CN110009985A - A kind of sand table producing device based on machine learning - Google Patents
A kind of sand table producing device based on machine learning Download PDFInfo
- Publication number
- CN110009985A CN110009985A CN201910239479.7A CN201910239479A CN110009985A CN 110009985 A CN110009985 A CN 110009985A CN 201910239479 A CN201910239479 A CN 201910239479A CN 110009985 A CN110009985 A CN 110009985A
- Authority
- CN
- China
- Prior art keywords
- sand table
- landforms
- data
- prediction model
- unit
- 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
-
- G—PHYSICS
- G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
- G09B—EDUCATIONAL OR DEMONSTRATION APPLIANCES; APPLIANCES FOR TEACHING, OR COMMUNICATING WITH, THE BLIND, DEAF OR MUTE; MODELS; PLANETARIA; GLOBES; MAPS; DIAGRAMS
- G09B25/00—Models for purposes not provided for in G09B23/00, e.g. full-sized devices for demonstration purposes
-
- G—PHYSICS
- G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
- G09F—DISPLAYING; ADVERTISING; SIGNS; LABELS OR NAME-PLATES; SEALS
- G09F19/00—Advertising or display means not otherwise provided for
- G09F19/12—Advertising or display means not otherwise provided for using special optical effects
- G09F19/18—Advertising or display means not otherwise provided for using special optical effects involving the use of optical projection means, e.g. projection of images on clouds
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Educational Administration (AREA)
- Educational Technology (AREA)
- Accounting & Taxation (AREA)
- Marketing (AREA)
- Holo Graphy (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The present invention relates to a kind of sand table producing device based on machine learning, belongs to sand table manufacture technology field, solves the problems, such as that existing sand table fabrication cycle is long, shows and lack authenticity and can not be adjusted in real time.Include: controlling terminal, for constructing prediction model, and seeks laser hologram projector parameter and complete relief data in real time;Laser hologram projector carries out sand table projection imaging;Mechanical arm carries out the production of sand table landforms;Binocular camera, laser ranging inductor, spatial digitizer, for data needed for collection model training.Laser hologram projection can really show the state of landforms in the present apparatus, and mechanical arm is quickly piled up and adjusted to landform;It can seek optimal projective parameter and geomorphology information in real time using the trained model of machine learning, automatic correction and adjustment, the sand table of production in real time are carried out to projection state and landforms and reach optimal projection size and effect.
Description
Technical field
The present invention relates to sand table manufacture technology field more particularly to a kind of sand table producing devices based on machine learning.
Background technique
Military sand table generally comprises two main parts: one is bottom tray, one be mountains and rivers river simulation product.
Production about mountains and rivers river is at present based on two kinds of materials, and one is plastic simulation product, and one is sands.
Sand table based on plastic simulation product needs to take the form of specific customization, carries out 3D figure to true landforms first
The drafting of paper then carries out the shape of 3D printing or injection molding to drawing, simulation product production is worked it out, then these simulation products are put
On pallet, for showing landforms to other people.This sand table, on the one hand, the preparatory period is longer, and the preparation of landforms needs to pass through 3D
The mode of printing or injection molding, needs to be prepared scene;In the state of burst, the preparation of landforms is carried out without the time,
Landforms can not adjust in time, be unable to satisfy the case where tracking to the variation of real scene.On the other hand, display area by
Limit, since landforms are all that 3D printing or mould-injection are made, to the size of plastic parts by certain requirement, and for super
Conventional size out, can also accordingly increase, while lacking reusability after use in expense, each landforms be it is unique,
Existing landforms can not be applied to other areas and be shown, and cause to waste.
Sand table based on sand covers required sand on pallet first, then according to existing drawing to each
Feature measures and marks (such as mountain range, river), and finally sand is piled up or digged pit according to mark point, is piled up
Part indicates mountains and rivers, and the part digged pit indicates rivers and creeks.This sand table, on the one hand, prepare and fabrication cycle is longer, need to true
Landforms carry out the scaling of equal proportion, and be labeled in sand table;Meanwhile the whole process of production requires manually to carry out
The production of landforms, such as sand pile up expression mountain range, and sand, which is digged pit, indicates lake.On the other hand, in scene service stage, first
Lack accuracy, same exhibition method can represent different landforms, be easy that people is allowed to give rise to misunderstanding, for example, by using the shape in sand pit
Formula can indicate lake and hawthorn landform simultaneously, need to mark to distinguish;Next lacks authenticity, and the color of sand is single,
It can not show true scene;Finally adjustment is inconvenient, and the change of landforms in use needs manually to be adjusted sand,
According to the increase of the complexity of variation and size, difficulty increases, inefficiency.
Summary of the invention
In view of above-mentioned analysis, the present invention is intended to provide a kind of sand table producing device based on machine learning, to solve
Existing sand table fabrication cycle is long, shows the problem of lacking authenticity and can not being adjusted in real time.
The purpose of the present invention is mainly achieved through the following technical solutions:
Provide a kind of sand table producing device based on machine learning, comprising:
Controlling terminal for constructing laser hologram projection prediction model and geomorphology information prediction model, and is trained, leads to
It crosses trained model and seeks laser hologram projector parameter and complete relief data in real time;
Laser hologram projector carries out sand table projection imaging according to the numerical value of above-mentioned projective parameter;
Mechanical arm carries out the production of sand table landforms according to above-mentioned complete relief data;
Binocular camera, laser ranging inductor, hang on the position directly above of sand table, and with controlling terminal phase
Even, for data needed for acquiring laser hologram projection prediction model training;
Spatial digitizer, for being scanned to true landforms, number needed for acquiring above-mentioned geomorphology information prediction model training
According to.
The present invention has the beneficial effect that: present apparatus sand table manufacturing process quickly and efficiently, is thrown by machine learning, laser hologram
Shadow, mechanical arm production can effectively solve the problems, such as the instantaneity and authenticity questions of traditional sand table;Laser hologram projection can be more
Add the true state for showing simulation landforms (such as mountains and rivers river etc.), quickly landform can be piled up using mechanical arm
And adjustment.Simultaneously by constantly monitoring external environment, when sand table size changes with external show surroundings, engineering is utilized
Optimal projective parameter and geomorphology information can be sought in real time by practising trained model, and the state and landforms to projection carry out automatic
It corrects and adjustment, the sand table produced reaches optimal projection size and effect.
On the basis of above scheme, the present invention has also done following improvement:
Further, the mechanical arm, comprising: mobile arm, chassis, laser distance inductor, Dextrous Hand;
One end of the mobile arm is equipped with Dextrous Hand, and the other end is mounted on chassis, mobile arm with chassis
Tie point is axle center, is rotated;
The chassis is equipped with moving assembly, and the position for mechanical arm is mobile;
The Dextrous Hand is rotated using the tie point with mobile arm as axle center, is piled up for landforms;
The quantity of the laser distance inductor is at least two, is separately mounted in Dextrous Hand and chassis, for moving
Barrier monitors in the process.
Further, the controlling terminal includes:
Data acquisition module, for number needed for obtaining trained laser hologram projection prediction model and geomorphology information prediction model
According to collection;
Prediction model building and training module, for constructing laser hologram projection sand table prediction model and geomorphology information prediction
Model, and two models are trained respectively using the data set that the data acquisition module obtains, it obtains trained sharp
Light line holographic projections sand table prediction model and geomorphology information prediction model;
Projective parameter and complete relief data seek module, for by the length of sand table, wide, high, depth information to be input to
Trained laser hologram projection prediction model is stated, projective parameter is obtained;It is also used to the profile of landforms being input to above-mentioned training
Good geomorphology information prediction model, obtains complete relief data;
Projective parameter and complete relief data are sought the projective parameter that module is sought and are transmitted to laser by signal transmission module
Holographic projector;The complete relief data sought is transmitted to mechanical arm simultaneously.
It further, further include that there is " ten " word cursor, the truing tool for position correction;The data acquisition module packet
Include: area division unit, length and width dimensions collect unit, depth size collects unit;
The area division unit, for being divided to sand table interior zone, and by the friendship of all subregion marked off
Point is used as datum mark;
The length and width dimensions collect unit, depth size collects unit, for incuding with binocular camera, laser ranging
Device, laser hologram projector and truing tool cooperation, carry out data acquisition to each datum mark, obtain number required for training pattern
According to collection;
The data set includes plan-position coordinate, vertical position coordinate, focusing range and the clarity of datum mark.
Further, the data acquisition module further include: profile scan unit, key point picking unit, data supplement are single
Member;
The profile scan unit, receive spatial digitizer scanning landforms profile, and to the turning occurred in profile into
Row is collected;
The key point picking unit grabs length, width, the height of landforms key point according to the position at above-mentioned turning
Information;
The data supplementary units, for being decomposed again to the part between adjacent key point, obtain decomposite it is each
Partial length, width, altitude information, obtain complete relief data, for training geomorphology information prediction model.
It is further, described that data acquisition is carried out to each datum mark, comprising:
Length and width dimensions collect unit and control laser hologram projector projects " ten " word cursor, and " ten " word cursor is successively projected
In the position of each datum mark, the position of truing tool is adjusted, until the position of " ten " word of " ten " word and cursor in truing tool
Set overlapping, length and width dimensions collect the plan-position coordinate of unit record datum mark at this time, and the length and width dimensions for completing sand table are collected;
Depth size collects unit and controls laser hologram projector projects " ten " word cursor, and " ten " word cursor is successively projected
In the position of each datum mark, in the state of keeping the position overlapping of " ten " of " ten " word and cursor in truing tool, upwards/to
Lower mobile truing tool reaches optimal display effect, the depth until being somebody's turn to do " ten " word cursor in binocular camera display interface
Size collects vertical height, focusing range and the sharpness information of the unit record position, the upright position as the datum mark
Coordinate, focusing range and clarity are completed sand table depth size and are collected.
Further, the prediction model building and training module, comprising:
Parameter initialization unit carries out initialization process to network weight and biasing, and concentrates at random in training data
Select the first input sample;
Parameter processing unit carries out neuronal activation forward-propagating, is carried out by hidden layer to above-mentioned first input sample
Weight and bias treatment, and seek the result and error of output layer;
Parameter adjustment unit carries out backpropagation according to above-mentioned error, is adjusted to network weight and biasing;
Whether training monitoring unit, terminate according to preset termination condition training of judgement.
Further, the projective parameter and complete relief data seek module, comprising:
High complexity zone marker unit, the region in complexity high in true landforms region and current sand table is compared
It is right, the point region that high complexity is met in sand table is marked;
Weight determining unit determines position, the quantity, the power of requirement degree in the point region of the above-mentioned high complexity marked
Weight;
Projective parameter seeks unit, according to the weight of the length of sand table, depth information and above-mentioned determination, utilizes training
Good model seeks focusing range and clarity.
Further, the projective parameter and complete relief data seek module, in length, the depth letter for obtaining sand table
It after breath, scans for matching in the existing data set of building, when matching consistent, then directly chooses corresponding focusing range
And sharpness information then is sought focusing model when matching inconsistent using trained laser hologram projection sand table prediction model
It encloses and sharpness information;It after obtaining landforms profile, scans for matching in the existing database of building, when matching is consistent
When, then directly choose corresponding complete relief data;When matching inconsistent, then asked using trained geomorphology information prediction model
Take whole relief data.
Further, described device further includes landforms monitoring sensor, for carrying out real-time monitoring to true geomorphology information
And controlling terminal is passed to, when controlling terminal judges that landforms change, by the interval time of landforms variation and preset benchmark
It compares, when the interval time of landforms variation being less than benchmark, then ignores;It is greater than base when the interval time of landforms variation
On time, on the one hand, projective parameter and complete relief data seek module and carry out weight to landforms whole region or high complexity region
New parameter prediction obtains new focusing range and clarity parameter;On the other hand, projective parameter and complete relief data are sought
Module carries out the collection of key point to new landforms region of variation and landforms supplement, and obtains complete relief data.
It in the present invention, can also be combined with each other between above-mentioned each technical solution, to realize more preferred assembled schemes.This
Other feature and advantage of invention will illustrate in the following description, also, certain advantages can become from specification it is aobvious and
It is clear to, or understand through the implementation of the invention.The objectives and other advantages of the invention can by specification, claims with
And it is achieved and obtained in specifically noted content in attached drawing.
Detailed description of the invention
Attached drawing is only used for showing the purpose of specific embodiment, and is not to be construed as limiting the invention, in entire attached drawing
In, identical reference symbol indicates identical component.
Fig. 1 is the sand table producing device structure chart based on machine learning in the embodiment of the present invention;
Fig. 2 is sand table region division and truing tool schematic diagram in the embodiment of the present invention;
Fig. 3 is to carry out relief data in the embodiment of the present invention to collect schematic diagram;
Fig. 4 is neural network structure schematic diagram in the embodiment of the present invention;
Fig. 5 is to be trained flow chart to the prediction model of building in the embodiment of the present invention;
Fig. 6 is to seek projective parameter flow chart in the embodiment of the present invention;
Fig. 7 is to seek relief data schematic diagram in the embodiment of the present invention;
Fig. 8 is key point amalgamation result schematic diagram in the embodiment of the present invention;
Fig. 9 is laser-projector schematic illustration in the embodiment of the present invention;
Figure 10 is mechanical arm structural schematic diagram in the embodiment of the present invention.
Specific embodiment
Specifically describing the preferred embodiment of the present invention with reference to the accompanying drawing, wherein attached drawing constitutes the application a part, and
Together with embodiments of the present invention for illustrating the principle of the present invention, it is not intended to limit the scope of the present invention.
A specific embodiment of the invention discloses a kind of sand table producing device based on machine learning, such as Fig. 1 institute
Show, controlling terminal, for constructing laser hologram projection prediction model and geomorphology information prediction model, and is trained, passes through instruction
The model perfected seeks laser hologram projector parameter and complete relief data in real time;
Laser hologram projector carries out sand table projection imaging according to the numerical value of above-mentioned projective parameter;
Mechanical arm carries out the production of sand table landforms according to above-mentioned complete relief data;
Binocular camera, laser ranging inductor, hang on the position directly above of sand table, and with controlling terminal phase
Even, for data needed for acquiring laser hologram projection prediction model training;
Spatial digitizer, for being scanned to true landforms, number needed for acquiring above-mentioned geomorphology information prediction model training
According to.
Compared with prior art, the sand table producing device provided in this embodiment based on machine learning, phase between each equipment
Mutually cooperation, sand table manufacturing process quickly and efficiently, can be solved effectively by machine learning, laser hologram projection, mechanical arm production
The instantaneity problem and authenticity questions of traditional sand table;Laser hologram projection can more really show simulation landforms (such as
Mountains and rivers river etc.) state, can quickly landform be piled up and be adjusted using mechanical arm.It is outer by constantly monitoring simultaneously
Portion's environment can be asked when sand table size changes with external show surroundings using the trained model of machine learning in real time
Optimal projective parameter and geomorphology information are taken, the state and landforms to projection are corrected and adjusted automatically, the sand table produced
Reach optimal projection size and effect.
Specifically, the present apparatus give full play to artificial intelligence, laser hologram shadow casting technique, mechanical arm production advantage, control
Terminal processed is automatic by machine learning and immediately obtains optimal parametric solution, and laser hologram projects more true simulation shows
Actual scene, mechanical arm replace artificial automatic progress landform production and adjustment.Controlling terminal includes: data acquisition module, prediction
Model construction and training module, projective parameter and complete relief data seek module, signal transmission module;Wherein,
Data acquisition module, for number needed for obtaining trained laser hologram projection prediction model and geomorphology information prediction model
According to collection;
Prediction model building and training module, for constructing laser hologram projection sand table prediction model and geomorphology information prediction
Model, and be trained using the data set that the data acquisition module obtains, obtain trained laser hologram projection sand table
Prediction model and geomorphology information prediction model;
Projective parameter and complete relief data seek module, for by the length of sand table, wide, high, the information such as depth to be input to
Above-mentioned trained laser hologram projects prediction model, obtains projective parameter (focusing range and definition values etc.);It is also used to, it will
The profile of landforms is input to above-mentioned trained geomorphology information prediction model, obtains complete relief data;
Signal transmission module is used for laser hologram projector, binocular camera, laser ranging inductor, mechanical arm, three-dimensional
Signal transmission between scanner and controlling terminal, seeks the focusing range that module is sought for projective parameter and complete relief data
It is transmitted to laser hologram projector with projective parameters such as definition values, carries out projection imaging;The complete landforms that will be sought simultaneously
Data are transmitted to mechanical arm, carry out piling up and adjusting for landforms.
Further, data acquisition module is divided into two parts: laser projection data-acquisition submodule, landforms production data acquisition
Submodule;Data set needed for laser projection data-acquisition submodule is used to acquire trained laser hologram projection sand table prediction model,
Landforms production data-acquisition submodule is for data set needed for acquiring trained geomorphology information prediction model.
Laser projection data-acquisition submodule includes: area division unit, length and width dimensions collect unit, depth size is collected
Unit, each unit are counted by cooperating with laser hologram projector, binocular camera, laser ranging inductor, truing tool
According to acquisition, obtain the data set containing mass data (this data set has target value).When carrying out real data acquisition:
Area division unit divides sand table interior zone, and using the intersection point of all subregion marked off as datum mark;Binocular
Camera, laser ranging inductor and laser hologram projector, lock the overall region of sand table, and position above-mentioned all benchmark
Point;Finally, length and width dimensions collect unit and depth size collect unit control respectively binocular camera, laser ranging inductor and
Laser hologram projector, and cooperate truing tool, data acquisition successively is carried out to each datum mark, is obtained required for training pattern
Data set.
Specifically, area division unit divides sand table interior zone, and by the intersection point of all subregion marked off
As datum mark.In the present embodiment by taking rectangle sand table as an example, as shown in Fig. 2, the sand table frame of pre-production is set (not prefabricated
When sand table frame, can have been built with brick clay or level land under dig 15~25 centimetres), and it is consistent with existing place as far as possible, spread in sand table frame
3-5CM humidity sandy soil appropriate (are picked up sandy soil with hand, agglomerating to have held, leaving behind to dissipate is advisable), and real with plank drag concora crush,
Datum level as sand table.Area division unit divides in sand table region automatically, and division mode can be according to the shape of sand table
Shape and projection show that required precision is selected, and illustratively, rectangle sand table are separated 20 sub- squares according to the matrix etc. of 4*5
Shape, and using 12 intersection points of rectangle inside sand table as datum mark.
After completing region division, can cover sand table position installation binocular camera, laser ranging inductor and
Laser hologram projector (is being typically mounted at rectangle sand table central point just using the overall region of binocular camera locking sand table
Top), and above-mentioned 12 datum marks are oriented, controlling terminal control laser hologram projector (is typically mounted at and sand table plane
The position at angle at 45 °) so that the projection clarity in sand table datum level is best, and cooperate truing tool, carries out sand table size number
According to (including: length and width dimensions, depth size) is collected and recorded, summarize to obtain " -12 position-degree of focus of sand table size/clear
Between degree " data set.
Length and width dimensions collect unit, with binocular camera, laser ranging inductor, laser hologram projector and truing tool
Cooperation, the length and width dimensions for completing sand table are collected;Specifically, laser hologram projector projects " ten " word cursor is controlled, by " ten " word
Cursor is incident upon the position of " 1 " point, and cursor color to be calibrated at this time is red.Calibration personnel holds truing tool, will
The position of " ten " word of " ten " word and cursor in truing tool is overlapped, and the focal length value of binocular camera is kept fixed, control
Laser ranging inductor processed (being mounted on right above sand table) is calibrated, the position coordinates of record " 1 " point (using sand table surface as
Datum level).It should be noted that further being calibrated using laser ranging inductor, can correct on local detail
Dimension data, so that the location information of the benchmark of acquisition is more accurate;It is finished if the position of " 1 " point carries out calibration, by light
Mark color is changed to green, indicates that the position of the point has been collected and finishes.When " 1 " point position correction finish, then control laser
Holographic projector " ten " word cursor is moved to the position of " " at 2 points, and calibrating mode is same as above, and so on, until 12 datum marks
Data are all collected and are finished, then the length and width dimensions collection of sand table finishes.
Depth size collects unit, on the basis of completing the collection of length and width dimensions, throws with binocular camera, laser hologram
Shadow instrument and truing tool cooperation, carry out the collection of sand table height and depth.Controlling terminal controls laser hologram projector projects
" ten " word cursor, is successively incident upon the position of each datum mark by " ten " word cursor, and operator keeps " ten " in truing tool
The state of the position overlapping of " ten " of word and cursor, the mobile truing tool of up/down, until in binocular camera display interface
In should " ten " word cursor reach optimal display effect, depth size collect the unit record position vertical height (relative to
Datum level), focusing range and sharpness information, as the vertical position coordinate of the datum mark, focusing range and clarity.
In specific work process, the collection of depth size can be divided into height collection and collect two kinds of situations with depth, carry out
When height is collected: depth size collects unit control laser-projector and projects " ten " word cursor, and " ten " word cursor is incident upon " 1 "
The position of point, and cursor color to be calibrated at this time is red.Calibration personnel holds truing tool, will be in truing tool
The position of " ten " of " ten " word and cursor is overlapped, and is moved up simultaneously, pay attention to during moving up require " ten " word and
Cursor " ten " keeps the state of overlapping.As " ten " in truing tool are constantly raised, in the position of " 1 " point, pass through binocular
Camera carries out calibration positioning, searches out optimal display effect (display is clearest) in height, obtains the height value of the point
The as height coordinate (binocular camera automatic positioning) of " 1 " point;Depth size collects unit for cursor color from red at this time
Become green, indicates the data collection for completing " 1 " point height.It collects and finishes when the position height of " 1 " point, then it is mobile to control cursor
Want the position of " " at 2 points, collection mode is same as above, and so on, it all collects and finishes until the data of 12 points, then the height of sand table
Collection finishes.When carrying out depth collection: depth size collects unit control laser-projector and projects " ten " word cursor, by " ten "
Word cursor is incident upon the position of " 1 " point, and cursor color to be calibrated at this time is red.Operator digs in " 1 " point region
One dell then holds truing tool, the position of " ten " word of " ten " word and cursor in truing tool is overlapped, together
When to dell bottom direction, move down, pay attention to requiring " ten " word and cursor " ten " to keep overlapping during moving down
State.As " ten " in truing tool are constantly moved down, by binocular camera in the position of " 1 " point, found in depth
To optimal display effect.Depth size collects unit and cursor color is become green from red at this time, indicates that completion " 1 " point is deep
The data collection of degree.When the depth collection of " 1 " point finishes, then cursor is moved to the position of " " at 2 points, and collection mode is same as above,
And so on, it all collects and finishes until the data of 12 points, then the depth collection of sand table finishes;Obtaining above-mentioned height or depth
After spending signal, depth size collects unit combination laser-projector in datum level focusing range and clarity, and it is optimal to seek this
The focusing range and sharpness information of display position, focusing range and clarity as corresponding datum mark.In addition, when laser is thrown
When shadow instrument carries out true environment displaying, in the feux rouges, green light and blue light information of above-mentioned each datum mark, data collection under also can record
Statistics is as shown in table 1.
1 data collection statistical form of table:
Landforms make data-acquisition submodule, cooperate with spatial digitizer, carry out data acquisition, obtain one containing a large amount of
The data set of data (this data set has target value);During carrying out actual acquisition, it can be carried out according to existing landforms
The foundation of model directly acquires existing sand table landform model, such as the data on mountain range, river.Specific data such as table 2, table 3
It is shown;After spatial digitizer scans landforms profile, it is sent to landforms production data-acquisition submodule, which includes: profile
Scanning element, key point picking unit, data supplementary units;Wherein, profile scan unit: the turning occurred in profile is carried out
It is automatic to collect;Key point picking unit: according to the position at above-mentioned turning, the length on landforms key point (vertex at each turning) is grabbed
Degree, width, elevation information;Data supplementary units: as shown in figure 3, according to true Geomorphological States, by the length of landforms, width,
Height is further decomposed into tiny part (can be decomposed according to the requirement of the complexity and displaying precision of practical landforms),
Further, the length of thin between each turning, width are collected, the data of height obtain more detailed landforms number
According to as the landforms sample on mountain range, so that prediction model is learnt.
The data on 2 mountain range of table
The data in 3 river of table
Prediction model building and training module, including prediction model construction unit project prediction model for laser hologram
And the building of geomorphology information prediction model;It is real when sand table size and external environment change by the model of building
When obtain optimal laser hologram projective parameter data, reach optimal display effect;It, can be quick when landforms change
The profile of landforms is analyzed, be automatically replenished details therein, obtain complete relief data.Prediction model construction unit
Two prediction models of building are all made of the machine learning algorithm of supervised, can be constructed and be predicted by a variety of machine learning algorithms
Model, in the present embodiment, the neural network of building is as shown in figure 3, in the model can be automatic by input sand table size
Obtain degree of focus/clarity numerical information;Input landforms profile can automatically derive complete relief data.
Prediction model building and training module further include: parameter initialization unit, parameter processing unit, parameter adjustment are single
Member, training monitoring unit;The training dataset obtained using data acquisition module two that prediction model construction unit is constructed
Prediction model is trained respectively, in order to which the value for returning to prediction model is right as far as possible in target value, training process
The weight of feature does continuous adjustment, until the value of prediction model return meets required precision;Specifically, as shown in figure 4,
Parameter initialization unit, in initial phase, to network weight and biasing one random number of setting, and in above-mentioned number
First input sample is selected at random according to concentrating.In machine-learning process, the part of this network weight and biasing can be with
Difference between actual value and target value is constantly adjusted.
Parameter processing unit carries out neuronal activation forward-propagating, carries out weight to above-mentioned input sample by hidden layer
And bias treatment, and seek the result and expected error of output layer;In the neural network of Fig. 3,Indicate the i-th of l layers
The weight that l layers are denoted as between Ll, L1 and L2 layers is by the weight between a node and l+1 layers of j-th of node
Weight between L2 and L3 layers isRepresent the bias term of l+1 i-th of node of layer;WithIndicate l+1 layers
The input value of j-th of node.As l=1,Represent l+1 j-th of node of layer
Output valve after activation primitive θ (x).Formula is as follows:
This completes primary training, have obtained output result hW,b(x)。
Parameter adjustment unit carries out backpropagation according to error, and adjusts network weight and biasing.Wherein, for output
Layer, Errj=Oj(1-Oj)(Tj-Oj) for hidden layer, Errj=Oj(1-Oj)∑Errkwjk;Weight updates, Δ Wij=(l)
ErrjOj, Wij'=Wij+ΔWij;Biasing updates, Δ θj=(l) Errj, θj=j+ Δ θj。
Whether training monitoring unit, training of judgement terminate, and according to pre-set trained termination condition, whether training of judgement
It completes, illustratively, when the update of weight is lower than some threshold value, the error rate of prediction is lower than some threshold value, it is default certain to reach
The conditions such as cycle-index when, terminate training, otherwise continue to choose new sample being trained.
Projective parameter and complete relief data seek module, can further be divided into: laser hologram projector parameter is asked
Submodule, landforms partial data is taken to seek submodule;Wherein,
Laser hologram projector parameter seeks submodule, after inputting the length of sand table, depth information, utilizes instruction
The model perfected seeks laser imager imaging parameters, as shown in figure 5, when the length of sand table, depth are changed
It waits, it is only necessary to which the length of the dimension information of typing sand table or 12 datum marks, depth data, laser hologram projector are thrown
Shadow parameter seeks submodule can calculate the parameter of focusing range and clarity automatically, for example, in table 1 x2 and y2 parameter;And it will
Projective parameter (feux rouges, green light, blue light, focusing range, clarity) is packaged, and is transferred to laser hologram by signal transmission module
Projector, laser-projector are projected on sand table according to the supplemental characteristic of projection.
In order to improve the authenticity of laser projection sand table displaying, degree of focus and clarity are being obtained by trained model
Before equal projective parameters, laser hologram projector parameter, which seeks submodule, can choose the region of high complexity and be added
Power, to improve the prediction effect of prediction model, the submodule is further can include: high complexity zone marker unit, Gao Fu
Weight determining unit, the projective parameter in miscellaneous degree region seek unit;Specifically,
High complexity zone marker unit will need the region of high complexity and current sand table in true landforms to be simulated
Region is compared, and the point region that high complexity is met in sand table is marked.Such as high complexity region is 1 point,
Or 4,5 points ... (note, if required to some specific region, before processing terminal carries out automatic comparison, people
It is high complexity region to mark this region).
Weight determining unit determines the position in the point region of high complexity, quantity, it is desirable that the key factors weight such as degree.
In view of the complexity in each region in simulation true environment is distinct, need to integrate corresponding key factor to confirm that clear
Clear degree and focusing range require some highest point or certain several point, reach optimal drop shadow effect.
Projective parameter seeks unit, wide according to the length of sand table, high, the weight of depth and determination, using above-mentioned trained
The value of prediction model, the focusing range and clarity that included to the region covered is handled, and finds most suitable focusing
Range and definition values.For example, if the region selected after weight for point 1 and point 2 position, as shown in table 4, then to X1,
X2 and Y1, Y2 value are handled, and corresponding X value and Y value are obtained.
4 focusing range of table and clarity are sought
It should be noted that in order to improve real-time, it is wide in the length of sand table, it is high, when depth changes, projection
Parameter seeks unit and scans for matching in the existing database of building first, when matching consistent, then directly selection pair
The focusing range and sharpness information answered.
Landforms partial data seeks submodule, using the geomorphology information prediction model being trained to, when landforms occur
When variation, the profile of landforms can be analyzed, select the data of several key points, in the geomorphology information database of above-mentioned foundation
Middle crawl needs the relief data that presents, or according to landforms information prediction model, is automatically replenished details therein, reach with very
The effect that real geomorphological environment is no different.
It is carrying out in actual work, in order to improve real-time, which can carry out the key point of extraction further
Screening, reduce the input of unnecessary key point.Specifically, it is first determined the key point of profile is right in the profile scanned
The turning of appearance is collected, using the vertex at each turning as key point;As shown in fig. 7, the coordinate for point 1 to point N carries out
It collects;Since first turning, the distance between all adjacent corners, such as distance 1-2, distance 2-3 ... are calculated;So
Afterwards, key point is screened according to above-mentioned distance, landforms partial data seeks submodule and presets a distance value x as base
Above-mentioned all distances are compared by standard with this benchmark, if being less than this reference value, are abandoned, if more than this benchmark
Value is retained;For example, distance 4-5 is less than a reference value x, then puts 5 and be abandoned.Distance 4-6 then is taken, then is carried out and a reference value x
It is compared, if distance 4-6 is less than a reference value x, abandon ... and so on.After screening, obtained key point, which exists, not to be connected
When continuous situation, processing is merged to key point, corresponding processing is done for discontinuous part and being merged.Such as Fig. 8
In, 5 points are abandoned, then point 4 and point 6 carry out processing on x and y-axis and obtain new 6 position of point.After merging,
It resequences to all point positions, as shown in figure 8, for example new 6 position of point is then changed to a little 5, connects after arrangement in sequence
Continuous key point is prepared for adjustment next time.Finally, each key point after merging to screening is input to trained landforms
Information prediction model automatically supplements the details lacked in the profile between each key point, using prediction model to missing
X2, the data of x4, x6, x8, h2 are supplemented;And the complete relief data obtained after supplement is transmitted into mould by receiving signal
Block is sent to mechanical arm in real time.
Laser hologram projector, the projective parameter numerical value that real-time reception signal transmission module is sent carry out sand table and project into
Picture.As shown in figure 9, laser hologram projector uses tricolor laser technology in the present embodiment, transmitted using laser beam
Picture, wherein the optical component of laser-projector is mainly by redgreenblue light valve, conjunction beam X prism, projection lens and driving light
Valve.There is three color laser of red, green, blue in laser hologram projector.Laser passes through corresponding optical element and processing in machine
It is transmitted to X prism again after the expanding of chip to integrate three beams of laser, is then again transmitted to the laser after integration by projection objective
On projection screen, completes entire laser hologram projector and show process.
Mechanical arm, the complete relief data that real-time reception signal transmission module is sent, mechanical arm is according to complete landforms number
According to quickly landform is piled up and is adjusted, to save the time that manually landforms establish with model.Mechanical arm, such as
Shown in Figure 10, comprising: mobile arm, chassis, laser distance inductor, Dextrous Hand;Wherein, one end of mobile arm is equipped with spirit
Dab hand, the other end are mounted on chassis, and mobile arm carries out all-directional rotation using the tie point with chassis as axle center;Chassis peace
Equipped with moving assembly (such as wheel);All-directional rotation may be implemented in Dextrous Hand;The quantity of laser distance inductor is at least two,
It installs on Dextrous Hand and chassis respectively, for barrier monitoring in moving process.
When specific works, firstly, determining the basic point of mechanical arm: before mechanical arm work, mechanical arm being placed on sand table periphery
Some position, using the coordinate typing controlling terminal of this position as the position basic point of mechanical arm.Then, it is determined that mechanical arm
Shift position, controlling terminal carry out the calculating of complexity for entire sand table landforms, are arranged according to the height of complexity, and
The position in complexity region successively (complexity is by high -> low) is sent to mechanical arm.Finally, mechanical arm passes through chassis moving portion
Point, the regional location sequence according to the high low complex degree of regional location-of the high complexity of base position-is mobile;In mobile process
In, barrier is measured by ranging inductor, prevents from colliding, when being often moved to a designated position, is held by Dextrous Hand
Row piles up task, piles up and finishes until all position landforms, then mechanical arm playbacks, and is moved to original base position, so far,
Mechanical arm completion landforms pile up work.
It should be noted that during landforms are piled up, can more sand table be carried out efficiently and accurately using Dextrous Hand
It piles up.Dextrous Hand is equipped with laser away from inductor, for hiding obstacle, when piling up, and the process that is moved out of sand table extroversion sand table
In, constantly whether there are obstacles in detection front, if the mountain range ... that heap puzzles stops moving if detecting barrier, adjusts
Whole dexterity hand position, is changed to move up, and is further continued for mobile to target position.
It needs to illustrate when landforms change, to carry out, it is emphasized that poor in order to solve traditional sand table instantaneity
The problem of adjusting in real time, can not reflecting true landforms.Device further includes (illustratively, landforms monitor sensor in the present embodiment
Video or Image Acquisition transmission device), landforms monitor sensor and carry out real-time monitoring to true geomorphology information and pass to control
Terminal processed, controlling terminal judge whether geomorphology information changes, and when landforms change, are according to preset condition judgement
It is no that the geomorphology information of building and projection are updated;Specifically, preset on the basis of a time interval that (time interval can
Set according to actual required precision, hardware condition, sand table scale etc.), when monitoring that landforms change, by landforms
The interval time of variation is compared with benchmark, if the interval time of landforms variation is less than benchmark, ignores;If landforms become
The interval time of change is greater than benchmark, on the one hand, projective parameter seek submodule to landforms whole region or high complexity region into
The parameter prediction of row again, obtains new focusing range and clarity parameter, and be sent to laser hologram projector and projected
Imaging;On the other hand, landforms partial data seeks the collection that submodule carries out key point to new landforms region of variation, carries out ground
Looks data supplement, and send and piled up to mechanical arm;
Consider that mechanical arm piles up required time cost to landform again, is greater than benchmark when the interval time of landforms variation, it is right
After new landforms carry out the collection of key point, controlling terminal takes different modes of piling up according to the degree that landforms change, specifically
The new key point collected is carried out intermittently error range (specific number by ground with the key point parameter collected recently
Value can be set according to conditions such as practical bandwagon effect, hardware device, requirement of real-time) it is compared;When the key point of update is joined
When number is greater than this error range, then the mode of " refigure " is taken, when the key point parameter of update is less than this error
When range, then take the mode of " small range amendment ".
Wherein, under refigure mode, landforms can be smoothed out refigure again, and mechanical arm first can be according to initial parameter
(the height Z coordinate under original state, X, Y axis coordinate) smoothes out the landforms in the region for initial state;Then using seeking
Newest geomorphology information, pile up again (execute step S202).Under small range modification model, further become with landforms
The region of change is reference, and the collection of key point, the newest landforms sought are carried out to a certain range of profile of this area peripheral edge
Information is piled up again.
The present apparatus increases artificial intelligence, laser hologram imaging technique, the production of mechanical arm landforms in military sand table production, can
To solve the problems, such as brought by current sand table;Laser hologram imaging technique and the fast automatic production of mechanical arm landforms can solve standard
The problem of true property and authenticity.Again on the basis of this, the present apparatus also has range measurement, the interaction functions such as position mark, by swashing
Ligh-ranging inductor, the distance between two o'clock in available sand table, and point-of-interest is demarcated, it can obtain immediately
Information on more battlefields.
It will be understood by those skilled in the art that realizing all or part of the process of above-described embodiment method, meter can be passed through
Calculation machine program instruction relevant hardware is completed, and the program can be stored in computer readable storage medium.Wherein, described
Computer readable storage medium is disk, CD, read-only memory or random access memory etc..
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
In the technical scope disclosed by the present invention, any changes or substitutions that can be easily thought of by anyone skilled in the art,
It should be covered by the protection scope of the present invention.
Claims (10)
1. a kind of sand table producing device based on machine learning characterized by comprising
Controlling terminal for constructing laser hologram projection prediction model and geomorphology information prediction model, and is trained, passes through instruction
The model perfected seeks laser hologram projector parameter and complete relief data in real time;
Laser hologram projector carries out sand table projection imaging according to the numerical value of above-mentioned projective parameter;
Mechanical arm carries out the production of sand table landforms according to above-mentioned complete relief data;
Binocular camera, laser ranging inductor hang on the position directly above of sand table, and are connected with controlling terminal, use
The data needed for acquisition laser hologram projection prediction model training;
Spatial digitizer, for being scanned to true landforms, data needed for acquiring above-mentioned geomorphology information prediction model training.
2. the apparatus according to claim 1, which is characterized in that the mechanical arm, comprising: mobile arm, chassis, laser away from
From inductor, Dextrous Hand;
One end of the mobile arm is equipped with Dextrous Hand, and the other end is mounted on chassis, and mobile arm is with the connection with chassis
Point is axle center, is rotated;
The chassis is equipped with moving assembly, and the position for mechanical arm is mobile;
The Dextrous Hand is rotated using the tie point with mobile arm as axle center, is piled up for landforms;
The quantity of the laser distance inductor is at least two, is separately mounted in Dextrous Hand and chassis, and moving process is used for
Middle barrier monitoring.
3. the apparatus of claim 2, which is characterized in that the controlling terminal includes:
Data acquisition module, for data needed for obtaining trained laser hologram projection prediction model and geomorphology information prediction model
Collection;
Prediction model building and training module, for constructing laser hologram projection sand table prediction model and geomorphology information prediction mould
Type, and two models are trained respectively using the data set that the data acquisition module obtains, obtain trained laser
Line holographic projections sand table prediction model and geomorphology information prediction model;
Projective parameter and complete relief data seek module, for by the length of sand table, wide, high, depth information to be input to above-mentioned instruction
The laser hologram projection prediction model perfected, obtains projective parameter;It is also used to for the profile of landforms being input to above-mentioned trained
Geomorphology information prediction model obtains complete relief data;
Projective parameter and complete relief data are sought the projective parameter that module is sought and are transmitted to laser hologram by signal transmission module
Projector;The complete relief data sought is transmitted to mechanical arm simultaneously.
4. device according to claim 3, which is characterized in that further include with " ten " word cursor, for position correction
Truing tool;The data acquisition module includes: area division unit, length and width dimensions collect unit, depth size collects unit;
The area division unit is made for dividing to sand table interior zone, and by the intersection point of all subregion marked off
On the basis of point;
The length and width dimensions collect unit, depth size collects unit, for binocular camera, laser ranging inductor, swash
Light holographic projector and truing tool cooperation, carry out data acquisition to each datum mark, obtain data set required for training pattern;
The data set includes plan-position coordinate, vertical position coordinate, focusing range and the clarity of datum mark.
5. device according to claim 4, which is characterized in that the data acquisition module further include: profile scan unit,
Key point picking unit, data supplementary units;
The profile scan unit, receives the landforms profile of spatial digitizer scanning, and searches to the turning occurred in profile
Collection;
The key point picking unit grabs length, the width, elevation information of landforms key point according to the position at above-mentioned turning;
The data supplementary units obtain each section decomposited for decomposing again to the part between adjacent key point
Length, width, altitude information, complete relief data is obtained, for training geomorphology information prediction model.
6. device according to claim 5, which is characterized in that described to carry out data acquisition to each datum mark, comprising:
Length and width dimensions collect unit and control laser hologram projector projects " ten " word cursor, and " ten " word cursor is successively incident upon respectively
The position of datum mark adjusts the position of truing tool, until the position weight of " ten " word of " ten " word and cursor in truing tool
Folded, length and width dimensions collect the plan-position coordinate of unit record datum mark at this time, and the length and width dimensions for completing sand table are collected;
Depth size collects unit and controls laser hologram projector projects " ten " word cursor, and " ten " word cursor is successively incident upon respectively
The position of datum mark, in the state of keeping the position overlapping of " ten " of " ten " word and cursor in truing tool, up/down is moved
Dynamic truing tool reaches optimal display effect, depth size until being somebody's turn to do " ten " word cursor in binocular camera display interface
Collect the unit record position vertical height, focusing range and sharpness information, as the datum mark vertical position coordinate,
Focusing range and clarity are completed sand table depth size and are collected.
7. device according to claim 6, which is characterized in that the prediction model building and training module, comprising:
Parameter initialization unit carries out initialization process to network weight and biasing, and selects at random in training data concentration
First input sample;
Parameter processing unit carries out neuronal activation forward-propagating, carries out weight to above-mentioned first input sample by hidden layer
And bias treatment, and seek the result and error of output layer;
Parameter adjustment unit carries out backpropagation according to above-mentioned error, is adjusted to network weight and biasing;
Whether training monitoring unit, terminate according to preset termination condition training of judgement.
8. device according to claim 7, which is characterized in that the projective parameter and complete relief data seek module,
Include:
High complexity zone marker unit, the region in complexity high in true landforms region and current sand table is compared, right
The point region for meeting high complexity in sand table is marked;
Weight determining unit determines position, the quantity, the weight of requirement degree in the point region of the above-mentioned high complexity marked;
Projective parameter seeks unit, and according to the weight of the length of sand table, depth information and above-mentioned determination, utilization is trained
Model seeks focusing range and clarity.
9. device according to claim 8, which is characterized in that the projective parameter and complete relief data seek module,
It after the length, depth information of acquisition sand table, scans for matching in the existing data set of building, when matching is consistent
When, then corresponding focusing range and sharpness information are directly chosen, when matching inconsistent, then uses trained laser hologram
Projection sand table prediction model seeks focusing range and sharpness information;After obtaining landforms profile, in the existing data of building
It scans for matching in library, when matching consistent, then directly chooses corresponding complete relief data;When matching inconsistent, then adopt
Complete relief data is sought with trained geomorphology information prediction model.
10. device described in one of -9 according to claim 1, which is characterized in that described device further includes landforms monitoring sensor,
For carrying out real-time monitoring to true geomorphology information and passing to controlling terminal, when controlling terminal judges that landforms change,
The interval time that landforms change is compared with preset benchmark, when the interval time of landforms variation being less than benchmark,
Then ignore;When the interval time of landforms variation being greater than benchmark, on the one hand, projective parameter and complete relief data seek module pair
Landforms whole region or high complexity region carry out parameter prediction again, obtain new focusing range and clarity parameter;Separately
On the one hand, projective parameter and complete relief data seek collection and the landforms that module carries out key point to new landforms region of variation
Supplement, obtains complete relief data.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910239479.7A CN110009985B (en) | 2019-03-27 | 2019-03-27 | Sand table making devices based on machine learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910239479.7A CN110009985B (en) | 2019-03-27 | 2019-03-27 | Sand table making devices based on machine learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110009985A true CN110009985A (en) | 2019-07-12 |
CN110009985B CN110009985B (en) | 2020-12-25 |
Family
ID=67168421
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910239479.7A Active CN110009985B (en) | 2019-03-27 | 2019-03-27 | Sand table making devices based on machine learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110009985B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111291729A (en) * | 2020-03-26 | 2020-06-16 | 北京百度网讯科技有限公司 | Human body posture estimation method, device, equipment and storage medium |
CN112785949A (en) * | 2021-01-22 | 2021-05-11 | 北京邮电大学 | Three-dimensional scene projection method and device and three-dimensional sand table |
CN114782901A (en) * | 2022-06-21 | 2022-07-22 | 深圳市禾讯数字创意有限公司 | Sand table projection method, device, equipment and medium based on visual change analysis |
Citations (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101699237A (en) * | 2009-11-20 | 2010-04-28 | 中国航空工业空气动力研究院 | Three-dimensional model attitude angle video measuring system for wind tunnel model test |
FR2970663A1 (en) * | 2011-01-26 | 2012-07-27 | Snecma | Processing a metal part by sintering and laser fusion for reducing its roughness, comprises automatically sanding a surface of the metal part with grains having different diameters, where the sanding process is defined by a program |
CN103389076A (en) * | 2013-07-24 | 2013-11-13 | 国家海洋局第二海洋研究所 | Submarine topography change detection and analysis method based on mesh reconstruction |
CN104134389A (en) * | 2014-08-19 | 2014-11-05 | 福建四创软件有限公司 | Interactive sand table demonstration system and method |
CN104259669A (en) * | 2014-09-11 | 2015-01-07 | 苏州菲镭泰克激光技术有限公司 | Precise three-dimensional curved surface laser marking method |
CN104567546A (en) * | 2014-12-23 | 2015-04-29 | 哈尔滨工业大学 | Infrared scene sand table system for testing infrared seeker |
CN105045389A (en) * | 2015-07-07 | 2015-11-11 | 深圳水晶石数字科技有限公司 | Demonstration method for interactive sand table system |
CN105894945A (en) * | 2016-04-13 | 2016-08-24 | 苏州育龙科教设备有限公司 | Digital sand table having landform teaching demonstration function |
CN106094721A (en) * | 2016-07-01 | 2016-11-09 | 西安工业大学 | The processing method of three-dimensional live topography sand table |
CN106980366A (en) * | 2017-02-27 | 2017-07-25 | 合肥安达创展科技股份有限公司 | Landform precisely catches system and fine high definition projection imaging system |
CN206594941U (en) * | 2016-05-06 | 2017-10-27 | 柳州铁道职业技术学院 | A kind of indoor digital mapping teaching platform device |
CN107545788A (en) * | 2017-10-17 | 2018-01-05 | 北京华如科技股份有限公司 | Goods electronic sand map system is deduced based on the operation that augmented reality is shown |
CN107886826A (en) * | 2017-11-22 | 2018-04-06 | 上海未石影视文化有限公司 | A kind of laser induced interactive digital sand table model |
CN108320311A (en) * | 2018-01-19 | 2018-07-24 | 杭州映墨科技有限公司 | A kind of AR sand table calibrating patterns computational methods for capableing of automatic deviation correction |
US20180210394A1 (en) * | 2017-01-26 | 2018-07-26 | The Charles Stark Draper Laboratory, Inc. | Method and Apparatus for Light Field Generation |
CN108510592A (en) * | 2017-02-27 | 2018-09-07 | 亮风台(上海)信息科技有限公司 | The augmented reality methods of exhibiting of actual physical model |
CN108789414A (en) * | 2018-07-17 | 2018-11-13 | 五邑大学 | Intelligent machine arm system based on three-dimensional machine vision and its control method |
CN208422342U (en) * | 2018-08-08 | 2019-01-22 | 深圳市大峡谷科技发展有限公司 | The projection of dual-projection three-dimensional imaging digital sand table controls equipment |
-
2019
- 2019-03-27 CN CN201910239479.7A patent/CN110009985B/en active Active
Patent Citations (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101699237A (en) * | 2009-11-20 | 2010-04-28 | 中国航空工业空气动力研究院 | Three-dimensional model attitude angle video measuring system for wind tunnel model test |
FR2970663A1 (en) * | 2011-01-26 | 2012-07-27 | Snecma | Processing a metal part by sintering and laser fusion for reducing its roughness, comprises automatically sanding a surface of the metal part with grains having different diameters, where the sanding process is defined by a program |
CN103389076A (en) * | 2013-07-24 | 2013-11-13 | 国家海洋局第二海洋研究所 | Submarine topography change detection and analysis method based on mesh reconstruction |
CN104134389A (en) * | 2014-08-19 | 2014-11-05 | 福建四创软件有限公司 | Interactive sand table demonstration system and method |
CN104259669A (en) * | 2014-09-11 | 2015-01-07 | 苏州菲镭泰克激光技术有限公司 | Precise three-dimensional curved surface laser marking method |
CN104567546A (en) * | 2014-12-23 | 2015-04-29 | 哈尔滨工业大学 | Infrared scene sand table system for testing infrared seeker |
CN105045389A (en) * | 2015-07-07 | 2015-11-11 | 深圳水晶石数字科技有限公司 | Demonstration method for interactive sand table system |
CN105894945A (en) * | 2016-04-13 | 2016-08-24 | 苏州育龙科教设备有限公司 | Digital sand table having landform teaching demonstration function |
CN206594941U (en) * | 2016-05-06 | 2017-10-27 | 柳州铁道职业技术学院 | A kind of indoor digital mapping teaching platform device |
CN106094721A (en) * | 2016-07-01 | 2016-11-09 | 西安工业大学 | The processing method of three-dimensional live topography sand table |
US20180210394A1 (en) * | 2017-01-26 | 2018-07-26 | The Charles Stark Draper Laboratory, Inc. | Method and Apparatus for Light Field Generation |
CN106980366A (en) * | 2017-02-27 | 2017-07-25 | 合肥安达创展科技股份有限公司 | Landform precisely catches system and fine high definition projection imaging system |
CN108510592A (en) * | 2017-02-27 | 2018-09-07 | 亮风台(上海)信息科技有限公司 | The augmented reality methods of exhibiting of actual physical model |
CN107545788A (en) * | 2017-10-17 | 2018-01-05 | 北京华如科技股份有限公司 | Goods electronic sand map system is deduced based on the operation that augmented reality is shown |
CN107886826A (en) * | 2017-11-22 | 2018-04-06 | 上海未石影视文化有限公司 | A kind of laser induced interactive digital sand table model |
CN108320311A (en) * | 2018-01-19 | 2018-07-24 | 杭州映墨科技有限公司 | A kind of AR sand table calibrating patterns computational methods for capableing of automatic deviation correction |
CN108789414A (en) * | 2018-07-17 | 2018-11-13 | 五邑大学 | Intelligent machine arm system based on three-dimensional machine vision and its control method |
CN208422342U (en) * | 2018-08-08 | 2019-01-22 | 深圳市大峡谷科技发展有限公司 | The projection of dual-projection three-dimensional imaging digital sand table controls equipment |
Non-Patent Citations (3)
Title |
---|
蔡凌豪: "基于增强实境的地形设计沙盘系统研究", 《西部人居环境学刊》 * |
郭岚,杜建丽: "电子沙盘的概念及其制作方法的比较与分析", 《测绘科学》 * |
陈靓,路海涛,谢庆胜,刘世靖,叶峰: "基于3D GIS和RS的安徽省地震监测台网电子沙盘制作的实现", 《西北地震学报》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111291729A (en) * | 2020-03-26 | 2020-06-16 | 北京百度网讯科技有限公司 | Human body posture estimation method, device, equipment and storage medium |
CN111291729B (en) * | 2020-03-26 | 2023-09-01 | 北京百度网讯科技有限公司 | Human body posture estimation method, device, equipment and storage medium |
CN112785949A (en) * | 2021-01-22 | 2021-05-11 | 北京邮电大学 | Three-dimensional scene projection method and device and three-dimensional sand table |
CN114782901A (en) * | 2022-06-21 | 2022-07-22 | 深圳市禾讯数字创意有限公司 | Sand table projection method, device, equipment and medium based on visual change analysis |
Also Published As
Publication number | Publication date |
---|---|
CN110009985B (en) | 2020-12-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109994036B (en) | Sand table manufacturing method based on machine learning | |
WO2022142759A1 (en) | Lidar and camera joint calibration method | |
CN1669069B (en) | System for texturizing electronic representations of objects | |
US20240153143A1 (en) | Multi view camera registration | |
CN106327573B (en) | A kind of outdoor scene three-dimensional modeling method for urban architecture | |
US7196730B2 (en) | Method and system for complete 3D object and area digitizing | |
CN104330074B (en) | Intelligent surveying and mapping platform and realizing method thereof | |
CN110009985A (en) | A kind of sand table producing device based on machine learning | |
CN109978955A (en) | A kind of efficient mask method for combining laser point cloud and image | |
CN106204656A (en) | Target based on video and three-dimensional spatial information location and tracking system and method | |
CN103954970B (en) | A kind of topographic(al) feature acquisition method | |
CN1149916A (en) | Method for collection, analysis, measurement and storage of geographical data | |
CN107105147A (en) | A kind of bionical super-resolution imaging sensor and imaging method | |
JPH03102213A (en) | Computer-aided survey system for obtaining digital three-dimensional geomorphological information | |
CN101919235A (en) | Orthophotographic image creating method and imaging device | |
CN104361603B (en) | Gun camera image target designating method and system | |
CN114332385A (en) | Monocular camera target detection and spatial positioning method based on three-dimensional virtual geographic scene | |
CN106846413A (en) | The device and method that three-dimensional tree-shaped rectangular coordinate system is built and image space is demarcated | |
CN107066975A (en) | Video identification and tracking system and its method based on depth transducer | |
CN109856946B (en) | Method for manufacturing laser holographic projection sand table | |
CN116883604A (en) | Three-dimensional modeling technical method based on space, air and ground images | |
CN109035400A (en) | A method of urban planning is established using oblique photograph | |
CN110415289A (en) | A kind of generation method and device of threedimensional model | |
CN111986320B (en) | Smart city application-oriented DEM and oblique photography model space fitting optimization method | |
US6922484B2 (en) | Information presentation apparatus and information presentation method |
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 |