CN110095751A - The target localization and tracking system of data-driven modeling is realized based on Method Using Relevance Vector Machine - Google Patents
The target localization and tracking system of data-driven modeling is realized based on Method Using Relevance Vector Machine Download PDFInfo
- Publication number
- CN110095751A CN110095751A CN201910222776.0A CN201910222776A CN110095751A CN 110095751 A CN110095751 A CN 110095751A CN 201910222776 A CN201910222776 A CN 201910222776A CN 110095751 A CN110095751 A CN 110095751A
- Authority
- CN
- China
- Prior art keywords
- data
- rss
- target
- vector machine
- relevance vector
- 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
- 238000000034 method Methods 0.000 title claims abstract description 144
- 230000004807 localization Effects 0.000 title claims abstract description 51
- 238000012549 training Methods 0.000 claims abstract description 79
- 238000005562 fading Methods 0.000 claims abstract description 35
- 230000003068 static effect Effects 0.000 claims abstract description 29
- 238000013480 data collection Methods 0.000 claims abstract description 18
- 238000012545 processing Methods 0.000 claims abstract description 18
- 239000011159 matrix material Substances 0.000 claims description 46
- 238000003384 imaging method Methods 0.000 claims description 25
- 238000003325 tomography Methods 0.000 claims description 21
- 238000005259 measurement Methods 0.000 claims description 16
- 238000004587 chromatography analysis Methods 0.000 claims description 11
- 238000013499 data model Methods 0.000 claims description 5
- 238000012935 Averaging Methods 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 3
- 241000208340 Araliaceae Species 0.000 claims 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims 1
- 235000003140 Panax quinquefolius Nutrition 0.000 claims 1
- 235000008434 ginseng Nutrition 0.000 claims 1
- 230000000875 corresponding effect Effects 0.000 abstract description 22
- 230000000694 effects Effects 0.000 abstract description 8
- 230000002596 correlated effect Effects 0.000 abstract description 6
- 239000004615 ingredient Substances 0.000 abstract description 6
- 238000010586 diagram Methods 0.000 description 12
- 230000007547 defect Effects 0.000 description 9
- 230000002708 enhancing effect Effects 0.000 description 9
- 230000006870 function Effects 0.000 description 9
- 238000004458 analytical method Methods 0.000 description 8
- 238000010276 construction Methods 0.000 description 7
- 230000035945 sensitivity Effects 0.000 description 6
- 230000008901 benefit Effects 0.000 description 4
- 230000015556 catabolic process Effects 0.000 description 4
- 238000006731 degradation reaction Methods 0.000 description 4
- 230000006698 induction Effects 0.000 description 4
- 238000012800 visualization Methods 0.000 description 4
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 230000007423 decrease Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 230000004888 barrier function Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S5/00—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations
- G01S5/02—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations using radio waves
- G01S5/0294—Trajectory determination or predictive filtering, e.g. target tracking or Kalman filtering
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The RSS data modeling method based on Method Using Relevance Vector Machine that this application discloses a kind of includes: to be used as output training RSS data after the RSS data of multiple preset scenes of acquisition is carried out RSS data pretreatment according to preset processing mode；By the RSS data collection of acquisition according to different links, carries out target corresponding with link and deviate sighting distance degree, target close to the three dimensional parameters solutions of node degree and static level of fading as input training data；The three-dimensional parameterized RSS model based on Method Using Relevance Vector Machine is established by exporting training RSS data and input training data.The influence of uncertainty caused by environment interference contribution is weakened by reducing unmodeled ingredient, remains to obtain effective guarantee so that target localization and tracking system correlated performance interferes in more complicated indoor scene in multipath effect.Disclosed herein as well is the target localization and tracking systems that data-driven modeling is realized based on Method Using Relevance Vector Machine.
Description
Technical field
This application involves radio frequency sensor technical field more particularly to a kind of RSS data modeling based on Method Using Relevance Vector Machine
Method and target locating method, the base that the target localization and tracking system of data-driven modeling is realized based on Method Using Relevance Vector Machine
The target localization and tracking system and computer readable storage medium of data-driven modeling are realized in Method Using Relevance Vector Machine.
Background technique
In the location-based intelligent Service that people, machine are interacted with environment depth, the acquisition of target position information is real
The important clue of existing basis Intellisense task.For the application scenarios of most of intelligent Services, target based on equipment
Positioning means applicability and in terms of it is all quite limited.
In recent years, it is based on the equipment hand-free of radio frequency receiving signal intensity (Received Signal Strength, RSS)
Positioning and tracking (Device-Free Localization and Tracking, DFLT) technology, utilize radio frequency sensing net
Link RSS observes pad value and is finally inversed by sensing region the distribution image of the shadow fading due to caused by target in network, and then to mesh
Mark carries out location estimation and tracking.Based on the RSS equipment hand-free positioning realized and tracking (RSS-DFLT) technology in wisdom man
It is concerned in the multiple fields such as front yard, intelligent medical treatment, intelligent monitoring.
However, the defect that current single dimensional model is not high to RSS data information utilization, so that RSS-
DFLT is that the multipath fading of barrier generation in hand-free target and environment interferes so that RSS observation data generation is not more true
It is qualitative, link RSS observation is weakened to the capture ability of hand-free target, causes the performance of RSS-DFLT system in multipath
There is the technical issues of degradation phenomena in the complex environment of serious interference.
Summary of the invention
The RSS data modeling method and realized based on Method Using Relevance Vector Machine that this application provides a kind of based on Method Using Relevance Vector Machine
The target locating method of the target localization and tracking system of data-driven modeling realizes data-driven based on Method Using Relevance Vector Machine
The target localization and tracking system and computer readable storage medium of modeling compensate for existing single dimensional model and believe RSS data
The not high defect of utilization rate is ceased, while enhancing expression ability of the model to data, is weakened by reducing unmodeled ingredient
Uncertainty caused by environment interference contribution influence so that target localization and tracking system correlated performance multipath effect interfere compared with
To remain to obtain effective guarantee in complicated indoor scene.
In view of this, the application first aspect provides a kind of RSS data modeling method based on Method Using Relevance Vector Machine, institute
The method of stating includes:
By the RSS data of multiple preset scenes of acquisition according to conduct after the progress RSS data pretreatment of preset processing mode
Export training RSS data；
By the RSS data collection of acquisition according to different links, carries out target corresponding with link and deviate sighting distance degree, mesh
The nearly node degree of tag splice and static three dimensional parameters of level of fading are solved as input training data；
The three-dimensional parameterized RSS mould based on Method Using Relevance Vector Machine is established by exporting training RSS data and input training data
Type.
Preferably, the RSS data of multiple preset scenes of acquisition RSS data are carried out according to preset processing mode to locate in advance
As training RSS data is exported after reason, specifically include:
The spacious outdoor scene of acquisition, general indoor scene, complex indoor scene human body target be located at and preset multiple positions
RSS data collection when setting；
It is poor make after averaging in certain time window to RSS data collection, and obtained RSS measurement changing value is as output
Training RSS data.
The three-dimensional parameterized RSS model parameterization for being preferably based on Method Using Relevance Vector Machine is expressed as
Sighting distance degree will be deviateed by target, target is combined close to node degree and static three dimensional parameters of level of fading
The output training RSS data of domination carries out parametrization expression:
Wherein, yiFor the corresponding trained RSS data of link i, M is number of links in sensor network, λi, γiAnd FiRespectively
The target of corresponding corresponding link deviates sighting distance degree, target close to node degree and static level of fading, eiFor model training mistake
Difference, K (θ, θi) it is the gaussian kernel function trained by three dimensional parameters, relation vector
The weight parameter ω=[ω estimated1,...,ωM]T。
It is real based on Method Using Relevance Vector Machine that the application second aspect provides a kind of RSS data model based on Method Using Relevance Vector Machine
The target locating method of the target localization and tracking system of existing data-driven modeling, which comprises
Target localization and tracking system based on Method Using Relevance Vector Machine realization data-driven modeling refers to the application any one
Three-dimensional parameterized RSS model described in kind carries out data-driven conjunctive model and calculates needed for generating radio frequency chromatography imaging task
Observing matrix；
The object to be tracked that the target localization and tracking system of data-driven modeling will acquire is realized based on Method Using Relevance Vector Machine
Real-time RSS data domain observation combine sparse restructing algorithm carry out radio frequency tomography in image reconstruction, determine mesh to be tracked
Mark the target positioning and tracking of object.
Preferably, observing matrix specifically includes:
Wherein, φijIndicate that j-th of pixel is to the contribution weight of the link in the i-th row in observing matrix.
Be preferably based on Method Using Relevance Vector Machine realize data-driven modeling target localization and tracking system will acquire to
The real-time RSS data domain observation of track object combines sparse restructing algorithm to carry out the image reconstruction in radio frequency tomography, determines
The target of object to be tracked positions and tracking, specifically includes:
The object to be tracked that the target localization and tracking system of data-driven modeling will acquire is realized based on Method Using Relevance Vector Machine
Real-time RSS data domain observation combine sparse restructing algorithm carry out radio frequency tomography in image reconstruction；
Realize the target localization and tracking system of data-driven modeling by the imaging knot after image reconstruction based on Method Using Relevance Vector Machine
The maximum single or multiple location of pixels of attenuation degree are as single goal or the estimated location of multiple target in fruit figure；
Realize that the target localization and tracking system of data-driven modeling passes through Kalman filtering and target based on Method Using Relevance Vector Machine
Real-time positioning result combines, and carries out the dynamic trajectory tracking of object to be tracked.
The application third aspect provides a kind of RSS data model building device based on Method Using Relevance Vector Machine, and described device includes:
Preprocessing module, the RSS data for the multiple preset scenes that will be acquired carry out RSS according to preset processing mode
As output training RSS data after data prediction；
Multidimensional processing module, the RSS data collection for that will acquire carry out mesh corresponding with link according to different links
Mark deviates sighting distance degree, target solves to be used as and inputs training number close to node degree and static three dimensional parameters of level of fading
According to；
Modeling module, for by exporting train RSS data and input training data to establish based on Method Using Relevance Vector Machine three
Dimension parametrization RSS model.
The application fourth aspect provides a kind of target locating that data-driven modeling is realized based on Method Using Relevance Vector Machine
Target localization and tracking system, comprising:
Computing module, for carrying out data drive to any one of three-dimensional parameterized RSS model that the application refers to
Dynamic conjunctive model, which calculates, generates observing matrix needed for radio frequency chromatographs imaging task；
The real-time RSS data domain observation of tracking module, the object to be tracked for will acquire combines sparse restructing algorithm
The image reconstruction in radio frequency tomography is carried out, determines the target positioning and tracking of object to be tracked.
The 5th aspect of the application provides a kind of computer readable storage medium, and the computer readable storage medium is used for
Program code is stored, said program code is for executing method described in above-mentioned first aspect.
The 6th aspect of the application provides a kind of computer readable storage medium, and the computer readable storage medium is used for
Program code is stored, said program code is for executing method described in above-mentioned second aspect.
As can be seen from the above technical solutions, the embodiment of the present application has the advantage that
In the embodiment of the present application, a kind of RSS data modeling method based on Method Using Relevance Vector Machine is provided and its based on correlation
Vector machine realizes the target locating method of the target localization and tracking system of data-driven modeling, based on Method Using Relevance Vector Machine reality
The target localization and tracking system and computer readable storage medium of existing data-driven modeling, the application based on Method Using Relevance Vector Machine
RSS data modeling method include: by the RSS data of multiple preset scenes of acquisition according to preset processing mode carry out RSS number
As training RSS data is exported after Data preprocess, by the RSS data collection of acquisition according to different links, progress is answered with link pair
Target deviate sighting distance degree, target is solved close to three dimensional parameters of node degree and static level of fading as input instruction
Practice data, establishes the three-dimensional parameterized RSS mould based on Method Using Relevance Vector Machine by exporting training RSS data and input training data
Type is different from traditional radio propagation mechanism analysis method and empirical analysis method based on extensive utilization in RSS Modeling Research,
It the advantages of cross-coupled relationship between each dimensional parameter and comprehensive three dimensional parameters can effectively be extracted, compensates for existing
The single dimensional model defect not high to RSS data information utilization, while enhancing expression ability of the model to data,
The influence of uncertainty caused by environment interference contribution is weakened by reducing unmodeled ingredient, solves current single dimension
The model defect not high to RSS data information utilization rate, the caused performance for causing RSS-DFLT system are dry in multipath
It disturbs in serious complex environment and the technical issues of degradation phenomena occurs.
And the target of the target localization and tracking system of the invention that data-driven modeling is realized based on Method Using Relevance Vector Machine
Positioning and tracing method carries out data-driven conjunctive model by using the three-dimensional parameterized RSS model that the present invention refers to and calculates life
Observing matrix needed for chromatographing imaging task at radio frequency, the real-time RSS data domain for the object to be tracked that will acquire, which is observed, to be combined
Sparse restructing algorithm carries out the image reconstruction in radio frequency tomography, determines the target positioning and tracking of object to be tracked, by
In the influence and link for having fully considered target position measurement data being changed embodied under by environment multi-path jamming it is different
Sensitivity characteristic, so that more complicated in multipath effect interference using the target localization and tracking system correlated performance of the model
It remains to obtain effective guarantee in indoor scene.
Detailed description of the invention
Fig. 1 is the process of one embodiment of the RSS data modeling method based on Method Using Relevance Vector Machine in the embodiment of the present application
Figure；
Fig. 2 is the stream of another embodiment of the RSS data modeling method based on Method Using Relevance Vector Machine in the embodiment of the present application
Cheng Tu；
Fig. 3 is that the RSS data model based on Method Using Relevance Vector Machine is based on Method Using Relevance Vector Machine realization data in the embodiment of the present application
Drive the flow chart of target locating method one embodiment of the target localization and tracking system of modeling；
Fig. 4 is the target localization and tracking system for realizing data-driven modeling in the embodiment of the present application based on Method Using Relevance Vector Machine
Target locating method another embodiment flow chart；
Fig. 5 is a structural schematic diagram of the RSS data model building device based on Method Using Relevance Vector Machine in the embodiment of the present invention；
Fig. 6 is the target localization and tracking system for realizing data-driven modeling in the embodiment of the present invention based on Method Using Relevance Vector Machine
A structural schematic diagram；
Fig. 7 is the application scenarios schematic diagram that the present invention is embodied；
Fig. 8 (a) to Fig. 8 (c) is the experiment scene schematic diagram that the present invention implements data acquisition and locating and tracking task；
Fig. 9 is that radio frequency chromatographs imaging observation matrix construction schematic diagram；
Figure 10 (a) to Figure 10 (c) is results of property figure of the model of the present invention in object locating system.
Specific embodiment
In order to make those skilled in the art more fully understand application scheme, below in conjunction in the embodiment of the present application
Attached drawing, the technical scheme in the embodiment of the application is clearly and completely described, it is clear that described embodiment is only
Some embodiments of the present application, instead of all the embodiments.Based on the embodiment in the application, ordinary skill people
Member's every other embodiment obtained without making creative work, shall fall in the protection scope of this application.
The RSS data modeling method that this application provides a kind of based on Method Using Relevance Vector Machine and its real based on Method Using Relevance Vector Machine
The target locating method of the target localization and tracking system of existing data-driven modeling realizes that data are driven based on Method Using Relevance Vector Machine
The target localization and tracking system and computer readable storage medium of dynamic modeling, compensate for existing single dimensional model to RSS data
The not high defect of information utilization is weakened while enhancing expression ability of the model to data by reducing unmodeled ingredient
Uncertainty caused by environment interference contribution influences, so that target localization and tracking system correlated performance is interfered in multipath effect
It remains to obtain effective guarantee in more complicated indoor scene.
In order to make it easy to understand, referring to Fig. 1, Fig. 1 is a kind of RSS number based on Method Using Relevance Vector Machine in the embodiment of the present application
According to the method flow diagram of modeling method, as shown in Figure 1, specifically:
101, after the RSS data of multiple preset scenes of acquisition being carried out RSS data pretreatment according to preset processing mode
RSS data is trained as exporting；
In the present embodiment, in order to effectively reduce uncertain shadow of the multipath fading interference to RSS observation of environmental induction
It rings, so that the performance of RSS-DFLT system is improved, the angle that the present invention is modeled from RSS data first, in conjunction with multiple dimensional parameters
Unmodeled composition information in RSS observation is effectively extracted, the joint RSS model of data-driven is proposed.
Firstly, the RSS data of multiple preset scenes of acquisition is carried out RSS data pretreatment according to preset processing mode
Afterwards as output training RSS data.It should be noted that this step specifically acquires and pretreatment mode will be in subsequent implementation
Example is described in detail.
102, the RSS data collection of acquisition is carried out target corresponding with link and deviates sighting distance journey according to different links
Degree, target are solved close to node degree and static three dimensional parameters of level of fading as input training data；
While step 101, by the RSS data collection of acquisition according to different links, target corresponding with link is carried out
Deviate sighting distance degree, target is solved close to node degree and static three dimensional parameters of level of fading as input training data.
In this specific embodiment, using the data set acquired, according to different links, to related to link
Target deviate sighting distance degree, the value of target close to node degree and static three dimensional parameters of level of fading is solved, make
To be originally inputted training data.
103, it is established by exporting training RSS data and input training data based on the three-dimensional parameterized of Method Using Relevance Vector Machine
RSS model.
In the present embodiment, the three-dimensional based on Method Using Relevance Vector Machine is established by exporting training RSS data and input training data
RSS model is parameterized, which is to utilize related Method Using Relevance Vector Machine method regression estimates model.
The RSS data modeling method based on Method Using Relevance Vector Machine of the application includes: by multiple preset scenes of acquisition
RSS data is used as output training RSS data after carrying out RSS data pretreatment according to preset processing mode, by the RSS number of acquisition
According to collection according to different links, target deviation sighting distance degree corresponding with link is carried out, target declines close to node degree and static state
Overboard flat three dimensional parameters are solved as input training data, are established by exporting training RSS data and input training data
Three-dimensional parameterized RSS model based on Method Using Relevance Vector Machine is different from traditional nothing based on extensive utilization in RSS Modeling Research
Line electric transmission mechanism analysis method and empirical analysis method can effectively extract the cross-coupled relationship between each dimensional parameter and comprehensive
The advantages of closing three dimensional parameters compensates for the existing single dimensional model defect not high to RSS data information utilization,
While enhancing expression ability of the model to data, weakened caused by environment interference contribution not by reducing unmodeled ingredient
Certainty influences, and solves the current single dimensional model defect not high to RSS data information utilization, caused makes
There is the technical issues of degradation phenomena in the serious complex environment of multi-path jamming at the performance of RSS-DFLT system.
In order to make it easy to understand, referring to Fig. 2, Fig. 2 is a kind of RSS number based on Method Using Relevance Vector Machine in the embodiment of the present application
According to the method flow diagram of modeling method, as shown in Fig. 2, specifically:
201, spacious outdoor scene, general indoor scene, the human body target of complex indoor scene acquired is more positioned at presetting
RSS data collection when a position；
In the present embodiment, in order to effectively reduce uncertain shadow of the multipath fading interference to RSS observation of environmental induction
It rings, so that the performance of RSS-DFLT system is improved, the angle that the present invention is modeled from RSS data first, in conjunction with multiple dimensional parameters
Unmodeled composition information in RSS observation is effectively extracted, the joint RSS model of data-driven is proposed.
Firstly, standing on the radio frequency receiving signal intensity of multiple predeterminated positions in radio frequency sensor network to human body target
(RSS) data carry out more scenes acquisition (such as Fig. 8 (a): empty of spacious outdoor scene, general indoor scene, complex indoor scene
Spacious outdoor scene, (b): general indoor scene, (c): more scenes acquisition of complex indoor scene).
202, make after averaging in certain time window RSS measurement changing value conduct that is poor, obtaining to RSS data collection
Export training RSS data.
Specifically, using in indifferent to human body target monitoring area RSS data collected as experiment base value
According to the RSS data being located at when presetting multiple positions to human body target respectively is acquired, to two parts data in certain time
It is poor to make after averaging in window, obtains RSS measurement changing value as output training RSS data.
203, the RSS data collection of acquisition is carried out target corresponding with link and deviates sighting distance journey according to different links
Degree, target are solved close to node degree and static three dimensional parameters of level of fading as input training data；
While step 201 and 202, by the RSS data collection of acquisition according to different links, progress is answered with link pair
Target deviate sighting distance degree, target is solved close to three dimensional parameters of node degree and static level of fading as input instruction
Practice data.
In this specific embodiment, using the data set acquired, according to different links, to related to link
Target deviate sighting distance degree, the value of target close to node degree and static three dimensional parameters of level of fading is solved, make
To be originally inputted training data.
204, it is established by exporting training RSS data and input training data based on the three-dimensional parameterized of Method Using Relevance Vector Machine
RSS model.
In the present embodiment, the three-dimensional based on Method Using Relevance Vector Machine is established by exporting training RSS data and input training data
RSS model is parameterized, which is to utilize related Method Using Relevance Vector Machine method regression estimates model.
Consider to be joined by target deviation sighting distance degree, target close to node degree and static three dimensional parameters of level of fading
It closes and dominates three-dimensional parameterized RSS model, the training data that is originally inputted about three dimensional parameters is trained into RSS number with output
According to Relation Parameters be expressed as: export the gaussian kernel function relation vector that is trained by three dimensional parameters of training RSS data
It is obtained with after Unknown weights weight parameter product plus model training error.
Sighting distance degree will be deviateed by target, target is combined close to node degree and static three dimensional parameters of level of fading
The output training RSS data of domination carries out parametrization expression:
Wherein, yiFor the corresponding trained RSS data of link i, M is number of links in sensor network, λi, γiAnd FiRespectively
The target of corresponding corresponding link deviates sighting distance degree, target close to node degree and static level of fading, eiFor model training mistake
Difference, K (θ, θi) it is the gaussian kernel function trained by three dimensional parameters, relation vector
The weight parameter ω=[ω estimated1,...,ωM]T。
The estimation problem of analysis model weight parameter belongs to regression estimation problem, using be usually used in solve compressed signal weight
Structure problem, the sparse restructing algorithm for being equally applicable to solve regression estimation problem estimate Model Weight parameter.
The RSS data modeling method based on Method Using Relevance Vector Machine of the application includes: by multiple preset scenes of acquisition
RSS data is used as output training RSS data after carrying out RSS data pretreatment according to preset processing mode, by the RSS number of acquisition
According to collection according to different links, target deviation sighting distance degree corresponding with link is carried out, target declines close to node degree and static state
Overboard flat three dimensional parameters are solved as input training data, are established by exporting training RSS data and input training data
Three-dimensional parameterized RSS model based on Method Using Relevance Vector Machine is different from traditional nothing based on extensive utilization in RSS Modeling Research
Line electric transmission mechanism analysis method and empirical analysis method can effectively extract the cross-coupled relationship between each dimensional parameter and comprehensive
The advantages of closing three dimensional parameters compensates for the existing single dimensional model defect not high to RSS data information utilization,
While enhancing expression ability of the model to data, weakened caused by environment interference contribution not by reducing unmodeled ingredient
Certainty influences, and solves the current single dimensional model defect not high to RSS data information utilization, caused makes
There is the technical issues of degradation phenomena in the serious complex environment of multi-path jamming at the performance of RSS-DFLT system.
In order to make it easy to understand, referring to Fig. 3, Fig. 3 is a kind of based on Method Using Relevance Vector Machine realization data in the embodiment of the present application
The flow chart for driving the target locating method of the target localization and tracking system of modeling, as shown in Fig. 3, specifically:
301, realize the target localization and tracking system of data-driven modeling to Fig. 1 and Fig. 2 embodiment based on Method Using Relevance Vector Machine
The three-dimensional parameterized RSS model established carries out sight needed for data-driven conjunctive model calculates generation radio frequency chromatography imaging task
Survey matrix；
In the present embodiment, in the target for the target localization and tracking system for realizing data-driven modeling based on Method Using Relevance Vector Machine
In the realization of locating and tracking, two stages, off-line phase and application on site stage can be, off-line phase is based on Fig. 1 and figure
The three-dimensional parameterized RSS model that 2 embodiments are established is directed to the building plan in existing radio frequency chromatographic imaging system to observing matrix
Slightly Shortcomings, the present embodiment is the observing matrix building method based on data-driven joint RSS model, for different sensitivities
The link and different location of pixels for spending characteristic are that each pixel distributes decaying or enhancing contribution, enable observing matrix sufficiently sharp
With shadow fading information and weaken multi-path jamming.
302, the mesh to be tracked that the target localization and tracking system based on Method Using Relevance Vector Machine realization data-driven modeling will acquire
Mark object real-time RSS data domain observation combine sparse restructing algorithm carry out radio frequency tomography in image reconstruction, determine to
The target of track object positions and tracking.
In the present embodiment, on-line stage is that the real-time RSS data domain observation combination for the object to be tracked that will acquire is sparse
Restructing algorithm carries out the image reconstruction in radio frequency tomography, determines the target positioning and tracking of object to be tracked.
The target positioning of the of the invention target localization and tracking system that data-driven modeling is realized based on Method Using Relevance Vector Machine with
Track method carries out data-driven conjunctive model by using the three-dimensional parameterized RSS model that the present invention refers to and calculates generation radio frequency
The observing matrix of tomography required by task, the real-time RSS data domain observation for the object to be tracked that will acquire combine sparse heavy
Structure algorithm carries out the image reconstruction in radio frequency tomography, the target positioning and tracking of object to be tracked is determined, due to abundant
It considers influence that target position changes measurement data and link is embodying different susceptibilitys under by environment multi-path jamming
Characteristic, so that using the target localization and tracking system correlated performance of the model in the more complicated indoor field of multipath effect interference
It remains to obtain effective guarantee in scape.
In order to make it easy to understand, referring to Fig. 4, Fig. 4 is a kind of based on Method Using Relevance Vector Machine realization data in the embodiment of the present application
The flow chart for driving the target locating method of the target localization and tracking system of modeling, as shown in Fig. 4, specifically:
401, realize the target localization and tracking system of data-driven modeling to Fig. 1 and Fig. 2 embodiment based on Method Using Relevance Vector Machine
The three-dimensional parameterized RSS model established carries out sight needed for data-driven conjunctive model calculates generation radio frequency chromatography imaging task
Survey matrix；
In the present embodiment, in the target for the target localization and tracking system for realizing data-driven modeling based on Method Using Relevance Vector Machine
In the realization of locating and tracking, two stages, off-line phase and application on site stage can be, off-line phase is based on Fig. 1 and figure
The three-dimensional parameterized RSS model that 2 embodiments are established is directed to the building plan in existing radio frequency chromatographic imaging system to observing matrix
Slightly Shortcomings, the present embodiment is the observing matrix building method based on data-driven joint RSS model, for different sensitivities
The link and different location of pixels for spending characteristic are that each pixel distributes decaying or enhancing contribution, enable observing matrix sufficiently sharp
With shadow fading information and weaken multi-path jamming.
It obtains being originally inputted training data by data acquisition, pretreatment and after calculating and exports training RSS data, with
Parameterized form indicates three-dimensional parameterized RSS model, and the regression estimates of weight parameter are realized based on Method Using Relevance Vector Machine method.It examines
Consider in existing radio frequency chromatographic imaging system and there is not rigorous place to the construction strategy of observing matrix, proposes to be based on data-driven
The observing matrix building method of joint RSS model, link and different location of pixels for varying sensitivity characteristic are each picture
Element distribution decaying or enhancing contribution, enable observing matrix to make full use of shadow fading information and weaken multi-path jamming.Observe square
Battle array can be calculated in advance according to RSS model in off-line phase:
Wherein, φijIndicate contribution weight of j-th of pixel to the link, radio frequency chromatography in the i-th row in observing matrix
Imaging observation matrix construction schematic diagram is as shown in Fig. 9.It is calculated to more intuitively observe and analyze data-driven conjunctive model
The observing matrix of generation takes an observing matrix wherein row vector, such as the measurement vector Φ of link lR(l :) it is converted into two-dimentional shape
Formula is visualized, and link measures shown in vector visualization result such as attached drawing 10 (a).
402, the mesh to be tracked that the target localization and tracking system based on Method Using Relevance Vector Machine realization data-driven modeling will acquire
The real-time RSS data domain observation for marking object combines sparse restructing algorithm to carry out the image reconstruction in radio frequency tomography.
In the present embodiment, on-line stage is the target locating system that data-driven modeling is realized based on Method Using Relevance Vector Machine
The real-time RSS data domain observation for the object to be tracked that system will acquire combines sparse restructing algorithm to carry out in radio frequency tomography
Image reconstruction.
According to the measurement equation of radio frequency chromatographic imaging system in specific one embodiment:
yR=ΦRxR+eR
Wherein, eRTo measure noise vector.The observing matrix Φ obtained using off-line phaseRWith real time data yR, in conjunction with
Sparse restructing algorithm completes the image reconstruction step in radio frequency tomography, i.e., in sensing region by targets of interest induction
Decline image xRInverting is carried out, to realize the positioning and tracking of target.
It should be noted that the sparse restructing algorithm that uses of above-mentioned RSS data modeling be applied to solve radio frequency chromatography at
It is isomery Bayes compressed sensing (Heterogeneous Bayesian as the sparse restructing algorithm of image reconstruction is consistent
Compressive Sensing, HBCS) algorithm.
403, realized based on Method Using Relevance Vector Machine the target localization and tracking system of data-driven modeling by after image reconstruction at
As in result figure the maximum single or multiple location of pixels of attenuation degree as single goal or the estimated location of multiple target；
404, based on Method Using Relevance Vector Machine realize data-driven modeling target localization and tracking system by Kalman filtering with
The real-time positioning result of target combines, and carries out the dynamic trajectory tracking of object to be tracked.
The target positioning of the of the invention target localization and tracking system that data-driven modeling is realized based on Method Using Relevance Vector Machine with
Track method, by realizing what the target localization and tracking system of data-driven modeling was referred to using the present invention based on Method Using Relevance Vector Machine
Three-dimensional parameterized RSS model carries out observing matrix needed for data-driven conjunctive model calculates generation radio frequency chromatography imaging task,
The real-time RSS data domain observation for the object to be tracked that will acquire combines sparse restructing algorithm to carry out in radio frequency tomography
Image reconstruction determines the target positioning and tracking of object to be tracked, due to having fully considered that target position becomes measurement data
The influence of change and link are embodying different sensitivity characteristics under by environment multi-path jamming, so that using the target of the model
Locating and tracking system correlated performance interferes in more complicated indoor scene in multipath effect and remains to obtain effective guarantee.
The off-line phase of Fig. 3, Fig. 3 embodiment will be carried out with an application scenarios and in conjunction with Fig. 7 to facilitate the understanding of the present invention
Illustrate with on-line stage, referring to Fig. 7, including:
1) off-line training step:
Step 1: training data being acquired and is pre-processed, data is participated in and acquires the schematic diagram of a scenario such as attached drawing 8 for including
It is shown.
Step 2: three-dimensional parameterized RSS model construction is realized based on Method Using Relevance Vector Machine method:
Specifically, sighting distance degree will be deviateed by target, target is joined close to node degree and static three dimensions of level of fading
The output training RSS data parameter that number joint dominates is expressed as
Wherein, yiFor the corresponding trained RSS data of link i, M is number of links in sensor network, λi, γiAnd FiIt is right respectively
The target of corresponding link is answered to deviate sighting distance degree, target close to node degree and static level of fading, eiFor model training mistake
Difference, K (θ, θi) it is the gaussian kernel function trained by three dimensional parameters, relation vector
Our target is to realize weight parameter ω=[ω by Method Using Relevance Vector Machine method1,...,ωM]TEstimation.
Step 3: observing matrix needed for generating radio frequency chromatography imaging task is calculated using data-driven conjunctive model:
Specifically, observing matrix can be calculated in advance according to RSS model in off-line phase:
Wherein, φijIndicate contribution weight of j-th of pixel to the link, radio frequency chromatography in the i-th row in observing matrix
Imaging observation matrix construction schematic diagram is as shown in Fig. 9.It is calculated to more intuitively observe and analyze data-driven conjunctive model
The observing matrix of generation takes an observing matrix wherein row vector, such as the measurement vector Φ of link lR(l :) it is converted into two-dimentional shape
Formula is visualized, and link measurement vector visualization result such as attached drawing 10 (a) show link measurement vector visualization result.
2) the application on site stage:
Step 1: obtaining target body and be located at the real-time RSS data in sensor network, participate in the reality of locating and tracking task
It is as shown in Fig. 8 to test schematic diagram of a scenario.
Step 2: the observing matrix and real-time RSS data obtained using off-line phase is penetrated in conjunction with the completion of sparse restructing algorithm
Image reconstruction step in frequency tomography:
Specifically, according to the measurement equation of radio frequency chromatographic imaging system:
yR=ΦRxR+eR
Wherein, eRTo measure noise vector.The observing matrix Φ obtained using off-line phaseRWith real time data yR, in conjunction with
Isomery Bayes compressed sensing (HBCS) algorithm in sparse restructing algorithm completes the image reconstruction step in radio frequency tomography
Suddenly, i.e., to the decline image x induced in sensing region by targets of interestRCarry out inverting, thus realize target positioning and with
Track.
Step 3: using the maximum single or multiple location of pixels of attenuation degree in imaging results figure as single goal or more mesh
Target estimated location.The multi-path jamming under complex scene can be overcome to verify data-driven model established by the present invention, guaranteed
Target positioning performance, attached drawing 10 (b) are that Bi-objective positions as a result, giving the indoor office ring in multipath effect serious interference
The positioning result of Bi-objective in border (attached drawing 8 (c)).
Step 4: being combined using kalman filter method with the real-time positioning result of target, realize the dynamic trajectory of target
Tracking, attached drawing 10 (c) are monotrack as a result, giving the indoor office environments (attached drawing 8 in multipath effect serious interference
(c)) the track following result of the single goal in.
In order to make it easy to understand, referring to Fig. 5, Fig. 5 is a kind of RSS number based on Method Using Relevance Vector Machine in the embodiment of the present application
According to model building device structure chart, as shown in figure 5, specifically:
Preprocessing module 501, the RSS data for the multiple preset scenes that will be acquired are carried out according to preset processing mode
As output training RSS data after RSS data pretreatment；
In one embodiment, the radio frequency of multiple predeterminated positions in radio frequency sensor network is stood on to human body target
Received signal strength (RSS) data carry out more scenes acquisition of spacious outdoor scene, general indoor scene, complex indoor scene
(such as Fig. 8 (a): spacious outdoor scene, (b): general indoor scene, (c): more scenes acquisition of complex indoor scene).
Using in indifferent to human body target monitoring area RSS data collected as experiment reference data, it is right respectively
The RSS data that human body target is located at when presetting multiple positions is acquired, and is averaged in certain time window to two parts data
It is poor to make afterwards, obtains RSS measurement changing value as output training RSS data.
Multidimensional processing module 502, the RSS data collection for that will acquire carry out corresponding with link according to different links
Target deviates sighting distance degree, target is solved close to node degree and static three dimensional parameters of level of fading as input training
Data；
Meanwhile the RSS data collection of acquisition being carried out target corresponding with link and deviateing sighting distance journey according to different links
Degree, target are solved close to node degree and static three dimensional parameters of level of fading as input training data.
In this specific embodiment, using the data set acquired, according to different links, to related to link
Target deviate sighting distance degree, the value of target close to node degree and static three dimensional parameters of level of fading is solved, make
To be originally inputted training data.
Modeling module 503, for establishing by exporting training RSS data and input training data and being based on Method Using Relevance Vector Machine
Three-dimensional parameterized RSS model.
Consider to be joined by target deviation sighting distance degree, target close to node degree and static three dimensional parameters of level of fading
It closes and dominates three-dimensional parameterized RSS model, the training data that is originally inputted about three dimensional parameters is trained into RSS number with output
According to Relation Parameters be expressed as: export the gaussian kernel function relation vector that is trained by three dimensional parameters of training RSS data
It is obtained with after Unknown weights weight parameter product plus model training error.
Sighting distance degree will be deviateed by target, target is combined close to node degree and static three dimensional parameters of level of fading
The output training RSS data of domination carries out parametrization expression:
Wherein, yiFor the corresponding trained RSS data of link i, M is number of links in sensor network, λi, γiAnd FiRespectively
The target of corresponding corresponding link deviates sighting distance degree, target close to node degree and static level of fading, eiFor model training mistake
Difference, K (θ, θi) it is the gaussian kernel function trained by three dimensional parameters, relation vector
The weight parameter ω=[ω estimated1,...,ωM]T。
The estimation problem of analysis model weight parameter belongs to regression estimation problem, using be usually used in solve compressed signal weight
Structure problem, the sparse restructing algorithm for being equally applicable to solve regression estimation problem estimate Model Weight parameter.
In order to make it easy to understand, referring to Fig. 6, Fig. 6 is a kind of based on Method Using Relevance Vector Machine realization data in the embodiment of the present application
The target localization and tracking system structure chart of modeling is driven, as shown in fig. 6, specifically:
Computing module 601 combines mould for carrying out data-driven to three-dimensional parameterized RSS model according to any one of claims 8
Type, which calculates, generates observing matrix needed for radio frequency chromatographs imaging task；
It obtains being originally inputted training data by data acquisition, pretreatment and after calculating and exports training RSS data, with
Parameterized form indicates three-dimensional parameterized RSS model, realizes that the recurrence of weight parameter is estimated based on related Method Using Relevance Vector Machine method
Meter.Consider there is not rigorous place to the construction strategy of observing matrix in existing radio frequency chromatographic imaging system, proposes to be based on data
The observing matrix building method of joint RSS model is driven, link and different location of pixels for varying sensitivity characteristic are
Each pixel distribution decaying or enhancing contribution, enable observing matrix to make full use of shadow fading information and weaken multi-path jamming.It sees
Surveying matrix can be calculated in advance according to RSS model in off-line phase:
Wherein, φijIndicate contribution weight of j-th of pixel to the link, radio frequency chromatography in the i-th row in observing matrix
Imaging observation matrix construction schematic diagram is as shown in Fig. 9.It is calculated to more intuitively observe and analyze data-driven conjunctive model
The observing matrix of generation takes an observing matrix wherein row vector, such as the measurement vector Φ of link lR(l :) it is converted into two-dimentional shape
Formula is visualized, and link measures shown in vector visualization result such as attached drawing 10 (a).
The real-time RSS data domain observation of tracking module 602, the object to be tracked for will acquire combines sparse reconstruct
Algorithm carries out the image reconstruction in radio frequency tomography, determines the target positioning and tracking of object to be tracked.
According to the measurement equation of radio frequency chromatographic imaging system in specific one embodiment:
yR=ΦRxR+eR
Wherein, eRTo measure noise vector.The observing matrix Φ obtained using off-line phaseRWith real time data yR, in conjunction with
Sparse restructing algorithm completes the image reconstruction step in radio frequency tomography, i.e., in sensing region by targets of interest induction
Decline image xRInverting is carried out, to realize the positioning and tracking of target.
It should be noted that the sparse restructing algorithm that uses of above-mentioned RSS data modeling be applied to solve radio frequency chromatography at
It is isomery Bayes compressed sensing (Heterogeneous Bayesian as the sparse restructing algorithm of image reconstruction is consistent
Compressive Sensing, HBCS) algorithm.
Using the maximum single or multiple location of pixels of attenuation degree in the imaging results figure after image reconstruction as single goal
Or the estimated location of multiple target；
Combined by Kalman filtering with the real-time positioning result of target, carry out the dynamic trajectory of object to be tracked with
Track.
The embodiment of the present application also provides a kind of computer readable storage medium, for storing program code, the program code
For executing any one in a kind of RSS data modeling method based on Method Using Relevance Vector Machine described in foregoing individual embodiments
Embodiment.
The embodiment of the present application also provides a kind of computer readable storage medium, for storing program code, the program code
For executing a kind of target locating for realizing data-driven modeling based on Method Using Relevance Vector Machine described in foregoing individual embodiments
Any one embodiment in system.
In several embodiments provided herein, it should be understood that disclosed system, device and method can be with
It realizes by another way.For example, the apparatus embodiments described above are merely exemplary, for example, the module
It divides, only a kind of logical function partition, there may be another division manner in actual implementation, such as multiple modules or group
Part can be combined or can be integrated into another system, or some features can be ignored or not executed.Another point, it is shown
Or the mutual coupling, direct-coupling or communication connection discussed can be through some interfaces, between device or module
Coupling or communication connection are connect, can be electrical property, mechanical or other forms.
The module as illustrated by the separation member may or may not be physically separated, as module
The component of display may or may not be physical module, it can and it is in one place, or may be distributed over more
On a network module.Some or all of the modules therein can be selected to realize this embodiment scheme according to the actual needs
Purpose.
It, can also be in addition, can integrate in a processing module in each functional module in each embodiment of the application
It is that modules physically exist alone, can also be integrated in two or more modules in a module.Above-mentioned integrated mould
Block both can take the form of hardware realization, can also be realized in the form of software function module.
If the integrated module is realized in the form of software function module and sells or use as independent product
When, it can store in a computer readable storage medium.Based on this understanding, the technical solution essence of the application
On all or part of the part that contributes to existing technology or the technical solution can be with the shape of software product in other words
Formula embodies, which is stored in a storage medium, including some instructions are used so that a calculating
Machine equipment (can be personal computer, server or the network equipment etc.) executes each embodiment the method for the application
All or part of the steps.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (full name in English: Read-
Only Memory, english abbreviation: ROM), random access memory (full name in English: RandomAccess Memory, English contracting
Write: RAM), the various media that can store program code such as magnetic or disk.
The above, above embodiments are only to illustrate the technical solution of the application, rather than its limitations；Although referring to before
Embodiment is stated the application is described in detail, those skilled in the art should understand that: it still can be to preceding
Technical solution documented by each embodiment is stated to modify or equivalent replacement of some of the technical features；And these
It modifies or replaces, the spirit and scope of each embodiment technical solution of the application that it does not separate the essence of the corresponding technical solution.
Claims (10)
1. a kind of RSS data modeling method based on Method Using Relevance Vector Machine characterized by comprising
Output is used as after the RSS data of multiple preset scenes of acquisition is carried out RSS data pretreatment according to preset processing mode
Training RSS data；
By the RSS data collection of acquisition according to different links, it is close to carry out target deviation sighting distance degree corresponding with link, target
Node degree and static three dimensional parameters of level of fading are solved as input training data；
The three-dimensional parameterized RSS model based on Method Using Relevance Vector Machine is established by exporting training RSS data and input training data.
2. the method according to claim 1, wherein by the RSS data of multiple preset scenes of acquisition according to pre-
It sets after processing mode carries out RSS data pretreatment and is used as output training RSS data, specifically include:
The spacious outdoor scene of acquisition, general indoor scene, complex indoor scene human body target be located at when presetting multiple positions
RSS data collection；
It is poor make after averaging in certain time window to RSS data collection, and obtained RSS measurement changing value is as output training
RSS data.
3. the method according to claim 1, wherein the three-dimensional parameterized RSS model based on Method Using Relevance Vector Machine is joined
Numberization is expressed as
It is dominated sighting distance degree, target is deviateed by target close to three dimensional parameter joints of node degree and static level of fading
It exports training RSS data and carries out parametrization expression:
Wherein, yiFor the corresponding trained RSS data of link i, M is number of links in sensor network, λi, γiAnd FiRespectively correspond phase
The target of link is answered to deviate sighting distance degree, target close to node degree and static level of fading, eiFor model training error, K (θ,
θi) it is the gaussian kernel function trained by three dimensional parameters, relation vectorIt estimates
Weight parameter ω=[ω1,...,ωM]T。
4. a kind of target locating method for the target localization and tracking system that data-driven modeling is realized based on Method Using Relevance Vector Machine,
It is characterised by comprising:
Realize the target localization and tracking system of data-driven modeling to any one of claims 1 to 3 based on Method Using Relevance Vector Machine
The three-dimensional parameterized RSS model carries out observation needed for data-driven conjunctive model calculates generation radio frequency chromatography imaging task
Matrix；
The reality for the object to be tracked that the target localization and tracking system of data-driven modeling will acquire is realized based on Method Using Relevance Vector Machine
When RSS data domain observation combine sparse restructing algorithm carry out radio frequency tomography in image reconstruction, determine object to be tracked
Target positioning and tracking.
5. method according to claim 4, which is characterized in that observing matrix specifically includes:
Wherein, φijIndicate that j-th of pixel is to the contribution weight of the link in the i-th row in observing matrix.
6. method according to claim 4, which is characterized in that realize the mesh of data-driven modeling based on Method Using Relevance Vector Machine
The real-time RSS data domain observation for the object to be tracked that mark locating and tracking system will acquire combines sparse restructing algorithm to carry out radio frequency
Image reconstruction in tomography determines the target positioning and tracking of object to be tracked, specifically includes:
The reality for the object to be tracked that the target localization and tracking system of data-driven modeling will acquire is realized based on Method Using Relevance Vector Machine
When RSS data domain observation combine sparse restructing algorithm carry out radio frequency tomography in image reconstruction；
Realize the target localization and tracking system of data-driven modeling by the imaging results figure after image reconstruction based on Method Using Relevance Vector Machine
The middle maximum single or multiple location of pixels of attenuation degree are as single goal or the estimated location of multiple target；
Realize that the target localization and tracking system of data-driven modeling is real-time by Kalman filtering and target based on Method Using Relevance Vector Machine
Positioning result combines, and carries out the dynamic trajectory tracking of object to be tracked.
7. a kind of RSS data model building device based on Method Using Relevance Vector Machine characterized by comprising
Preprocessing module, the RSS data for the multiple preset scenes that will be acquired are pre- according to preset processing mode progress RSS data
As output training RSS data after processing；
Multidimensional processing module, the RSS data collection for that will acquire carry out target corresponding with link and deviate according to different links
Sighting distance degree, target are solved close to node degree and static three dimensional parameters of level of fading as input training data；
Modeling module, for establishing the three-dimensional ginseng based on Method Using Relevance Vector Machine by exporting training RSS data and inputting training data
Numberization RSS model.
8. a kind of target localization and tracking system for the target locating for realizing data-driven modeling based on Method Using Relevance Vector Machine, special
Sign is, comprising:
Computing module calculates life for carrying out data-driven conjunctive model to three-dimensional parameterized RSS model as claimed in claim 7
Observing matrix needed for chromatographing imaging task at radio frequency；
The real-time RSS data domain observation of tracking module, the object to be tracked for will acquire combines sparse restructing algorithm to carry out
Image reconstruction in radio frequency tomography determines the target positioning and tracking of object to be tracked.
9. a kind of computer readable storage medium, which is characterized in that the computer readable storage medium is for storing program generation
Code, said program code require the described in any item RSS data modeling sides based on Method Using Relevance Vector Machine 1-3 for perform claim
Method.
10. a kind of computer readable storage medium, which is characterized in that the computer readable storage medium is for storing program generation
Code, said program code require 4-6 described in any item and realize data-driven modeling based on Method Using Relevance Vector Machine for perform claim
Target localization and tracking system target locating method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910222776.0A CN110095751A (en) | 2019-03-22 | 2019-03-22 | The target localization and tracking system of data-driven modeling is realized based on Method Using Relevance Vector Machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910222776.0A CN110095751A (en) | 2019-03-22 | 2019-03-22 | The target localization and tracking system of data-driven modeling is realized based on Method Using Relevance Vector Machine |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110095751A true CN110095751A (en) | 2019-08-06 |
Family
ID=67443078
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910222776.0A Pending CN110095751A (en) | 2019-03-22 | 2019-03-22 | The target localization and tracking system of data-driven modeling is realized based on Method Using Relevance Vector Machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110095751A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112394320A (en) * | 2020-04-26 | 2021-02-23 | 南京邮电大学 | Indoor high-precision centroid positioning method based on support vector machine |
| CN112720244A (en) * | 2020-12-18 | 2021-04-30 | 杭州电子科技大学 | Crystal grinding control method and system based on open-loop cross-coupling iterative learning |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120009882A1 (en) * | 2010-07-09 | 2012-01-12 | Neal Patwari | Statistical Inversion Method and System for Device-Free Localization in RF Sensor Networks |
| CN102427602A (en) * | 2011-12-06 | 2012-04-25 | 南京师范大学 | Sparse-based direct position determination method |
| CN104093202A (en) * | 2014-07-02 | 2014-10-08 | 南京信息工程大学 | Environment-adaptive device-free target positioning method |
| CN104200695A (en) * | 2014-08-15 | 2014-12-10 | 北京航空航天大学 | Vehicle co-location method based on special short range communication for vehicular access |
| CN107666705A (en) * | 2017-08-23 | 2018-02-06 | 中山大学 | A kind of dual spaces back projection radio frequency tomography method base, localization method and device |
-
2019
- 2019-03-22 CN CN201910222776.0A patent/CN110095751A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120009882A1 (en) * | 2010-07-09 | 2012-01-12 | Neal Patwari | Statistical Inversion Method and System for Device-Free Localization in RF Sensor Networks |
| CN102427602A (en) * | 2011-12-06 | 2012-04-25 | 南京师范大学 | Sparse-based direct position determination method |
| CN104093202A (en) * | 2014-07-02 | 2014-10-08 | 南京信息工程大学 | Environment-adaptive device-free target positioning method |
| CN104200695A (en) * | 2014-08-15 | 2014-12-10 | 北京航空航天大学 | Vehicle co-location method based on special short range communication for vehicular access |
| CN107666705A (en) * | 2017-08-23 | 2018-02-06 | 中山大学 | A kind of dual spaces back projection radio frequency tomography method base, localization method and device |
Non-Patent Citations (2)
| Title |
|---|
| JOEY WILSON AND NEAL PATWARI: "Fade-Level Skew-Laplace Signal Strength Model for Device-Free Localization with Wireless Networks", 《IEEE TRANSACTIONS ON MOBILE COMPUTING》 * |
| XIAOBING HONG等: "A RVM-based RSS model for Device-Free localization", 《PROCEEDINGS OF THE 36TH CHINESE CONTROL CONFERENCE》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112394320A (en) * | 2020-04-26 | 2021-02-23 | 南京邮电大学 | Indoor high-precision centroid positioning method based on support vector machine |
| CN112394320B (en) * | 2020-04-26 | 2023-06-23 | 南京邮电大学 | Indoor high-precision centroid positioning method based on support vector machine |
| CN112720244A (en) * | 2020-12-18 | 2021-04-30 | 杭州电子科技大学 | Crystal grinding control method and system based on open-loop cross-coupling iterative learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10853970B1 (en) | System for estimating a three dimensional pose of one or more persons in a scene | |
| US11182924B1 (en) | System for estimating a three dimensional pose of one or more persons in a scene | |
| Zhao et al. | mid: Tracking and identifying people with millimeter wave radar | |
| US11348279B1 (en) | System for estimating a three dimensional pose of one or more persons in a scene | |
| US11521373B1 (en) | System for estimating a three dimensional pose of one or more persons in a scene | |
| Rogister et al. | Asynchronous event-based binocular stereo matching | |
| CN109917818A (en) | Collaboratively searching based on ground robot contains method | |
| US11688139B1 (en) | System for estimating a three dimensional pose of one or more persons in a scene | |
| CN107529222A (en) | A kind of WiFi indoor locating systems based on deep learning | |
| CN107066806B (en) | Data Association and device | |
| CN112153736A (en) | Personnel action identification and position estimation method based on channel state information | |
| Wang et al. | RFID & vision based indoor positioning and identification system | |
| CN109766758A (en) | A kind of vision SLAM method based on ORB feature | |
| CN109859266A (en) | Vision positions and drawing practice simultaneously under a kind of big visual angle change based on pre-transform | |
| CN108430105A (en) | Distributed sensor networks cooperate with target state estimator and interference source passive location method | |
| CN111601253A (en) | Passive passive intelligent tracking and positioning method and system, storage medium and tracking and positioning terminal | |
| CN110095751A (en) | The target localization and tracking system of data-driven modeling is realized based on Method Using Relevance Vector Machine | |
| CN110223351A (en) | A kind of depth camera localization method based on convolutional neural networks | |
| Tian et al. | Resilient and distributed multi-robot visual slam: Datasets, experiments, and lessons learned | |
| CN111812978B (en) | Cooperative SLAM method and system for multiple unmanned aerial vehicles | |
| CN109902675A (en) | The method and apparatus of the pose acquisition methods of object, scene reconstruction | |
| CN116861779A (en) | Intelligent anti-unmanned aerial vehicle simulation system and method based on digital twinning | |
| CN104239719B (en) | Formation target plot-track Association Algorithm based on dual fuzzy topology under systematic error | |
| Dubois et al. | AirMuseum: a heterogeneous multi-robot dataset for stereo-visual and inertial simultaneous localization and mapping | |
| Kaplan et al. | Simulations studies of multisensor track association and fusion methods |
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: 20190806 |