve middle flat section.
The ROI foreground area change curve that can be seen that the road II video image from Fig. 5-Fig. 9 has following characteristics:
1. entering elevator event for personnel, S (t) is equal to zero point；Elevator event is left to personnel, S (t') is equal to zero point；
To there is personnel to enter and leave elevator event simultaneously, S (t) and S (t') are not equal to zero point.
2. entering or leaving the ordinate that elevator number determines curve middle flat section.
3. the duration of curve ascent stage and descending branch is of substantially equal for various disengaging elevator number increment types.
According to Fig. 5-Fig. 9 and fusion formula, the results are shown in Table 2 for the increment of calculating disengaging elevator personnel.Here with one
People illustrates how to calculate for entering the case where elevator (the case where i.e. in table 2 1. number is) | io |.As shown in Figure 5: n1=
102,n2=8766, then n (t1,t′1)=(n1+n2)/2=4434.It then passes through searching algorithm and obtains t1=1.20, t '1=
7.68 T1=t1'-t1=6.48 by T1Value substitute into formula (6), calculate | io | ≈ 1.It is verified, result shown in table 2 and artificial
The true value that observation video obtains is equal.
The calculated result of the disengaging elevator number increment of table 2
From table 2 it can be seen that more people enter elevator, more people leave elevator and more people enter and leave the three classes situation such as elevator
Under, number increment absolute value is equal to 2, and at this moment I (t, t') can further discriminate between different disengaging types as directivity function.This
Show through fusion I, the road II video analysis as a result, so that it is correct effective for calculating into the method for elevator number increment.
Although the embodiments of the present invention has been shown and described above, it is to be understood that above-described embodiment is example
Property, it is not considered as limiting the invention, those skilled in the art are not departing from the principle of the present invention and objective
In the case where can make changes, modifications, alterations, and variations to the above described embodiments within the scope of the invention.
Claims (7)
1. a kind of lift car carrying number detection method based on the analysis of two-way monitor video, which is characterized in that including walking as follows
It is rapid:
Step S1, the video data of outer interior I, II two-way of elevator is monitored, and is transferred to hard disk video recorder；
Step S2, hard disk video recorder respectively gives two-path video algoritic module 1 and algoritic module 2 is analyzed；
Wherein, algoritic module 1 analyzes the road I video data, and disengaging carriage is calculated according to elevator door opening-closing process duration
Total number of persons；Algoritic module 2 analyzes the road II video data, according to foreground area variation before and after elevator door opening-closing be calculated personnel into
Directivity function out；
Step S3, algoritic module 1 and algoritic module 2 give calculated result to algoritic module 3, and algoritic module 3 is according to disengaging carriage
The product of total number of persons and directivity function obtains the net number into carriage.
2. the lift car carrying number detection method according to claim 1 based on the analysis of two-way monitor video, feature
It is, the road I video data is obtained by the road the I video camera acquisition being deployed in outside lift car, and the road I is camera supervised
Elevator outlet and its nearby aisle region；The road II video data is by being deployed in vertex and visual angle inclination in lift car
The road II video camera acquire obtain, the camera supervised elevator inner region in the road II.
3. the lift car carrying number detection method according to claim 1 based on the analysis of two-way monitor video, feature
It is, in step S2, algoritic module 1 analyzes the road I video data, and disengaging is calculated according to elevator door opening-closing process duration
The implementation of carriage total number of persons is as follows:
Step S101, pass through the parameter alpha and β of experimental calibration lift car；
Step S102, by obtaining elevator door opening-closing process duration T to the road I video analysis, and step S101 is obtained
The parameter alpha and β of lift car substitute into following formula:
| io |=α T+ β
By above formula, can be calculated into carriage personnel and the sum for leaving carriage personnel | io |.
4. the lift car carrying number detection method according to claim 3 based on the analysis of two-way monitor video, feature
It is, in step S102, by obtaining the specific implementation of elevator door opening-closing process duration T to the road I video analysis such as
Under:
If foreground pixel number is n in the ROI as caused by noise when elevator door is closed1, prospect picture in ROI when elevator door fully opens
Prime number is n2, the start/stop time that opening and closing event occurs for elevator is t1、t′1；The then ROI prospect picture that an elevator opening and closing event generates
The waveform available functions n (t of prime number variation1,t′1) description:
n(t1,t′1)=n1+(n2-n1)[S(k(t-t1))-S(k(t-t′1))]
In formula, coefficient k is the steep for influencing waveform raising and lowering, and S (t) is Sigmoid function, i.e.,
If occurring to be followed successively by t at the time of elevator opening and closing event in a period1,t′1,t2,t′2,…,tN,t′N, then ROI prospect picture
Prime number change curve may be expressed as:
During elevator opens or closes, waveform that ROI foreground pixel number changes over time has one and rises or falls
Slope section, for the ordinate of slope section there are maximum value and minimum value, capture prime number is equal to the average value of maximum value and minimum value
Moment is the starting and ending moment that elevator cab door is opened and closed process, even
And n1And n2It can be acquired by the analysis to the road I video data；
Searching algorithm is recycled to acquire t1With t '1；
Further, T=t ' is utilized1-t1Elevator door opening-closing process duration T is calculated.
5. the lift car carrying number detection method according to claim 4 based on the analysis of two-way monitor video, feature
It is, it is described to acquire t using searching algorithm1With t '1Implementation it is as follows:
In continuous monitoring data sequence (t, n), pixel number n must can be found equal to (n1+n2Two points of)/2, the two points
In, the lesser point of abscissa is the starting that elevator cab door is opened and closed process, and the biggish point of abscissa is that elevator cab door was opened and closed
The end of journey, i.e. t1Equal to lesser abscissa, t '1Equal to biggish abscissa.
6. the lift car carrying number detection method according to claim 1 based on the analysis of two-way monitor video, feature
It is, in step S2, algoritic module 2 analyzes the road II video data, is calculated according to foreground area variation before and after elevator door opening-closing
The implementation of the directivity function passed in and out to personnel is as follows:
If the foreground area of the road II video image is respectively S (t), S (t') before and after elevator door opening-closing, if moment t foreground area S
It (t) is more than preset area threshold value S0, then it is assumed that someone in elevator；If foreground area S (t) is less than preset area threshold value S0, then it is assumed that
In elevator nobody；The state for whether being detained personnel in elevator is indicated with function I (t), then
Remember that in is the net number into lift car, i.e. enter in elevator opening and closing event procedure carriage number x (t') with from
Open the difference of carriage number x (t):
In=x (t')-x (t)
Note is t and t' by the start/stop time that the road I video analysis obtains wherein elevator opening and closing event, in and | io | ratio be
Directivity function I (t, t'), then
In=I (t, t') | io |
| io | it is into carriage personnel and the sum for leaving carriage personnel；
Due to x (t), x (t') meets following linear equation in two unknowns group
So
Up to directivity function
Thus the net number in into lift car can be calculated.
7. a kind of lift car carrying number detection system based on the analysis of two-way monitor video characterized by comprising
The road I video camera, is deployed in outside lift car, the outlet of acquisition elevator and its nearby video data in aisle region；
The road II video camera is deployed in vertex in lift car, and visual angle is tilted towards outside elevator, acquires the row of personnel in elevator
For and abnormal video data；
Hard disk video recorder, the hard disk video recorder are connected with the road I video camera, the road II video camera, the HD recording
Machine receives, stores the video data from the road I video camera, the road II video camera；
Algoritic module 1 is connected with the hard disk video recorder, the road the I video data in the hard disk video recorder is analyzed, according to elevator
Disengaging carriage total number of persons is calculated in door opening and closing process duration；
Algoritic module 2 is connected with the hard disk video recorder, the road hard disk video recorder Zhong II video data is analyzed, according to electricity
The directivity function of personnel's disengaging is calculated in the foreground area variation of terraced door opening and closing front and back；
Algoritic module 3 is connected with the algoritic module 1, algoritic module 2, receives the analysis of the algoritic module 1, algoritic module 2
As a result, simultaneously further progress convergence analysis, the product that will pass in and out carriage total number of persons and directivity function is used as net people into carriage
Number；
Database is connected with the algoritic module 3, stores all previous output of the algoritic module 3 as a result, for data mining point
Analysis；
Alarm is connected with the algoritic module 3, issues alarm signal when lift car carrying number transfinites.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910298652.0A CN110040592B (en) | 2019-04-15 | 2019-04-15 | Elevator car passenger number detection method and system based on double-path monitoring video analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910298652.0A CN110040592B (en) | 2019-04-15 | 2019-04-15 | Elevator car passenger number detection method and system based on double-path monitoring video analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110040592A true CN110040592A (en) | 2019-07-23 |
| CN110040592B CN110040592B (en) | 2020-11-20 |
Family
ID=67277216
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910298652.0A Active CN110040592B (en) | 2019-04-15 | 2019-04-15 | Elevator car passenger number detection method and system based on double-path monitoring video analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110040592B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112573316A (en) * | 2020-12-08 | 2021-03-30 | 成都睿瞳科技有限责任公司 | Elevator trapping detection method based on computer vision |
| CN113044694A (en) * | 2021-03-05 | 2021-06-29 | 苏州威视通智能科技有限公司 | Construction site elevator people counting system and method based on deep neural network |
| CN116506473A (en) * | 2023-06-29 | 2023-07-28 | 北京格林威尔科技发展有限公司 | Early warning method and device based on intelligent door lock |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101920896A (en) * | 2009-06-10 | 2010-12-22 | 株式会社日立制作所 | Lift facility |
| CN102334142A (en) * | 2009-02-24 | 2012-01-25 | 三菱电机株式会社 | Human tracking device and human tracking program |
| JP5197115B2 (en) * | 2008-04-10 | 2013-05-15 | 三菱電機株式会社 | Elevator control device |
| CN103910272A (en) * | 2013-01-07 | 2014-07-09 | 东芝电梯株式会社 | Elevator system |
| CN107609597A (en) * | 2017-09-26 | 2018-01-19 | 嘉世达电梯有限公司 | A kind of number of people in lift car detecting system and its detection method |
| CN207713204U (en) * | 2017-06-26 | 2018-08-10 | 西安电子科技大学 | Number detecting system in lift car based on fuzzy reasoning |
| CN109384108A (en) * | 2017-08-08 | 2019-02-26 | 东芝电梯株式会社 | The control system of elevator |
-
2019
- 2019-04-15 CN CN201910298652.0A patent/CN110040592B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5197115B2 (en) * | 2008-04-10 | 2013-05-15 | 三菱電機株式会社 | Elevator control device |
| CN102334142A (en) * | 2009-02-24 | 2012-01-25 | 三菱电机株式会社 | Human tracking device and human tracking program |
| CN101920896A (en) * | 2009-06-10 | 2010-12-22 | 株式会社日立制作所 | Lift facility |
| CN103910272A (en) * | 2013-01-07 | 2014-07-09 | 东芝电梯株式会社 | Elevator system |
| CN207713204U (en) * | 2017-06-26 | 2018-08-10 | 西安电子科技大学 | Number detecting system in lift car based on fuzzy reasoning |
| CN109384108A (en) * | 2017-08-08 | 2019-02-26 | 东芝电梯株式会社 | The control system of elevator |
| CN107609597A (en) * | 2017-09-26 | 2018-01-19 | 嘉世达电梯有限公司 | A kind of number of people in lift car detecting system and its detection method |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112573316A (en) * | 2020-12-08 | 2021-03-30 | 成都睿瞳科技有限责任公司 | Elevator trapping detection method based on computer vision |
| CN113044694A (en) * | 2021-03-05 | 2021-06-29 | 苏州威视通智能科技有限公司 | Construction site elevator people counting system and method based on deep neural network |
| CN113044694B (en) * | 2021-03-05 | 2023-10-24 | 苏州威视通智能科技有限公司 | System and method for counting number of persons in building elevator based on deep neural network |
| CN116506473A (en) * | 2023-06-29 | 2023-07-28 | 北京格林威尔科技发展有限公司 | Early warning method and device based on intelligent door lock |
| CN116506473B (en) * | 2023-06-29 | 2023-09-22 | 北京格林威尔科技发展有限公司 | Early warning method and device based on intelligent door lock |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110040592B (en) | 2020-11-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111860282B (en) | Subway section passenger flow volume statistics and pedestrian retrograde motion detection method and system | |
| CN112257557B (en) | High-altitude parabolic detection and identification method and system based on machine vision | |
| CN106980829B (en) | Abnormal behaviour automatic testing method of fighting based on video analysis | |
| CN110040592A (en) | Lift car carrying number detection method and system based on the analysis of two-way monitor video | |
| US9451214B2 (en) | Indoor surveillance system and indoor surveillance method | |
| US7369680B2 (en) | Method and apparatus for detecting an event based on patterns of behavior | |
| CN101465033B (en) | Automatic tracking recognition system and method | |
| CN106241534B (en) | More people's boarding abnormal movement intelligent control methods | |
| CN106006266B (en) | A kind of machine vision method for building up applied to elevator safety monitoring | |
| CN106373240A (en) | Intelligent monitoring system of self-service bank and monitoring method thereof | |
| CN103198605A (en) | Indoor emergent abnormal event alarm system | |
| US20030059081A1 (en) | Method and apparatus for modeling behavior using a probability distrubution function | |
| US20140177946A1 (en) | Human detection apparatus and method | |
| CN110781723B (en) | Group abnormal behavior identification method | |
| CN110425005A (en) | The monitoring of transportation of belt below mine personnel's human-computer interaction behavior safety and method for early warning | |
| CN106128053A (en) | A kind of wisdom gold eyeball identification personnel stay hover alarm method and device | |
| CN106210634A (en) | A kind of wisdom gold eyeball identification personnel fall down to the ground alarm method and device | |
| CN101295405A (en) | Portrait and vehicle recognition alarming and tracing method | |
| CN102663452A (en) | Suspicious act detecting method based on video analysis | |
| CN101635834A (en) | Automatic tracing identification system for artificial neural control | |
| Nguyen et al. | Multistage real-time fire detection using convolutional neural networks and long short-term memory networks | |
| CN105426820A (en) | Multi-person abnormal behavior detection method based on security monitoring video data | |
| CN104899895B (en) | A kind of mobile target trajectory complexity detection method of electric transmission line channel pyrotechnics video | |
| CN106127814A (en) | A kind of wisdom gold eyeball identification gathering of people is fought alarm method and device | |
| CN101968848A (en) | Video monitoring method and system and video monitoring alarm system |
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 |