CN107928675A - A kind of trunk measuring method being combined based on deep learning and red dot laser - Google Patents
A kind of trunk measuring method being combined based on deep learning and red dot laser Download PDFInfo
- Publication number
- CN107928675A CN107928675A CN201711171226.8A CN201711171226A CN107928675A CN 107928675 A CN107928675 A CN 107928675A CN 201711171226 A CN201711171226 A CN 201711171226A CN 107928675 A CN107928675 A CN 107928675A
- Authority
- CN
- China
- Prior art keywords
- red
- trunk
- image
- point
- video camera
- 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
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/107—Measuring physical dimensions, e.g. size of the entire body or parts thereof
- A61B5/1075—Measuring physical dimensions, e.g. size of the entire body or parts thereof for measuring dimensions by non-invasive methods, e.g. for determining thickness of tissue layer
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/107—Measuring physical dimensions, e.g. size of the entire body or parts thereof
- A61B5/1079—Measuring physical dimensions, e.g. size of the entire body or parts thereof using optical or photographic means
Abstract
The present invention proposes a kind of trunk measuring method being combined based on deep learning and red dot laser, the trunk of destination object is detected from image using deep learning, two light sources are fixed and parallel red dot laser is beaten on human body, and the position of red point is detected from image by red detection algorithm, calculate the pixel distance of red point, and the conversion of the pixel distance and actual range of image is obtained according to the actual range of red point, and then the physical length of trunk can be calculated according to the length in pixels of trunk.It is of the invention with existing trunk measure compared with, have the advantages that the non-cpntact measurement based on identification can be carried out to objectives thing under reality scene.
Description
Technical field
The present invention provides a kind of trunk measuring method being combined based on deep learning and red dot laser, it is related to depth
Degree study, technical field of computer vision.
Background technology
Anthropometry is developed so far, survey tool by initial tape measure, tape measure, developed into computer, sensor with
And the high-acruracy survey instrument such as laser, the method for measurement pass through constantly improve, there is very high precision and efficiency.From technology
From the point of view of development, human body measurement technology can be divided into common survey technology and three-dimensional digital human body measurement technology.
1st, average person measures
The related instrument that average person's measuring instrument can be measured using general Human Physiology, including anthropometer,
Right-angle gauge, bent angle rule, coordinate caliper, tape, tooth caliper, cubic craniophor, parallel fixed point apparatus etc., its data processing uses people
Work processing is manually entered the mode being combined with computer disposal.Such a measurement method takes time and effort, and data processing is easy
Error, data application is dumb, but of low cost, has certain applicability.
2nd, three-dimensional digital human body measurement technology
In recent years, domestic and international textile garment industrial computer Computer Aided Design (CAD) technology reaches its maturity, and promotes three-dimensional people
The development of bulk measurement technology.The flourishing country of the rag trades such as the U.S., France, Canada, Britain, Germany and Japan is in 20th century 70
Mid-nineties 90 starts to propose many new Principle and method of measurement.The research of China in this respect is started late, but near several
Year has some research institutions successively to carry out research work, and achieves some achievements in research and have accumulated certain experience.From
For the principle of instrument body, three-dimensional digital anthropological measuring be divided into manual contact formula, manually contactless, automatic contacting,
Automatically it is contactless etc..
2.1 3 D human body manual measurement technologies
Manual three-dimensional e measurement technology can be divided into contact type measurement technology and non-contact measuring technology again.
(1) manual contact formula three-dimensional digitized measurement
The FaroArm of Florida, US Faro technology companies is typical manual contact formula digitalized measuring apparatus.Measurement
When, operator holds Faro arms, and the probe contact of its end presses lower button when being tested the surface of human body, measures human body surface point
Locus.X, Y, Z coordinate and the probe handle direction of probe institute measuring point under three-dimensional data information record, and use DSP skills
Art is connected to various application software by RS232 Serial Port Lines and wraps.
(2) manual non-contact 3-D digitized measurement
The developed country such as the research for non-contact 3-D human body measurement technology, the U.S., Britain, Germany starts more early
Typically the LASS technologies of Britain and the white light phase measurement of U.S. TC2 exploitations.
2.2 three-dimensional human body automatic measuring technique
Three-dimensional human body automatic measuring is before realizing clothes Design of digital, three-dimensional virtual fitting and the application such as making to measure
Carry.It can be divided into two kinds of contact and non-contact measurement.
(1) contact automatic measurement
Contact automatic measurement is mainly used for the measurement of standard Ren Tai surfaces human body data point.Specific method is：Using three
Coordinate measuring apparatus measurement and the data point of recorder's platform different parts, the measurement interval of wherein curvature of curved surface larger part is close, offset
Point is more；The measurement interval at the flatter place of curved surface is dredged, and data point is few.This method is not suitable for the measurement of real human body.
(2) contactless automatic measurement
Contactless automatic measurement is mainly used for obtaining real human body characteristic.Non-contact three-dimensional automatic human body measuring is more
The deficiency of the contact automatic measurement of routine has been mended, has been mainly characterized by quick, accurate, efficiency is high etc..It is modern image measurement
One branch of technology, based on contemporary optics, melts photoelectronics, computer graphics, information processing, computer vision etc.
Science and technology is in the e measurement technology of one.It is when measuring measurand image as the means or load for detecting and transmitting information
Body is used, and the purpose is to useful information is extracted from image.
The present invention is combined red dot laser ranging technology with depth learning technology popular at present, it is proposed that one kind is based on
The distance measuring method that deep learning and red dot laser are combined.This method is a kind of contactless method for automatic measurement, by with base
It is combined in the trunk identification of deep learning, reaches the ability to trunk measurement, and can complete to identification
Multiple somatometric requirements.
The content of the invention
Present invention solves the technical problem that it is：For existing non-contact measurement method to trunk measuring apparatus excessively
The problem of expensive and computationally intensive, there is provided the measuring method that a kind of trunk identification is combined with red dot laser, having can
The advantages of non-cpntact measurement based on identification is carried out to trunk under reality scene.
The present invention provides a kind of trunk measuring method being combined based on deep learning and red dot laser, including following
Step：
Step (1), first fix two red dot laser transmitters and video camera, and ensures that two-beam line is parallel, and
It is vertical with the projection plane of video camera, i.e., it is parallel with the optical axis of video camera so that and shooting function captures clearly two red points,
Record the actual range of two red dot laser transmitters；In the step, video camera is demarcated, such as the ginseng of calibrating camera
Number, including inner parameter and/or external parameter.The benefit that two beam parallel lasers are used in step (1) is no matter what laser impinges upon
Place, the actual range of two red points is all the time in the picture arrived all the time with light ray parallel, camera acquisition due to the optical axis of video camera
It is constant.And when laser is impinged upon on human body, trunk is identical with the distance of red point to video camera, in image coordinate system he
Ratio and world coordinate system in ratio also approximately equal, therefore can easily calculate the length of trunk.
Step (2), by the image under step (1) calibrated camara module capture reality scene.
Step (3), by step (2) capture image be input to the red detection module based on opencv, detect and identify
Go out the red point of laser in image, if input picture includes the red point that can be recognized, two red points of output are in input picture
In coordinate position, if input picture does not include the red point that can recognize, back to step (2).
The image that camara module in step (2) captures, is input to the trunk inspection based on deep learning by step (4)
Module is surveyed, detects and identifies the trunk in image, if input picture includes the human body that can be recognized, output is each
The coordinate position of a human body key point in the input image, and the position of human body key point is linked, obtain complete human body
Trunk；If input picture does not include the human body that can be recognized, back to step (2).Step (4) is based on deep learning
The input picture of trunk detection module comes from the image of camara module capture.Propose a kind of human body of openpose
Trunk detects and recognizer, returns out the position of human body key point first, then links each key point, obtains complete people
Body trunk.
The position for the human body key point that step (5), the position of the two red points obtained according to step (3) and step (4) obtain
Put, respectively the length in pixels of the pixel distance of two red points of calculating and trunk；According to the actual range of red dot laser, obtain
The mapping of pixel distance and actual range；The reality of trunk is calculated according to the length in pixels of the mapping and trunk
Length.
In the above method, video camera is monocular camera.Image under calibrated monocular camera module capture reality scene,
Human body should face camera and keep vertical with laser direction at this time, be calculated with facilitating.
A certain range of space is needed in the above method, immediately ahead of video camera, such as measured people takes the photograph at station in distance
In the range of 2~10 meters of camera, it can so ensure that video camera can collect complete and clearly human body.
In the above method, the step of red detection module detects and identify red of laser in place's image in step (3), is specific
Including：Input picture is transformed into HSV space from rgb space first, for tri- passages of H, S, V respectively into row threshold division,
Then region merging technique is carried out, the bianry image of red point is obtained, finally carries out contours extract, obtains the seat of red point in the input image
Cursor position.
The principle of the present invention is：
The present invention proposes a kind of trunk measuring method being combined based on deep learning with red dot laser, overcomes
The shortcomings that existing non-contact measurement method is prohibitively expensive to trunk measuring apparatus and computationally intensive.This method is suitable for room
The measurement for the trunk that can be recognized using deep learning method in suitable distance under interior scene.This method includes four steps
Suddenly：Two red dot laser transmitters are fixed with video camera, and ensure that two-beam line is parallel, and are put down with the optical axis of video camera
OK so that shooting function captures clearly two red points, records the actual range of two red dot laser transmitters；Utilize calibration
The image in video camera capture reality scene afterwards；Then the image that video camera captures the red point based on opencv is input to examine
Module is surveyed, the position of two red points in image is provided by the module；The image that video camera captures is input to based on deep learning
Trunk identification module, the position of the human body key point included in image in the picture is provided by the module；Last basis
Red obtained position and the position of human body key point, can calculate the pixel distance of two red points and the pixel of trunk
Length, and according to the actual range of red dot laser, the mapping of pixel distance and actual range can be obtained, so as to calculate human body
The physical length of trunk.
Present disclosure mainly includes following aspect：
Acquisition system is built.Two red dot laser transmitters are fixed with video camera, and ensure that two-beam line is parallel, and
It is and parallel with the optical axis of video camera so that shooting function captures clearly two red points.One is needed immediately ahead of video camera
The space of scope is determined, it is necessary to which measured people station in the range of 2~10 meters of video camera, is ensureing that video camera can collect
Complete and clearly human body.The actual range of two red dot laser transmitters is recorded at the same time.
The acquisition of image.The input picture of trunk identification module and red identification module comes from calibrated shooting
The image of machine module capture, the video camera used is monocular camera.
The position detection of the red point of laser in image.The module uses the red recognition methods based on opencv, first
Image is transformed into HSV space from rgb space, for tri- passages of H, S, V respectively into row threshold division, then carries out region conjunction
And the bianry image of the red point of laser is obtained, contours extract is finally carried out, obtains the position of the red point of laser.
Trunk detection based on deep learning.The module is used to be detected based on deep learning trunk and calculated
Method-openpose.The main thought of this method is that big receptive field is obtained using big convolution kernel, to the people that is occurred into
Row returns, and returns the point in each personal joint, then gets rid of the response to other people according to center map, finally by
Repeatedly final result is obtained to predicting the heatmap come progress refine.
The calculating of trunk length.After the position of red position and human body key point is obtained, two can be calculated
The pixel distance of red point and the length in pixels of trunk, and according to the actual range of red dot laser, pixel distance can be obtained
With the mapping of actual range, so as to calculate the physical length of trunk.
The present invention compared with prior art the advantages of be：
The present invention proposes the method being combined based on deep learning with red dot laser for measuring trunk length.
This method can solve existing non-contact measurement method deficiency prohibitively expensive to trunk measuring apparatus and computationally intensive
Part.In actual scene, human body key point coordinate position is identified with identification module with reference to trunk detection, examined with red point
Survey module and be combined the actual range that can obtain trunk in the image.
Brief description of the drawings
Fig. 1 is a kind of flow for the trunk measuring method being combined based on deep learning and red dot laser of the present invention
Figure；
Fig. 2 is the red detection algorithm flow chart based on opencv；
Fig. 3 is 18 key point schematic diagrames of human body；
Fig. 4 is that the human body key point based on deep learning identifies schematic network structure.
Embodiment
Illustrate to be described in more detail the present invention below in conjunction with the accompanying drawings.
As shown in Figure 1, a kind of trunk measuring method being combined based on deep learning with red dot laser of the present invention,
Realize that flow includes：Image Acquisition (by calibrated monocular camera, red dot laser transmitter), the laser based on opencv are red
Point identification, the trunk detection based on deep learning, trunk length computation, specific embodiment are as follows：
1st, Image Acquisition
When carrying out Image Acquisition, two red dot laser transmitters are fixed with video camera first, two beam laser are parallel simultaneously
It is and vertical with the projection plane of video camera so that shooting function captures clearly two red points, the red dot laser hair of record two
The actual range of emitter；In the step, video camera is demarcated, such as the parameter of calibrating camera, including inner parameter and/or
External parameter.
There is certain requirement in the position that this measuring method stands red dot laser transmitter and the position of video camera and people,
The two beam laser requirement of wherein red dot laser transmitter transmitting is parallel, and their needs are vertical with the projection plane of video camera, i.e.,
It is parallel with the optical axis of video camera.Measured people station is needed to ensure video camera in the range of 2~10 meters of video camera at the same time
Complete and clearly human body, therefore a certain range of space is needed immediately ahead of video camera can be collected.Therefore this method
Acquisition system build and first fix two red dot laser transmitters with video camera, and ensure that two-beam line is parallel, and with
The optical axis of video camera is parallel so that shooting function captures clearly two red points, records the reality of two red dot laser transmitters
Border distance.Before collection, the inner parameter of video camera is demarcated.After the completion of acquisition system is built, gathered using this module
Image provides image input for red detection module and trunk identification module.
2nd, the red point identification of laser
The purpose of the module is come the red point in detection image using the red recognizer based on opencv.The algorithm
Main thought is that image is transformed into HSV space from rgb space first, for tri- passages of H, S, V respectively into row threshold division,
Then region merging technique is carried out, the bianry image of the red point of laser is obtained, finally carries out contours extract, obtains the position of the red point of laser,
Flow chart is as shown in Figure 2.
The parameter in hsv color space is respectively：Tone (H), saturation degree (S), lightness (V).Compared to rgb space, HSV face
The colour space more meets human eye perception characteristics, its effect in color segmentation field is also more preferable.
It is by the conversion formula of RGB to HSV：
V=max (R, G, B)
After image is transformed into hsv color space, for tri- passages of H, S, V respectively into row threshold division.Due to we
Method is using red dot laser, and the scope of H passages red is 0~10 and 156~180, and Binding experiment is as a result, true by H scopes
It is set to 0~20 and 160~180, while by experiment, the scope for determining channel S and V passages is respectively S：100~255, V：200
~256.
After the segmentation figure picture of three passages is obtained, pixel and operation, the segmentation after being merged are carried out to three passages
Image, the image are the bianry image of the red point of laser, then need to extract the position of the red point of laser from bianry image
Come.This method uses the contours extract function findContous of opencv, and the profile of segmentation result is extracted from bianry image,
And sort according to the size of profile, two regions of area maximum are the region where the red point of laser, constituency region
Coordinate of the central point as red dot laser.
3rd, trunk detects
In view of the speed and precision of identification, this method uses the openpose algorithms of current better performances, network structure
As shown in Figure 3.The effect of trunk identification module based on the algorithm is as follows：
According to input picture, the coordinate position of each human body key point recognized in the input image is exported.Human body closes
Key point is 18 key points preset in advance, this 18 key points are as shown in Figure 3.
Human body critical point detection using Convolutional Pose Machine (CPM) model improvement.CPM moulds
The algorithm idea of type mainly obtains big receptive field using big convolution kernel, and the people occurred is returned, and returns each
The point in personal joint, then gets rid of the response to other people according to center map, finally by repeatedly to prediction
Heatmap out carries out refine and obtains final result.
The main flow of algorithm is as follows：
1) under each scale, the response diagram in each joint is calculated.
2) for each joint, the response diagram for all scales that add up, obtains overall response figure.
3) on the overall response figure in each joint, the point of corresponding maximum is found out, for the position of the artis.
Openpose network models as shown in figure 4, the first stage is a basic convolutional network 1 (white convs), from
Coloured image directly predicts the response of each component.Bust form has 9 components, additionally comprises a background response, totally 10 layers
Response diagram.
Second stage is also to predict each unit response from coloured image, but a series connection layer more than the convolutional layer stage casing
(red concat), unifies following three data：
Convolution results (46*46*32) → textural characteristics of-stage
- previous stage each unit response (46*46*10) → space characteristics
- center constrains (46*46*1)
Result size constancy after series connection, depth are changed into 32+10+1=43.
Phase III does not use original image as input, but it is 128 to take out a depth from the midway of second stage
Characteristic pattern (feature image) as input.It is same to integrate three kinds of factors using series connection layer：Textural characteristics+space characteristics+
Center constrains.
Fourth stage structure is identical with the phase III.When designing more complicated network (such as whole body model), only
Number of components (being changed into 15 from 10) need to be adjusted, and repeats phase III structure.
The heatmap of the image is obtained after image is inputted openpose models, wherein containing the general of each key point
Rate is distributed, and for each key point, chooses prediction coordinate of the position as the key point of its maximum probability.
4th, trunk length computation
After the position of red position and human body key point is obtained, the pixel distance and human body of two red points can be calculated
The length in pixels of trunk, and according to the actual range of red dot laser, the mapping of pixel distance and actual range can be obtained, finally
The physical length of trunk is calculated according to the length in pixels of the mapping and trunk.Trunk physical length Lr's
Calculation formula is：
Wherein Lp represents the length in pixels of trunk, and Dr represents the actual range of red point, Dp represent the pixel of red point away from
From.
The technology contents that the present invention does not elaborate belong to the known technology of those skilled in the art.
Although the illustrative embodiment of the present invention is described above, in order to the technology people of this technology neck
Member understands the present invention, it should be apparent that the invention is not restricted to the scope of embodiment, to the ordinary skill of the art
For personnel, as long as various change, in the spirit and scope of the present invention that appended claim limits and determines, these become
Change is it will be apparent that all utilize the innovation and creation of present inventive concept in the row of protection.
Claims (4)
- A kind of 1. trunk measuring method being combined based on deep learning and red dot laser, it is characterised in that：Including following Step：Step (1)：Two red dot laser transmitters are fixed with video camera first, two beam laser are parallel and throwing with video camera Shadow plane is vertical so that shooting function captures clearly two red points, records the actual range of two red dot laser transmitters； In the step, video camera is demarcated；Step (2)：By the image under the calibrated video camera capture reality scene of step (1)；Step (3)：The image that step (2) captures is input to the red detection module based on opencv, detects and identifies figure The red point of laser as in, if input picture includes the red point that can recognize, two red points of output are in the input image Coordinate position, if input picture does not include the red point that can be recognized, back to step (2)；Step (4)：The image that camara module in step (2) captures is input to the trunk based on deep learning and detects mould Block, detects and identifies the trunk in image, if input picture includes the human body that can be recognized, exports each individual The coordinate position of body key point in the input image, and the position of human body key point is linked, obtain complete trunk； If input picture does not include the human body that can be recognized, back to step (2)；Step (5)：The position for the human body key point that the position of the two red points obtained according to step (3) and step (4) obtain, point Ji Suan not the pixel distance of two red points and the length in pixels of trunk；According to the actual range of red dot laser, pixel is obtained Distance and the mapping of actual range；The actual (tube) length of trunk is calculated according to the length in pixels of the mapping and trunk Degree.
- 2. according to the method described in claim 1, it is characterized in that：The video camera is monocular camera.
- 3. according to the method described in claim 1, it is characterized in that：In step (1) be measured people station apart from video camera 2~ In the range of 10 meters so that video camera can collect complete and clearly human body.
- 4. according to the method described in claim 1, it is characterized in that：Red detection module detects and identifies place's figure in step (3) Specifically included as in the step of red of laser：Input picture is transformed into HSV space from rgb space first, for H, S, V tri- Passage into row threshold division, then carries out region merging technique respectively, obtains the bianry image of red point, finally carries out contours extract, obtain To the coordinate position of red point in the input image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711171226.8A CN107928675A (en) | 2017-11-22 | 2017-11-22 | A kind of trunk measuring method being combined based on deep learning and red dot laser |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711171226.8A CN107928675A (en) | 2017-11-22 | 2017-11-22 | A kind of trunk measuring method being combined based on deep learning and red dot laser |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107928675A true CN107928675A (en) | 2018-04-20 |
Family
ID=61929715
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711171226.8A Pending CN107928675A (en) | 2017-11-22 | 2017-11-22 | A kind of trunk measuring method being combined based on deep learning and red dot laser |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107928675A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108829232A (en) * | 2018-04-26 | 2018-11-16 | 深圳市深晓科技有限公司 | The acquisition methods of skeleton artis three-dimensional coordinate based on deep learning |
| CN109145783A (en) * | 2018-08-03 | 2019-01-04 | 北京字节跳动网络技术有限公司 | Method and apparatus for generating information |
| CN110008818A (en) * | 2019-01-29 | 2019-07-12 | 北京奇艺世纪科技有限公司 | A kind of gesture identification method, device and computer readable storage medium |
| CN110119682A (en) * | 2019-04-04 | 2019-08-13 | 北京理工雷科电子信息技术有限公司 | A kind of infrared remote sensing Image Fire point recognition methods |
| CN110660072A (en) * | 2019-09-20 | 2020-01-07 | 北京神工科技有限公司 | Method and device for identifying straight line edge, storage medium and electronic equipment |
| CN110660073A (en) * | 2019-09-20 | 2020-01-07 | 北京神工科技有限公司 | Straight line edge identification equipment |
| CN111231892A (en) * | 2019-12-29 | 2020-06-05 | 的卢技术有限公司 | Automatic automobile unlocking control method and system based on face and gesture recognition |
| CN112132884A (en) * | 2020-09-29 | 2020-12-25 | 中国海洋大学 | Sea cucumber length measuring method and system based on parallel laser and semantic segmentation |
| CN112308858A (en) * | 2020-12-29 | 2021-02-02 | 南京派光智慧感知信息技术有限公司 | Multidimensional intelligent monitoring method and system for states of railway track and track slab |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102305589A (en) * | 2011-07-28 | 2012-01-04 | 深圳创维－Rgb电子有限公司 | Method and system for laser ranging |
| US20150078617A1 (en) * | 2013-09-13 | 2015-03-19 | Research & Business Foundation Sungkyunkwan University | Mobile terminal and method for generating control command using marker attached to finger |
| CN104858168A (en) * | 2015-04-28 | 2015-08-26 | 国家电网公司 | Water washing robot aiming system and method based on laser ranging and image recognition |
| CN105258640A (en) * | 2015-08-06 | 2016-01-20 | 大理大学 | Multifunctional object measuring instrument and using method thereof |
| CN205619927U (en) * | 2016-04-26 | 2016-10-05 | 西安理工大学 | Provide simple and easy device of engineer's scale in photo |
| CN107180438A (en) * | 2017-04-26 | 2017-09-19 | 清华大学 | Estimate yak body chi, the method for body weight and corresponding portable computer device |
| CN107220604A (en) * | 2017-05-18 | 2017-09-29 | 清华大学深圳研究生院 | A kind of fall detection method based on video |
| US20170315674A1 (en) * | 2016-04-28 | 2017-11-02 | Canon Kabushiki Kaisha | Information processing apparatus, control method for the information processing apparatus, and storage medium |
-
2017
- 2017-11-22 CN CN201711171226.8A patent/CN107928675A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102305589A (en) * | 2011-07-28 | 2012-01-04 | 深圳创维－Rgb电子有限公司 | Method and system for laser ranging |
| US20150078617A1 (en) * | 2013-09-13 | 2015-03-19 | Research & Business Foundation Sungkyunkwan University | Mobile terminal and method for generating control command using marker attached to finger |
| CN104858168A (en) * | 2015-04-28 | 2015-08-26 | 国家电网公司 | Water washing robot aiming system and method based on laser ranging and image recognition |
| CN105258640A (en) * | 2015-08-06 | 2016-01-20 | 大理大学 | Multifunctional object measuring instrument and using method thereof |
| CN205619927U (en) * | 2016-04-26 | 2016-10-05 | 西安理工大学 | Provide simple and easy device of engineer's scale in photo |
| US20170315674A1 (en) * | 2016-04-28 | 2017-11-02 | Canon Kabushiki Kaisha | Information processing apparatus, control method for the information processing apparatus, and storage medium |
| CN107180438A (en) * | 2017-04-26 | 2017-09-19 | 清华大学 | Estimate yak body chi, the method for body weight and corresponding portable computer device |
| CN107220604A (en) * | 2017-05-18 | 2017-09-29 | 清华大学深圳研究生院 | A kind of fall detection method based on video |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108829232B (en) * | 2018-04-26 | 2021-07-23 | 深圳市同维通信技术有限公司 | Method for acquiring three-dimensional coordinates of human skeletal joint points based on deep learning |
| CN108829232A (en) * | 2018-04-26 | 2018-11-16 | 深圳市深晓科技有限公司 | The acquisition methods of skeleton artis three-dimensional coordinate based on deep learning |
| CN109145783A (en) * | 2018-08-03 | 2019-01-04 | 北京字节跳动网络技术有限公司 | Method and apparatus for generating information |
| CN109145783B (en) * | 2018-08-03 | 2022-03-25 | 北京字节跳动网络技术有限公司 | Method and apparatus for generating information |
| CN110008818A (en) * | 2019-01-29 | 2019-07-12 | 北京奇艺世纪科技有限公司 | A kind of gesture identification method, device and computer readable storage medium |
| CN110119682A (en) * | 2019-04-04 | 2019-08-13 | 北京理工雷科电子信息技术有限公司 | A kind of infrared remote sensing Image Fire point recognition methods |
| CN110660073A (en) * | 2019-09-20 | 2020-01-07 | 北京神工科技有限公司 | Straight line edge identification equipment |
| CN110660072A (en) * | 2019-09-20 | 2020-01-07 | 北京神工科技有限公司 | Method and device for identifying straight line edge, storage medium and electronic equipment |
| CN110660072B (en) * | 2019-09-20 | 2022-08-26 | 北京神工科技有限公司 | Method and device for identifying straight line edge, storage medium and electronic equipment |
| CN110660073B (en) * | 2019-09-20 | 2022-10-25 | 北京神工科技有限公司 | Straight line edge recognition equipment |
| CN111231892A (en) * | 2019-12-29 | 2020-06-05 | 的卢技术有限公司 | Automatic automobile unlocking control method and system based on face and gesture recognition |
| CN112132884A (en) * | 2020-09-29 | 2020-12-25 | 中国海洋大学 | Sea cucumber length measuring method and system based on parallel laser and semantic segmentation |
| CN112308858A (en) * | 2020-12-29 | 2021-02-02 | 南京派光智慧感知信息技术有限公司 | Multidimensional intelligent monitoring method and system for states of railway track and track slab |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107928675A (en) | A kind of trunk measuring method being combined based on deep learning and red dot laser | |
| CN110017773B (en) | Package volume measuring method based on machine vision | |
| CA2731680C (en) | System for adaptive three-dimensional scanning of surface characteristics | |
| CN105354819B (en) | Depth data measuring system, depth data determine method and apparatus | |
| CN103959012B (en) | 6DOF position and orientation determine | |
| CN106918299B (en) | A kind of line-structured light machine vision tire wear measurement method | |
| CN103535960B (en) | Human body three-dimensional measurement method based on digital images | |
| CN108113120B (en) | Three-dimensional foot shape and plantar pressure integrated measuring instrument | |
| CN106767562B (en) | A kind of measurement method and human body measurement method based on machine vision and speckle | |
| CN102178530A (en) | Method for automatically measuring human body dimensions on basis of three-dimensional point cloud data | |
| CN205352391U (en) | Entity measurement system based on ground three -dimensional laser scanning | |
| CN110672020A (en) | Stand tree height measuring method based on monocular vision | |
| CN103948196A (en) | Human body data measuring method | |
| CN101611939A (en) | Remote intelligent testing method of three-dimensional human body image data network | |
| CN104408762A (en) | Method for obtaining object image information and three-dimensional model by using monocular unit and two-dimensional platform | |
| CN106705849A (en) | Calibration method of linear-structure optical sensor | |
| CN107607053A (en) | A kind of standing tree tree breast diameter survey method based on machine vision and three-dimensional reconstruction | |
| CN110604574A (en) | Human body height measuring method based on video imaging principle | |
| KR101090082B1 (en) | System and method for automatic measuring of the stair dimensions using a single camera and a laser | |
| CN108805980A (en) | A kind of horse body measurement system and data correcting method based on 3D | |
| CN107084671A (en) | A kind of recessed bulb diameter measuring system and measuring method based on three wire configuration light | |
| JP2013079854A (en) | System and method for three-dimentional measurement | |
| CN111353997B (en) | Real-time three-dimensional surface defect detection method based on fringe projection | |
| CN106384380A (en) | 3D human body scanning, modeling and measuring method and system | |
| CN108175535A (en) | A kind of dentistry spatial digitizer based on microlens array |
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: 20180420 |