CN107657215A - Indoor action trail movement semantic analytic method based on Passive Infrared Sensor - Google Patents
Indoor action trail movement semantic analytic method based on Passive Infrared Sensor Download PDFInfo
- Publication number
- CN107657215A CN107657215A CN201710798850.4A CN201710798850A CN107657215A CN 107657215 A CN107657215 A CN 107657215A CN 201710798850 A CN201710798850 A CN 201710798850A CN 107657215 A CN107657215 A CN 107657215A
- Authority
- CN
- China
- Prior art keywords
- neighborhood
- node
- semantic
- coding
- movement
- 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
- 230000033001 locomotion Effects 0.000 title claims abstract description 98
- 238000004458 analytical method Methods 0.000 title claims abstract description 25
- 230000009471 action Effects 0.000 title claims abstract description 14
- 230000004044 response Effects 0.000 claims abstract description 64
- 239000000178 monomer Substances 0.000 claims abstract description 20
- 238000000605 extraction Methods 0.000 claims abstract description 13
- 238000012731 temporal analysis Methods 0.000 claims abstract description 11
- 238000000700 time series analysis Methods 0.000 claims abstract description 11
- 238000012163 sequencing technique Methods 0.000 claims abstract description 10
- 238000000034 method Methods 0.000 claims description 25
- 238000012732 spatial analysis Methods 0.000 claims description 15
- 238000009434 installation Methods 0.000 claims description 4
- 238000011835 investigation Methods 0.000 claims description 3
- 239000000203 mixture Substances 0.000 abstract description 3
- 230000006399 behavior Effects 0.000 description 12
- 230000008569 process Effects 0.000 description 12
- 238000011160 research Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000009466 transformation Effects 0.000 description 4
- 238000006243 chemical reaction Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 235000012364 Peperomia pellucida Nutrition 0.000 description 2
- 240000007711 Peperomia pellucida Species 0.000 description 2
- 230000003542 behavioural effect Effects 0.000 description 2
- 238000000205 computational method Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000009412 basement excavation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000006854 communication Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
Landscapes
- Engineering & Computer Science (AREA)
- Human Computer Interaction (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of indoor action trail movement semantic analytic method based on Passive Infrared Sensor, including:(1) according to indoor plane figure, sensing station and neighbouring relations structure sensor network figure;(2)With indoor point of interest definition space analysis window, five neighbourhood models comprising center point of interest and four neighborhood direction sensors are established as minimum space unit, carry out neighborhood coding;(3)With the related interest moment definition time series analysis window of point of interest, forward and backward response sequence delimited with sensor response time and response sequencing;(4)Rely on five neighbourhood models and neighborhood to encode, establish monomer and encoded with group movement;(5)Extraction of semantics rule is established, calculates the motion encoded of forward and backward response sequence composition, corresponding minimum semantic primitive is explained to obtain final movement semantic.Movement semantic relation is dissolved into sensor information expression by the present invention, solves the semantization parsing problem of the mankind's activity track under only mixed and disorderly response sequence.
Description
Technical field
The present invention relates to the related track of computer and GIS-Geographic Information System field, particularly sensor network and human body row
For analysis.
Background technology
With Internet of Things, the fast development of mobile emerging technology, geography information location technology is increasingly mature, space-time track number
According to collection turn into normality.Wherein, pir sensor (passive infrared sensor) is also known as Passive Infrared Sensor,
Have benefited from its passive infrared mode and detect advantage of the human body infrared information in cost and power consumption, be widely used in room
The fields such as interior monitoring, security location-based service.Simultaneously because its data acquiring mode is easy and can effectively protect individual privacy, also often
It is used for analysis individual and group behavior.Individual and group behavior based on pir sensor analyze its difficult point and are abstract data
Huge space-time data collection is measured, generation can track corresponding with human behavior from mixed and disorderly response sequence.
Under traditional trajectory analysis pattern, for example in terms of the reconstruction of track and the mode excavation of track, have been achieved with
A series of achievements in research.Existing research method has the trajectory range reconstruct based on demarcation, the track reconstructing based on filtering, is based on
The 3D cycle behaviors track reconstructing of 2D behavior sequences, Frequent Trajectory Patterns based on Moving Objects from Surveillance Video excavate etc., energy
It is enough to generate trajectory path from a series of tracing point.But above-mentioned reconstruct mode is only reconstructing path, is not related to extraction of semantics.
For pir sensor, its record result be sense human infrared radiation period (including the time started and at the end of
Between), coordinate information is only the installation position of sensor.It can not determine to pass through the number of specific region, and also None- identified passes through
Object, cause to be difficult to the behavior state for extracting special object.In existing track applied analysis, such as the means such as video monitoring, all
Try hard to excavate behavioural information from track, and then as the support of the great researchs such as human behavioral mode, accident prediction
Means.And pir sensor data are completely absent dominant behavior semanteme, it is impossible to meet the semantic great demand of track behavior,
To analyzing and using huge difficulty is above caused, being badly in need of scientific and effective solution.
Therefore, from the visual angle of information integration, the inside semantic information of motor behavior is introduced, and combines the connectedness of sensor
Topology is to solve the above problems, and then extracts the possible approaches of effective track.And track data is in itself and in the absence of semanteme
Information, semantic information are the results of data mining.If trajectory analysis model forgives semantic feature and architectural feature, information excavating with
Track reduction is synchronous, can at utmost ensure information integrity, simplifies data analysis complexity.
The content of the invention
Goal of the invention:The problem of existing for existing trajectory analysis method, present invention aims at provide one kind to be based on quilt
The indoor action trail movement semantic analytic method of dynamic infrared sensor, this method are based on human body motion feature and Spatial Semantics mould
Type frame structure, human body behavior is illustrated using the position and response sequence of pir sensor node, carries out the motion language of indoor action trail
Justice parsing.
Technical scheme:The present invention is a kind of indoor action trail movement semantic parsing side based on Passive Infrared Sensor
Method, specific steps include:
(1) according to room area plan and infrared sensor distribution coordinate position, with reference to the syntople of sensor, with
Sensor is as node, the sensor network figure of the undirected no weight of structure;
(2) using selected analysis point of interest as search center, the most adjacent section of four direction up and down is outwards expanded
Point composition spatial analysis window, filters out selected point of interest and its sensor section in four neighborhood directions out of spatial analysis window
Point, connected subgraph is formed, establish five neighbourhood models comprising Centroid and four neighborhood direction nodes as minimum space list
Position, neighborhood coding is carried out to connected subgraph using the blade objects in Geometrical algebra, realizes and is encoded from connected subgraph to algebraization
Transformation;
(3) in spatial analysis window, the interest period is selected, according to sensor response time settling time analysis window.
Using the response data of Centroid as core, remaining neighborhood node splits data into preceding response sequence with after by response sequencing
Response sequence;
(4) minimum semantic primitive is established with five neighbourhood models in step (2), five neighbourhood model interior joints is showed
Different spaces condition conversion is that monomer is motion encoded or group movement coding, every kind of coding correspond to a kind of specific movement semantic;
(5) extraction of semantics rule by definition, the various combination mode of forward and backward response sequence is judged, in terms of neighborhood coding
Calculation obtains that monomer is motion encoded or group movement coding, and the corresponding minimum predefined movement semantic type of semantic primitive is converted into most
Whole movement semantic.
Further, the azimuth angle interval computational methods of the neighborhood in the step 2 and the blade of five neighbourhood models are encoded
Mode is respectively:
(2.1) spatial window is divided by azimuth, is divided into upper and lower, left and right totally four neighborhood directions, four neighborhoods
Azimuth angle interval is respectively:
Wherein θ is azimuth angle interval, and its value is the clockwise direction using point of interest due north as Fixed Initial Point.
(2.2) meter is applied to the connected subgraph structure that own node, upper node, right node, lower node, left sibling are formed
The coding mode of calculation, utilize the blade object groups in Geometrical algebra<eC, eT, eB, eL, eR>Algebraization coding is carried out to node,
eC, eT, eB, eL, eRCenter, upper and lower, left and right totally five nodes are corresponded to respectively.
Further, the computational methods of the time window T in the step 3 and the classification of node response sequence and computation sequence
Constraint be judged as:
(3.1) time window realizes that formula is
Wherein, v is the normal walking speed of human body, and s is sensor radius of investigation, and d is the feasible path of sensor installation
Distance.
(3.2) node response sequence is successively classified by the response time, and its classifying rules can be represented by following formula:
Wherein preNodeList and nextNodeList represents the forward and backward sequence of response, T respectivelyNeighborhoodAnd TCenterNeighbour is represented respectively
The response time of domain node and Centroid, T are time window, eFieldeCenterOr eCentereNeighborhoodRepresent the geometry product of neighbor domain of node coding
As a result, sequencing difference is calculated by response condition.
Further, the motion encoded tool with group movement coding of monomer during minimum semantic primitive is established in the step 4
Body method is:
The motion encoded geometry for neighborhood node and Centroid neighborhood coding of monomer accumulates result, and institute is comprising movement semantic
Stand, pass through, turning, into, leave and return to six kinds;The monomer that group movement coding connects different characteristic by "+" number is transported
Dynamic expression, the multivector structure formed in Geometrical algebra space, it includes movement semantic to separate with polymerizeing two kinds.
Further, semantic transformation rule is specially in the step 5:
Wherein, before and after preNodeList and nextNodeList two row represent in response sequence responsive node number, it is preceding
Sequence relation list shows whether forward and backward sequence responsive node is identical afterwards, motion encoded to represent to encode with neighborhood with movement semantic row
The monomer or group movement coding and its corresponding semantic classification that geometry product is calculated, eCRepresent Centroid coding, eiAnd ej
Represent section neighborhood of a point coding respectively, subscript meets i, j ∈ { R, L, T, B } and i ≠ j, R, L, T, B represent respectively it is right, left, upper,
Lower four direction.
Beneficial effect:Sensor response is converted into node by the present invention using the superior spatial and temporal expression ability of Geometrical algebra
Between incidence relation, utilization space and time series analysis window constrain track semanteme analysis condition, establish minimum semantic parsing
Unit, response sequence is completed to the one-to-one transformation rule of movement semantic, the automation semanteme for realizing indoor track solves
Analysis.
Brief description of the drawings
Fig. 1 is the core procedure schematic flow sheet of the embodiment of the present invention;
Fig. 2 is the blade coding schematic diagrams of five neighbourhood models in the embodiment of the present invention;
Fig. 3 is motion segmentation time window time calculating schematic diagram in the embodiment of the present invention;
Fig. 4 is minimum semantic primitive and motion encoded schematic diagram in the embodiment of the present invention;
Fig. 5 is movement locus extraction of semantics case result schematic diagram in the embodiment of the present invention.
Embodiment
Technical solution of the present invention is described in detail below, but protection scope of the present invention is not limited to the implementation
Example.
The embodiment of the present invention provides a kind of indoor action trail movement semantic analytic method based on Passive Infrared Sensor,
Its core procedure flow is as shown in Figure 1.First, coordinate position is distributed according to room area plan and infrared sensor, with reference to
The syntople of sensor, using sensor as node, build the sensor network figure of undirected no weight;Then, with selected
Point of interest is analyzed as search center, the most adjacent node for outwards expanding four direction up and down forms spatial analysis window,
Selected point of interest and its sensor node in four neighborhood directions are filtered out out of spatial analysis window, forms connected subgraph, is established
Five neighbourhood models comprising Centroid and four neighborhood direction nodes are as minimum space unit, using in Geometrical algebra
Blade objects carry out neighborhood coding to connected subgraph, realize the transformation encoded from connected subgraph to algebraization;Secondly, in space
In analysis window, the interest period is selected, according to sensor response time settling time analysis window, with the number of responses of Centroid
According to for core, remaining neighborhood node preceding response sequence and rear response sequence are splitted data into by response sequencing;Then, with five
Neighbourhood model establishes minimum semantic primitive, and the different spaces condition conversion that five neighbourhood model interior joints are showed moves for monomer
Coding or group movement coding, a kind of every kind of motion encoded specific movement semantic of correspondence;Finally, extraction of semantics rule by definition
Then, judge the various combination mode of forward and backward response sequence, be calculated that monomer is motion encoded or group movement is compiled with neighborhood coding
Code, and the corresponding minimum predefined movement semantic type of semantic primitive is converted into final movement semantic.When it is implemented, including with
Lower committed step:
Process one:Experimental data chooses and pretreatment, builds sensor network figure
Step 1:Laboratory data needed for choosing, obtains the essential information data of data, for example, comprising sensor
Number, sensor ID, record start/termination time, record total number etc..
Step 2:According to indoor plan, reference sensor distributing position and neighbouring relations, space inner sensor is established
Undirected no weight network.
Process two:Spatial analysis window based on point of interest determines
Step 1:Choose important sensor node in pilot region and, as point of interest, establish what is put centered on the point of interest
Spatial analysis window, outwards expands the adjacent node of four direction up and down, and its way of search is searching sensor network figure
In sensor node closest on the four direction that is directly connected with Centroid.
Step 2:Filtered out out of spatial analysis window comprising Centroid (selected point of interest) and its four neighborhoods (upper bottom left
It is right) sensor node in direction, its spatial distribution interval division determines according to equation below.
θ values are the clockwise direction using point of interest due north as Fixed Initial Point.And then filtered out from raw sensor network
Include all sensor connected subgraphs for choosing node.
Step 3:Neighborhood coding is carried out to sensor connected subgraph, by Centroid and four neighborhood directions (it is upper, right, under,
It is left) define five neighbourhood models<C, T, R, B, L>, the connection to Centroid, upper node, right node, lower node, left sibling composition
Subgraph structure is applied to the coding mode calculated, and the embodiment of the present invention uses the blade object groups in Geometrical algebra space<eC,
eT, eB, eL, eR>Encoded.Specifically five neighbourhood model centers and the node of sensor up and down are compiled as shown in Fig. 2
Code.
Process three, the time series analysis window based on the interest moment determine:
Step 1:According to the acquisition time span of experimental data, choose and meet period of analysis demand as the interest moment,
The time series analysis window excavated track is determined within the interest moment.
Step 2:Time series analysis window is used for catching the responsive state of Centroid and neighborhood node, need to ensure adjacent node
It is different according to sensor distribution distance to the continuous capturing of center joint movements.It is assumed that v is the normal walking speed of human body, s is
Sensor radius of investigation, d are the feasible path distance of sensor installation, and time series analysis window T can be by formulaIt is determined that.
If as shown in figure 3, when having adjacent node response out of a certain node the restriction response time, then it is assumed that the joint movements are continuous
And the two nodes are related, if without response, then it is assumed that motion terminates.
Step 3:With the sequencing of time series analysis window interior nodes response to node sequencing, its motion encoded computation sequence
Depending on response sequence, specific constraint rule can be represented by following formula
By the priority of the response time compared with Centroid, response sequence is divided into preceding response sequence
And rear response sequence (nextNodeList) (preNodeList).The motion encoded computation sequence of preceding response sequence is eNeighborhoodeCenter,
Then the computation sequence of response sequence is eCentereNeighborhood。
Process four, semantic basis unit with it is motion encoded:
Step 1:Because the motion of human body track is the process that node is propagated to neighborhood node, its motion state is in five nodes
In show different space structures, this spatiality is converted into monomer is motion encoded or group movement encodes, it is right respectively
Answer different movement semantics.
Step 2:Calculated as shown in figure 4, monomer is motion encoded with neighborhood node and Centroid neighborhood coding geometry product
Arrive k-blade (k=1,2,3) geometry product coding form, comprising movement semantic be a kind standing, 4 kinds pass through, 8 kinds turning, 4
Kind enter, 4 kinds leave and 8 kinds return.Such as when motion state is eL-eC-eRWhen, that is, represent from a left side into again to the right side it is simple
Motion is passed through, monomer is motion encoded to be designated as eLeCeR.By movement semantic classification specific coding, semanteme of standing is { O }, is represented not
Motion;It is { e to pass through semantemeBeCeT, eTeCeB, eLeCeR, eReCeL, such as eBeCeTTo pass through from top to bottom;Turn semanteme is
{eTeCeR, eReCeT, eReCeB, eBeCeR, eBeCeL, eLeCeB, eLeCeT, eTeCeL, the steering behavior by intermediate node is represented,
Such as eTeCeRFor by up to again to the right;It is { e into semantemeLeC, eTeC, eReC, eBeC, such as eLeCIn entering from left side
The heart;It is { e to leave semantemeCeL, eCeT, eCeR, eCeB, such as eCeLTo be left to the left from center;It is { e to return to semantemeReCeR,
eLeCeL, eBeCeB, eTeCeT, eCeLeC, eCeReC, eCeBeC, eCeTeC, such as eReCeRIt is right to enter middle return again from right side
Side.
Step 3:As shown in figure 4, the monomer motion table that group movement coding connects different characteristic with "+" number reaches, formed several
Multivector structure in what algebraic space, it includes 6 kinds and separated polymerize movement semantic with 6 kinds.For example, when motion state is
eC-eL, eC-eBWhen, that is, represent from Centroid simultaneously to left sibling and the split movement of lower node, its group movement coding note
For eCeL+eCeB.By movement semantic classification specific coding, it is { e to separate semantemeCeL+eCeB, eCeL+eCeT, eCeL+eCeR, eCeT+
eCeB, eCeT+eCeB, eCeB+eCeR, such as eCeL+eCeBTo be separated from Centroid to left sibling and lower node;Aggregation semantic is
{eBeC+eLeC, eTeC+eLeC, eLeC+eReC, eTeC+eReC, eBeC+eTeC, eReC+eBeC, wherein eLeC+eBeCFor from left sibling
Assemble with lower node to Centroid.
Process five, extraction of semantics rule convert with movement semantic
Step 1:According to the syntagmatic of forward and backward response sequence, extraction of semantics rule is defined, each semantic type is corresponding
A kind of rule information, realize that man-to-man semantic information converts, specific rules are as shown in the table:
Wherein, before and after preNodeList and nextNodeList two row represent in response sequence responsive node number, it is preceding
Sequence relation list shows whether forward and backward sequence responsive node is identical afterwards, motion encoded to represent to encode with neighborhood with movement semantic row
The monomer or group movement coding and its corresponding semantic classification that geometry product is calculated, eCRepresent Centroid coding, eiAnd ej
Represent section neighborhood of a point coding respectively, subscript meets i, j ∈ { R, L, T, B } and i ≠ j, R, L, T, B represent respectively it is right, left, upper,
Lower four direction
Step 2:The forward and backward response sequence that will split in time series analysis window, foundation is with the response time successively for constraint
Computation sequence, the calculation accumulated using responsive node neighborhood coding geometry, is tried to achieve motion encoded and right according to extraction of semantics rule
The predefined semantic type of minimum semantic primitive is answered to be converted into final semanteme.
Step 3:Above-mentioned semantic resolving is repeated, generation different time is interior, the semantic information of diverse location point, and carries out
Related statistics and visual analyzing.
Below by taking the sensor data analysis to certain electric research laboratory as an example, present invention method is illustrated
Specific implementation process.
Process one:Experimental data chooses and pretreatment, structure sensor connected graph and coding
Step 1:Selected object of experiment region, the data of the present embodiment are certain electric research laboratory, are amounted in laboratory
213 sensors, the data record for up to 1 year, Annual distribution is from March 24,21 days to 2007 March in 2006.Note
Record data are divided into four parts, respectively sensor id, target acquisition initial time, and target acquisition terminates time and detection
Validity, the data share 30239000 records.
Step 2:The plan and sensor distributing position in foundation laboratory, establish the undirected no weight of space inner sensor
Network.
Process two:Spatial analysis window based on point of interest determines
Step 1:Because experimental data is certain laboratory detection data, therefore important sensor node is as emerging in chosen area
Interesting point (such as important office, lift port, meeting room), establishes the spatial analysis window put centered on the point of interest, according to biography
Sensor network outwards expands the most adjacent node of four direction up and down.
Step 2:Filtered out out of spatial analysis window comprising Centroid (selected point of interest) and its four neighborhoods (upper bottom left
It is right) sensor node in direction, use geometry generation as sensor subgraph, and by Centroid and the node in four neighborhood directions
Blade object groups in number space<eC, eT, eB, eL, eR>Encoded.
Process three, the time series analysis window based on the interest moment determine:
Step 1:Embodiment packet chooses the period for meeting analysis demand as emerging containing the total length data record of 1 year
Interesting moment (such as morning working, lunch, come off duty at dusk, working day, weekend), determine that track excavates within the interest moment when
Between analysis window.
Step 2:Time series analysis window is used for catching the responsive state of Centroid and neighborhood node, assumes in the present embodiment
Normal person's speed of travel is 1.2 meter per seconds, and each sensor distribution is close, and seamless between sensor investigative range or gap is small,
The scope of each sensor is 2 meters, therefore people by a sensor and enters the adjoining sensor sensing time for 2 seconds~4 seconds,
Median is taken to set time window as 3 seconds.If have adjacent node response in 3 seconds from a certain node, then it is assumed that the node
Motion is continuous and the two nodes are related, if without response, then it is assumed that motion terminates.
Step 3:With the sequencing of time window interior nodes response to node sequencing, response sequence is divided into preceding response
Sequence (preNodeList) and rear response sequence (nextNodeList).
Process four, semantic basis unit with it is motion encoded:
Using five neighbourhood models as fundamental space unit, by between node and neighborhood node as caused by human motion communication process
Numeralization, spatially the difference of state be converted into that monomer is motion encoded or group movement coding,
Process five, extraction of semantics rule convert with movement semantic
Step 1:According to the syntagmatic of forward and backward response sequence, extraction of semantics rule is defined, each semantic type is corresponding
A kind of rule information, realize man-to-man semantic information conversion.
Step 2:It is adjacent using responsive node according to the computation sequence under response time constraint by the forward and backward response sequence of target
The calculation of domain coding geometry product, tries to achieve motion encoded according to extraction of semantics rule, and corresponding minimum semantic primitive is predefined
Semantic type is converted into final semanteme.
Step 3:Above-mentioned semantic resolving is repeated, extracts the track semantic information in different time, and carry out related system
Meter.So that " entrance " is semantic as an example, " entrance " extracted and count Monday to Wednesday in seclected time section is semantic, according to " entrance " language
The frequency and spatial distribution that justice occurs carry out Visualization, and deeper position generation " entrance " the behavior frequency that represents of color is got over
It is high.As shown in figure 5, a, b, c tri- opens the semantic frequency that subgraph respectively show Monday to Wednesday.
Claims (7)
- A kind of 1. indoor action trail movement semantic analytic method based on Passive Infrared Sensor, it is characterised in that including with Lower step:(1) according to room area plan and infrared sensor distribution coordinate position, with reference to the syntople of sensor, with sensing Device is as node, the sensor network figure of the undirected no weight of structure;(2) using selected analysis point of interest as search center, the most adjacent node group of four direction up and down is outwards expanded Into spatial analysis window, selected point of interest and its sensor node in four neighborhood directions, structure are filtered out out of spatial analysis window Into connected subgraph, five neighbourhood models comprising Centroid and four neighborhood direction nodes are established as minimum space unit, profit Neighborhood coding is carried out to connected subgraph with the blade objects in Geometrical algebra, realizes and turns from connected subgraph to what algebraization encoded Become;(3) in spatial analysis window, the interest period is selected, according to sensor response time settling time analysis window, in The response data of heart node is core, and remaining neighborhood node splits data into preceding response sequence and rear response by response sequencing Sequence;(4) minimum semantic primitive is established with five neighbourhood models in step (2), the difference that five neighbourhood model interior joints are showed Spatiality is converted into that monomer is motion encoded or group movement coding, a kind of every kind of corresponding specific movement semantic of coding;(5) extraction of semantics rule by definition, judge the various combination mode of forward and backward response sequence, calculated with neighborhood coding To monomer is motion encoded or group movement coding, the corresponding minimum predefined movement semantic type of semantic primitive is converted into final fortune Dynamic semanteme.
- A kind of 2. indoor action trail movement semantic parsing side based on Passive Infrared Sensor according to claim 1 Method, it is characterised in that the spatial analysis window in the step (2) is divided by azimuth, is divided into totally four, upper and lower, left and right Neighborhood direction, the azimuth angle interval of four neighborhoods are respectively:Wherein θ is azimuth angle interval, and its value is the clockwise direction using point of interest due north as Fixed Initial Point.
- A kind of 3. indoor action trail movement semantic parsing side based on Passive Infrared Sensor according to claim 1 Method, it is characterised in that the blade object groups in Geometrical algebra are utilized in the step (2)<eC,eT,eB,eL,eR>Centering respectively Heart node and node up and down carry out algebraization coding, eC,eT,eB,eL,eRCenter, upper and lower, left and right totally five are corresponded to respectively Individual node.
- A kind of 4. indoor action trail movement semantic parsing side based on Passive Infrared Sensor according to claim 1 Method, it is characterised in that the time series analysis window T in the step (3) determines according to below equation:<mrow> <mi>T</mi> <mo>=</mo> <mfrac> <mrow> <mi>d</mi> <mo>-</mo> <mn>2</mn> <mi>s</mi> </mrow> <mi>v</mi> </mfrac> </mrow>Wherein, v is the normal walking speed of human body, and s is sensor radius of investigation, and d is the feasible path distance of sensor installation.
- A kind of 5. indoor action trail movement semantic parsing side based on Passive Infrared Sensor according to claim 1 Method, it is characterised in that the classifying rules of preceding response sequence and rear response sequence can be represented by following formula in the step (3):Wherein preNodeList and nextNodeList represents forward and backward response sequence, T respectivelyNeighborhoodAnd TCenterNeighborhood section is represented respectively The response time of point and Centroid, T is analysis time window, eNeighborhoodeCenterOr eCentereNeighborhoodRepresent the geometry product of neighbor domain of node coding As a result, sequencing difference is calculated by response condition.
- A kind of 6. indoor action trail movement semantic parsing side based on Passive Infrared Sensor according to claim 1 Method, it is characterised in that in the step (4), the motion encoded geometry for neighborhood node and Centroid neighborhood coding of monomer accumulates As a result, comprising movement semantic to stand, passing through, turning, into, leave and return to six kinds;Group movement coding passes through "+" number The monomer motion table of connection different characteristic reaches, and the multivector structure formed in Geometrical algebra space, it is comprising movement semantic Separate with polymerizeing two kinds.
- A kind of 7. indoor action trail movement semantic parsing side based on Passive Infrared Sensor according to claim 1 Method, it is characterised in that the extraction of semantics rule in the step (5), be specially:Wherein, before and after preNodeList and nextNodeList two row represent in response sequence responsive node number, front and rear sequence Row relation list shows whether forward and backward sequence responsive node is identical, and motion encoded and movement semantic row represent to encode geometry with neighborhood Monomer or the group movement coding and its corresponding semantic classification that product is calculated, eCRepresent Centroid coding, eiAnd ejRespectively Section neighborhood of a point coding is represented, subscript meets i, j ∈ { R, L, T, B } and i ≠ j, R, L, T, B represent right, left, upper and lower four respectively Individual direction.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710798850.4A CN107657215B (en) | 2017-09-07 | 2017-09-07 | Indoor behavior track motion semantic analysis method based on passive infrared sensor |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710798850.4A CN107657215B (en) | 2017-09-07 | 2017-09-07 | Indoor behavior track motion semantic analysis method based on passive infrared sensor |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107657215A true CN107657215A (en) | 2018-02-02 |
CN107657215B CN107657215B (en) | 2020-01-21 |
Family
ID=61129306
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710798850.4A Active CN107657215B (en) | 2017-09-07 | 2017-09-07 | Indoor behavior track motion semantic analysis method based on passive infrared sensor |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107657215B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108966139A (en) * | 2018-03-28 | 2018-12-07 | 厦门上特展示系统工程有限公司 | Location information acquisition and analysis device and method |
CN109561079A (en) * | 2018-11-12 | 2019-04-02 | 兴科迪科技(泰州)有限公司 | A kind of semantic communication agreement, communication means and system |
CN109668094A (en) * | 2019-01-29 | 2019-04-23 | 盐城骏拔汽车零部件有限公司 | Intelligent road lamp system |
CN109871469A (en) * | 2019-02-28 | 2019-06-11 | 浙江大学城市学院 | Tuftlet crowd recognition method based on dynamic graphical component |
CN113750283A (en) * | 2021-09-26 | 2021-12-07 | 中国人民解放军联勤保障部队第九六〇医院 | Intelligent sterilization and disinfection and personnel accident guarantee system and method for shelter changing room |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101694629A (en) * | 2009-10-23 | 2010-04-14 | 北京邮电大学 | Context sensing application platform based on main body and work method thereof |
CN106570081A (en) * | 2016-10-18 | 2017-04-19 | 同济大学 | Semantic net based large scale offline data analysis framework |
-
2017
- 2017-09-07 CN CN201710798850.4A patent/CN107657215B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101694629A (en) * | 2009-10-23 | 2010-04-14 | 北京邮电大学 | Context sensing application platform based on main body and work method thereof |
CN106570081A (en) * | 2016-10-18 | 2017-04-19 | 同济大学 | Semantic net based large scale offline data analysis framework |
Non-Patent Citations (1)
Title |
---|
冯琳耀: "基于室内传感网数据的行为轨迹几何代数建模与特征分析方法", 《中国优秀硕士学位论文全文数据库》 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108966139A (en) * | 2018-03-28 | 2018-12-07 | 厦门上特展示系统工程有限公司 | Location information acquisition and analysis device and method |
CN109561079A (en) * | 2018-11-12 | 2019-04-02 | 兴科迪科技(泰州)有限公司 | A kind of semantic communication agreement, communication means and system |
CN109668094A (en) * | 2019-01-29 | 2019-04-23 | 盐城骏拔汽车零部件有限公司 | Intelligent road lamp system |
CN109871469A (en) * | 2019-02-28 | 2019-06-11 | 浙江大学城市学院 | Tuftlet crowd recognition method based on dynamic graphical component |
CN109871469B (en) * | 2019-02-28 | 2021-09-24 | 浙江大学城市学院 | Small cluster crowd identification method based on dynamic graphics primitives |
CN113750283A (en) * | 2021-09-26 | 2021-12-07 | 中国人民解放军联勤保障部队第九六〇医院 | Intelligent sterilization and disinfection and personnel accident guarantee system and method for shelter changing room |
CN113750283B (en) * | 2021-09-26 | 2022-11-15 | 中国人民解放军联勤保障部队第九六〇医院 | Intelligent sterilization and disinfection and personnel accident guarantee system and method for shelter dressing room |
Also Published As
Publication number | Publication date |
---|---|
CN107657215B (en) | 2020-01-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107657215A (en) | Indoor action trail movement semantic analytic method based on Passive Infrared Sensor | |
CN104933722B (en) | A kind of method for detecting image edge based on Spiking convolutional neural networks models | |
CN103021136B (en) | System and method for monitoring swimming pool drowning based on internet of things | |
Moore et al. | Visual crowd surveillance through a hydrodynamics lens | |
Zhang et al. | A review on node deployment of wireless sensor network | |
CN106774314A (en) | A kind of home-services robot paths planning method based on run trace | |
CN109299401A (en) | Metropolitan area space-time stream Predicting Technique based on deep learning model LSTM-ResNet | |
CN104933417A (en) | Behavior recognition method based on sparse spatial-temporal characteristics | |
CN108537818A (en) | Crowd's trajectory predictions method based on cluster pressure LSTM | |
Sedky et al. | Classification of smart video surveillance systems for commercial applications | |
CN103679215A (en) | Video monitoring method based on group behavior analysis driven by big visual big data | |
CN102663775A (en) | Target tracking method oriented to video with low frame rate | |
CN113985897B (en) | Mobile robot path planning method based on pedestrian track prediction and social constraint | |
CN103902966A (en) | Video interaction event analysis method and device base on sequence space-time cube characteristics | |
CN105959270A (en) | Network attack detection method based on spectral clustering algorithm | |
CN102224526A (en) | A system and a method for identifying human behavioural intention based on an effective motion analysis | |
Zhang et al. | Video-driven group behavior simulation based on social comparison theory | |
Nyayu et al. | Analysis on swarm robot coordination using fuzzy logic | |
CN114372503A (en) | Cluster vehicle motion trail prediction method | |
CN102567715A (en) | Human body action hierarchical identification method based on pyroelectric infrared detection | |
CN116682175A (en) | Workshop personnel dangerous behavior detection method under complex environment | |
CN113627326B (en) | Behavior recognition method based on wearable equipment and human skeleton | |
He et al. | Motion pattern analysis in crowded scenes by using density based clustering | |
CN109241952B (en) | Figure counting method and device in crowded scene | |
Shao et al. | A real-time algorithm for small group detection in medium density crowds |
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 |