CN107285148A - Interest region decision system and method based on elevator scene - Google Patents
Interest region decision system and method based on elevator scene Download PDFInfo
- Publication number
- CN107285148A CN107285148A CN201710667804.0A CN201710667804A CN107285148A CN 107285148 A CN107285148 A CN 107285148A CN 201710667804 A CN201710667804 A CN 201710667804A CN 107285148 A CN107285148 A CN 107285148A
- Authority
- CN
- China
- Prior art keywords
- unit
- human body
- feature point
- body feature
- sight
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B66—HOISTING; LIFTING; HAULING
- B66B—ELEVATORS; ESCALATORS OR MOVING WALKWAYS
- B66B1/00—Control systems of elevators in general
- B66B1/02—Control systems without regulation, i.e. without retroactive action
- B66B1/06—Control systems without regulation, i.e. without retroactive action electric
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B66—HOISTING; LIFTING; HAULING
- B66B—ELEVATORS; ESCALATORS OR MOVING WALKWAYS
- B66B1/00—Control systems of elevators in general
- B66B1/34—Details, e.g. call counting devices, data transmission from car to control system, devices giving information to the control system
- B66B1/3415—Control system configuration and the data transmission or communication within the control system
Abstract
Present invention firstly provides a kind of interest region decision system based on elevator scene, including image collection unit, human body feature point detection unit, stereoscopic features matching unit, sight towards analytic unit and concern time acquisition unit, image collection unit sets up data cube computation with human body feature point detection unit, human body feature point collecting unit sets up data cube computation with stereoscopic features matching unit, stereoscopic features matching unit sets up data cube computation with sight towards analytic unit, and sight sets up data cube computation towards analytic unit with concern time acquisition unit.The present invention also provides a kind of interest region determination methods using above-mentioned system.The present invention can purposefully deliver advertisement, improve the validity of advertisement putting.
Description
Technical field
Set the present invention relates to technical field of computer vision, more particularly to technical field of computer vision in lift car
Application in terms of meter.
Background technology
Dispensing advertisement is broadly divided into following several in elevator:The 1st, photo frame is installed in car, 2, installing advertisement, sedan-chair on elevator door
Media advertisement has following advantage in railway carriage or compartment:High frequency, the reading frequency of the high arrival rate of high frequency time, daily averagely more than four times;Performance
Property, strong, the stronger picture impulsive force of expressive force;It is mandatory, disturb small, mandatory strong, lift space closing is numerous from other
Advertisement is disturbed, mandatory to read big;Low cost:Low advertising cost, high performance-price ratio.
But one it is important the problem of, no one know advertisement putting effect how, specifically how many people concern
The advertisement of dispensing, direct corresponding topic is that the advertisement delivered can bring great value to brand, and two are related to here
Problem:On the one hand advertisement delivery effect quality is that the quality with advertisement in itself is relevant, and whether attractive, advertisement delivery effect is good
Bad on the other hand closely bound up with audience, such as high-grade cell needs the somewhat high-grade brand of some to deliver.It is how effective
Advertisement putting efficiency in geo-statistic lift car, degree is concerned, allows the advertisement putting in lift car more to collect neutralization high
Effect, is that lift car arrangement needs the problem of emphasis considers.
The content of the invention
The technical problem that the present invention first has to solve is to provide a kind of interest region decision system based on elevator scene, profit
The concerned degree of billboard in lift car can be counted with the system.
The present invention solves the technical scheme that is used of above-mentioned technical problem:Interest region decision system based on elevator scene
System, including image collection unit, human body feature point detection unit, stereoscopic features matching unit, sight towards analytic unit and
Time acquisition unit is paid close attention to, image collection unit is arranged in lift car and can gather the multi-angle of passenger in lift car
Image information, image collection unit and human body feature point detection unit set up data cube computation and send out the image information gathered
Human body feature point detection unit is delivered to, human body feature point detection unit carries out video analysis according to the image information to collection and obtained
Corresponding human body feature point, human body feature point collecting unit sets up data cube computation with stereoscopic features matching unit and will analysis gained
Human body feature point is sent to stereoscopic features matching unit, and human body feature point match obtaining each by stereoscopic features matching unit
The corresponding three-dimensional coordinate of point, stereoscopic features matching unit sets up data cube computation towards analytic unit with sight and will match what is obtained
Three-dimensional coordinate is sent to sight towards analytic unit, and sight is obtained in lift car towards analytic unit according to three-dimensional coordinate analysis
The sight direction of passenger, sight sets up data cube computation and by the sight court of passenger towards analytic unit with concern time acquisition unit
Sent to information to concern time acquisition unit, concern time acquisition unit calculates according to sight orientation information and obtains lift car
Interior billboard receives the time accounting of concern.
Further, image collection unit is binocular camera.
Another technical problem to be solved by this invention is to provide a kind of interest region decision side based on elevator scene
Method, this method utilizes above-mentioned system, and comprises the following steps:
(1) the multi-angle image information and transmission to human body feature point of passenger are examined in image collection unit collection lift car
Survey unit,
(2) human body feature point detection unit carries out video analysis to the image information of collection and obtains corresponding human body feature point
And send the human body feature point to stereoscopic features matching unit,
(3) human body feature point match obtaining the corresponding three-dimensional coordinate of every bit and should by stereoscopic features matching unit
Three-dimensional coordinate is sent to sight towards analytic unit,
(4) sight is analyzed the sight direction for obtaining passenger in lift car according to three-dimensional coordinate towards analytic unit and will regarded
Line orientation information is sent to concern time acquisition unit,
(5) concern time acquisition unit obtains billboard in lift car according to the calculating of sight orientation information and receives concern
Time accounting.
Further, image collection unit is binocular camera, and binocular camera gathers image and each preserved respectively.
Further, human body feature point detection unit obtains the position of human joint pointses in the picture using pose estimation technology
Put, and these positions are defined as human body feature point, these human body feature points include the crown, eyes, ears, both shoulders, duplex, double
Wrist, double sterns, double knees and double ankles, altogether 17 human body feature points.
Further, human body feature point is mapped to three dimensions by stereoscopic features matching unit, and utilizes Stereo matching skill
Art, obtains disparity map of each human body feature point in the image that binocular camera is gathered respectively, so that it is special to obtain each human body
Levy three-dimensional coordinate a little in the picture.
Further, sight towards analytic unit by the people of the human body feature point of the double sterns of two correspondences and two correspondence both shoulders
Four three-dimensional coordinates of body characteristicses point, the characteristic vector corresponding to minimal eigenvalue is obtained using singular value decomposition, this feature to
Amount characterizes the sight direction of people.
Further, concern time acquisition unit calculating sight towards vector, all around intersect with lift car by four faces
Situation, obtains the time that each face is noted, so as to obtain the time accounting that billboard in car attracts attention, pays close attention to the time
The calculation formula of accounting is as follows:
Per=people pays close attention to some billboard time/people total time × 100% in car.
The beneficial effects of the invention are as follows:First with pose estimation technology, human body feature point is detected, solid is then utilized
The topology distribution data-evaluation that these characteristic points are mapped to characteristic point in three dimensions, last utilization space is gone out people by matching technique
Direction, finally give the concern time accounting that people's some billboard in elevator is obtained, concern event accounting this refer to
Mark can be used for weighing the level index that some advertisement is concerned, so that the efficiency of the advertisement putting is judged, while being capable of root
According in some region of the indicator-specific statistics, such as in the lift car of some cell, what kind of billboard be easier by
To concern, so as to purposefully deliver advertisement, the validity of advertisement putting is improved.
Brief description of the drawings
Fig. 1 is the human body feature point distribution map of the present invention.
Fig. 2 is the structure chart of the present invention.
Fig. 3 is the flow chart of the present invention.
Fig. 4 is the illustraton of model of the camera imaging of the present invention.
Fig. 5 is the three-dimensional modeling figure in people's lift stand car.
Embodiment
Referring to the drawings.
The interest region decision system based on elevator scene of the present invention, it includes image collection unit 1, human body feature point
Detection unit 2, stereoscopic features matching unit 3, sight are towards analytic unit 4 and pay close attention to time acquisition unit 5, image collection list
Member 1 is arranged in lift car and can gather the image information of the multi-angle of passenger in lift car, image collection unit 1 and
Human body feature point detection unit 2 sets up data cube computation and sends the image information gathered to human body feature point detection unit 2,
Human body feature point detection unit 2 carries out video analysis according to the image information to collection and obtains corresponding human body feature point, human body
Characteristic point collecting unit 2 sets up data cube computation with stereoscopic features matching unit 3 and sends analysis gained human body feature point to vertical
Body characteristicses matching unit 3, human body feature point match by stereoscopic features matching unit 3 obtains the corresponding three-dimensional seat of every bit
Mark, stereoscopic features matching unit 3 and sight are set up towards analytic unit 4 to be connected and sends the obtained three-dimensional coordinate of matching to regarding
Line analyzes the sight court for obtaining passenger in lift car towards analytic unit 4 towards analytic unit 4, sight according to three-dimensional coordinate
To sight sets up data cube computation with concern time acquisition unit 5 towards analytic unit 4 and sends the sight orientation information of passenger
To concern time acquisition unit 5, concern time acquisition unit 5 calculates according to sight orientation information and obtains billboard in lift car
Receive the time accounting of concern.In said structure, only image collection unit 1 needs to be arranged in lift car, other
Structure can set up data cube computation by way of the wireless connection wireless (such as WIFI) with image collection unit 1, and differ
Provisioning request will be also arranged in lift car.
The method judged using the system of the present invention, specifically includes following steps:
(1) the multi-angle image information and transmission to human body feature point of passenger are examined in image collection unit collection lift car
Unit is surveyed, image collection unit is binocular camera, and binocular camera gathers image and each preserved respectively.
(2) human body feature point detection unit carries out video analysis to the image information of collection and obtains corresponding human body feature point
And send the human body feature point to stereoscopic features matching unit, human body feature point detection unit is obtained using pose estimation technology
The position of human joint pointses in the picture, and these positions are defined as human body feature point, these human body feature points include the crown,
Eyes, ears, both shoulders, duplex, double wrists, double sterns, double knees and double ankles, 17 human body feature points altogether, as shown in figure 1, profit at present
With cnn (convolutional neural networks) even depth learning art, human body feature point accuracy in detection has reached a suitable height.
(3) human body feature point match obtaining the corresponding three-dimensional coordinate of every bit and should by stereoscopic features matching unit
Three-dimensional coordinate is sent to sight towards analytic unit, and human body feature point is mapped to three dimensions by stereoscopic features matching unit, and
Using Stereo Matching Technology, disparity map of each human body feature point in the image that binocular camera is gathered respectively is obtained, so that
To the three-dimensional coordinate of each human body feature point in the picture.
The model of camera imaging is as shown in Figure 4:
P is the point in space, and P1 and P2 are imaging points of the point P in the image plane of left and right, and f is focal length, and OR and OT are left and right
The photocentre of camera.The optical axis of two cameras in left and right is parallel as seen from the figure.XRAnd XTIt is two imaging points in the image planes of left and right two
The distance of upper range image left hand edge.Z is Z axis coordinate of the P points under camera coordinates system in space, and parallax d=XR-XT.Simultaneously
The distance between binocular camera is b.If two cameras have corrected that completion reaches that polar curve is parallel, two optical axis directions are also put down
OK.Then the relational expression of parallax and Object Depth is as follows:
It can derive
Similarly it can be seen from similar triangles relation
Derive
DeriveWherein x, y are the image coordinate using picture centre as the origin of coordinates.So far, binocular
After camera calibration, it is known that point coordinates x, a y in image, the parallax d obtained according to left images may know that its corresponding sky
Between coordinate XYZ.
(4) sight is analyzed the sight direction for obtaining passenger in lift car according to three-dimensional coordinate towards analytic unit and will regarded
Line orientation information is sent to concern time acquisition unit, sight towards analytic unit by the human body feature point of the double sterns of two correspondences and
Four three-dimensional coordinates of the human body feature point of two correspondence both shoulders, the spy corresponding to minimal eigenvalue is obtained using singular value decomposition
Vector is levied, this feature vector characterizes the sight direction of people.
As shown in figure 5, in the middle of people's lift stand car, being derived from above, it is assumed that four human bodies of shoulder and buttocks
The three-dimensional coordinate X of characteristic point1Y1Z1, X2Y2Z2, X3Y3Z3, X4Y4Z4, it is expressed as p1,p2,p3,p4, obtain people's sight towards step such as
Under:
√ obtains the average of four points:
√ obtains covariance matrix:
√ carries out SVD decomposition to M, because M is symmetrical matrix, is equal to Orthogonal Decomposition, i.e. M=Q Λ QT, wherein Q is just
Battle array is handed over, each column vector is M characteristic vector, and Λ is diagonal matrix, each element is M characteristic value.Here only
Need Q the 3rd column vector q3。
The currently known average points of √And vector q3, linear equation can be obtained,
It is assumed that the people in car faces elevator doorway, then q3Approached with vectorial [0,1,0], if back to elevator door
Mouth q3Approached with vectorial [0, -1,0], towards elevator left side q3Approached with vectorial [- 1,0,0], towards elevator the right q3With vector
[1,0,0] approach.
(5) concern time acquisition unit obtains billboard in lift car according to the calculating of sight orientation information and receives concern
Time accounting.Calculating sight, all around situation is intersected in four faces with lift car towards vector, obtains each face and is noted
Time, so as to obtain the time accounting that billboard in car attracts attention, the calculation formula for paying close attention to time accounting is as follows:
Per=people pays close attention to some billboard time/people total time × 100% in car.
As described above, having obtained linear equation, that determination people sees that plane only needs to calculate straight line and four in car
The intersecting situation in individual face, if the length, width and height of elevator are respectively a, b, the equation in tetra- faces of c is respectively
This face of √ elevator switch door:
Y=b (0<x<a,0<z<C),
Behind √ elevators:
Y=0 (0<x<a,0<z<C),
The √ elevators left side:
X=0 (0<y<b,0<z<C),
The √ elevators right side:
X=a (0<y<b,0<z<C),
So straight line is only possible to intersect with a face in four faces for normal, and it is judged as that the face is concerned.
Claims (8)
1. the interest region decision system based on elevator scene, it is characterized in that, including the detection of image collection unit, human body feature point
Unit, stereoscopic features matching unit, sight are arranged on towards analytic unit and concern time acquisition unit, image collection unit
Lift car is interior and can gather the image information of the multi-angle of passenger in lift car, image collection unit and human body feature point
Detection unit sets up data cube computation and sends the image information gathered to human body feature point detection unit, human body feature point inspection
Survey unit and corresponding human body feature point, human body feature point collecting unit are obtained according to the image information progress video analysis to collection
Data cube computation is set up with stereoscopic features matching unit and sends analysis gained human body feature point to stereoscopic features matching unit, is stood
Body characteristicses matching unit carries out human body feature point to match and obtain the corresponding three-dimensional coordinate of every bit, stereoscopic features matching unit with
Sight sets up data cube computation towards analytic unit and will match obtained three-dimensional coordinate and sends to sight towards analytic unit, sight
The sight direction for obtaining passenger in lift car is analyzed according to three-dimensional coordinate towards analytic unit, sight is towards analytic unit with closing
Note time acquisition unit sets up data cube computation and sends the sight orientation information of passenger to concern time acquisition unit, during concern
Between acquiring unit according to sight orientation information calculate obtain the time accounting that billboard in lift car receives concern.
2. the interest region decision system according to claim 1 based on elevator scene, it is characterized in that, image collection unit
It is binocular camera.
3. the interest region determination methods based on elevator scene, it is characterized in that, this method utilizes the system described in claim 1,
And comprise the following steps:
(1) the multi-angle image information and transmission to human body feature point of passenger are detected singly in image collection unit collection lift car
Member,
(2) human body feature point detection unit obtains corresponding human body feature point to the image information progress video analysis of collection and will
The human body feature point is sent to stereoscopic features matching unit,
(3) human body feature point match by stereoscopic features matching unit obtains the corresponding three-dimensional coordinate of every bit and by the three-dimensional
Coordinate is sent to sight towards analytic unit,
(4) sight is analyzed towards analytic unit according to three-dimensional coordinate obtains the sight of passenger in lift car towards and by sight court
Sent to information to concern time acquisition unit,
(5) concern time acquisition unit calculates according to sight orientation information and obtains the time that billboard in lift car receives concern
Accounting.
4. the interest region determination methods according to claim 3 based on elevator scene, it is characterized in that, image collection unit
It is binocular camera, binocular camera gathers image and each preserved respectively.
5. the interest region determination methods according to claim 3 based on elevator scene, it is characterized in that, human body feature point inspection
Survey unit and obtain the position of human joint pointses in the picture using pose estimation technology, and these positions are defined as characteristics of human body
Point, these human body feature points include nose, eyes, ears, both shoulders, duplex, double wrists, double sterns, double knees and double ankles, altogether 17 people
Body characteristicses point.
6. the interest region determination methods according to claim 3 based on elevator scene, it is characterized in that, stereoscopic features matching
Human body feature point is mapped to three dimensions by unit, and utilizes Stereo Matching Technology, obtains each human body feature point in binocular phase
Disparity map in the image that machine is gathered respectively, so as to obtain the three-dimensional coordinate of each human body feature point in the picture.
7. the interest region determination methods according to claim 3 based on elevator scene, it is characterized in that, sight direction analysis
Unit is by the human body feature point of the double sterns of two correspondences and four three-dimensional coordinates of the human body feature point of two correspondence both shoulders, using strange
Different value decomposes the characteristic vector obtained corresponding to minimal eigenvalue, and this feature vector characterizes the sight direction of people.
8. the interest region determination methods according to claim 3 based on elevator scene, it is characterized in that, the concern time obtains
Unit calculates sight, and towards vector, all around situation is intersected in four faces with lift car, obtain that each face is noted when
Between, so as to obtain the time accounting that billboard in car attracts attention, the calculation formula for paying close attention to time accounting is as follows:
Per=people pays close attention to some billboard time/people total time × 100% in car.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710667804.0A CN107285148A (en) | 2017-08-07 | 2017-08-07 | Interest region decision system and method based on elevator scene |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710667804.0A CN107285148A (en) | 2017-08-07 | 2017-08-07 | Interest region decision system and method based on elevator scene |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107285148A true CN107285148A (en) | 2017-10-24 |
Family
ID=60104722
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710667804.0A Pending CN107285148A (en) | 2017-08-07 | 2017-08-07 | Interest region decision system and method based on elevator scene |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107285148A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107973198A (en) * | 2017-11-30 | 2018-05-01 | 成都科力夫科技有限公司 | The elevator card jettison system of Auto-matching occupant |
CN114604800B (en) * | 2022-03-31 | 2023-12-15 | 江苏西顿科技有限公司 | Explosion-proof AGV car |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006020131A (en) * | 2004-07-02 | 2006-01-19 | Nippon Telegr & Teleph Corp <Ntt> | Device and method for measuring interest level |
JP2010063001A (en) * | 2008-09-05 | 2010-03-18 | Mitsubishi Electric Corp | Person-tracking device and person-tracking program |
CN102081890A (en) * | 2009-11-30 | 2011-06-01 | 鸿富锦精密工业(深圳)有限公司 | Information control device and method and information display system with information control device |
CN102934458A (en) * | 2011-02-04 | 2013-02-13 | 松下电器产业株式会社 | Interest estimation device and interest estimation method |
JP2016169067A (en) * | 2015-03-11 | 2016-09-23 | 三菱電機株式会社 | Elevator system |
-
2017
- 2017-08-07 CN CN201710667804.0A patent/CN107285148A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006020131A (en) * | 2004-07-02 | 2006-01-19 | Nippon Telegr & Teleph Corp <Ntt> | Device and method for measuring interest level |
JP2010063001A (en) * | 2008-09-05 | 2010-03-18 | Mitsubishi Electric Corp | Person-tracking device and person-tracking program |
CN102081890A (en) * | 2009-11-30 | 2011-06-01 | 鸿富锦精密工业(深圳)有限公司 | Information control device and method and information display system with information control device |
CN102934458A (en) * | 2011-02-04 | 2013-02-13 | 松下电器产业株式会社 | Interest estimation device and interest estimation method |
JP2016169067A (en) * | 2015-03-11 | 2016-09-23 | 三菱電機株式会社 | Elevator system |
Non-Patent Citations (2)
Title |
---|
赵乐军等: "头盔式眼动仪的标定与算法实现", 《西安电子科技大学学报》 * |
黄静编著: "《计算机图形学及其实践教程》", 30 June 2015, 机械工业出版社 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107973198A (en) * | 2017-11-30 | 2018-05-01 | 成都科力夫科技有限公司 | The elevator card jettison system of Auto-matching occupant |
CN114604800B (en) * | 2022-03-31 | 2023-12-15 | 江苏西顿科技有限公司 | Explosion-proof AGV car |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102750695B (en) | Machine learning-based stereoscopic image quality objective assessment method | |
CN104036488B (en) | Binocular vision-based human body posture and action research method | |
CN107358631A (en) | A kind of binocular vision method for reconstructing for taking into account three-dimensional distortion | |
CN109118479A (en) | Defects of insulator identification positioning device and method based on capsule network | |
CN106920276B (en) | A kind of three-dimensional rebuilding method and system | |
CN107886477A (en) | Unmanned neutral body vision merges antidote with low line beam laser radar | |
CN105320271A (en) | HMD calibration with direct geometric modeling | |
CN101610425B (en) | Method for evaluating stereo image quality and device | |
CN101216304A (en) | Systems and methods for object dimension estimation | |
CN104754322B (en) | A kind of three-dimensional video-frequency Comfort Evaluation method and device | |
CN107656619A (en) | A kind of intelligent projecting method, system and intelligent terminal | |
CN205610834U (en) | Stereo display system | |
CN105825499A (en) | Reference plane determination method and determination system | |
CN104021538A (en) | Object positioning method and device | |
CN107578376A (en) | The fork division of distinguished point based cluster four and the image split-joint method of local transformation matrix | |
CN107016697A (en) | A kind of height measurement method and device | |
US20130274027A1 (en) | Golf swing classification method, classification system, classification apparatus, and non-transitory computer-readable storage medium with program recorded thereon | |
CN107397658A (en) | A kind of multiple dimensioned full convolutional network and vision blind-guiding method and device | |
CN107689060A (en) | Visual processing method, device and the equipment of view-based access control model processing of destination object | |
CN106447707A (en) | Image real-time registration method and system | |
CN110274598A (en) | A kind of robot monocular vision robust location estimation method | |
CN108362205A (en) | Space ranging method based on fringe projection | |
CN107285148A (en) | Interest region decision system and method based on elevator scene | |
CN106210710A (en) | A kind of stereo image vision comfort level evaluation methodology based on multi-scale dictionary | |
CN208754392U (en) | A kind of three lens camera and the education recorded broadcast device with it |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171024 |