CN107657215B - Indoor behavior track motion semantic analysis method based on passive infrared sensor - Google Patents
Indoor behavior track motion semantic analysis method based on passive infrared sensor Download PDFInfo
- Publication number
- CN107657215B CN107657215B CN201710798850.4A CN201710798850A CN107657215B CN 107657215 B CN107657215 B CN 107657215B CN 201710798850 A CN201710798850 A CN 201710798850A CN 107657215 B CN107657215 B CN 107657215B
- Authority
- CN
- China
- Prior art keywords
- neighborhood
- motion
- semantic
- node
- response
- 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.)
- Active
Links
Images
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
Abstract
The invention discloses a semantic analysis method for indoor behavior track motion based on a passive infrared sensor, which comprises the following steps: (1) constructing a sensor network diagram according to the indoor plan, the sensor positions and the adjacent relation; (2) defining a space analysis window by using an indoor interest point, establishing a five-neighborhood model containing a central interest point and four neighborhood direction sensors as a minimum space unit, and performing neighborhood coding; (3) defining a time analysis window by the interest time related to the interest point, and defining a front response sequence and a rear response sequence by the response time and the response sequence of the sensor; (4) building monomer and group motion codes by relying on a five-neighborhood model and neighborhood codes; (5) establishing a semantic extraction rule, calculating motion codes formed by the front and rear response sequences, and interpreting a corresponding minimum semantic unit to obtain a final motion semantic. The invention integrates the motion semantic relation into the sensor information expression, and solves the semantic analysis problem of the human activity track under only a disordered response sequence.
Description
Technical Field
The invention relates to the field of computers and geographic information systems, in particular to analysis of tracks and human body behaviors related to a sensor network.
Background
With the rapid development of the internet of things and mobile emerging technologies, the geographic information positioning technology is mature day by day, and the acquisition of the space-time trajectory data becomes a normal state. Among them, the PIR sensor (passive infrared sensor) is also called a passive infrared sensor, and has been widely used in the fields of indoor monitoring, security location service, etc. owing to the advantages of its passive infrared mode in detecting infrared information of a human body in terms of cost and power consumption. Meanwhile, the method is easy in data acquisition mode and can effectively protect individual privacy, and the method is also often used for analyzing individual and group behaviors. The difficulty in PIR sensor-based analysis of individual and group behaviors is to abstract a massive amount of spatiotemporal data sets, generating trajectories from a chaotic response sequence that can correspond to human behavior.
Under the traditional track analysis mode, for example, in the aspects of track reconstruction and track pattern mining, a series of research results are obtained. The existing research methods include track space reconstruction based on calibration, track reconstruction based on filtering, 3D periodic behavior track reconstruction based on 2D behavior sequences, frequent track mode mining based on monitoring video moving targets and the like, and can generate an out-of-track path from a series of track points. However, the above reconstruction method is only path reconstruction and does not involve semantic extraction. For the PIR sensor, the recording result is the time period (including the starting time and the ending time) of sensing the infrared radiation of the human body, and the coordinate information is only the arrangement position of the sensor. It cannot determine the number of people crossing a specific area or identify crossing objects, resulting in difficulty in extracting the behavior state of a specific object. In the conventional track application analysis, for example, means such as video monitoring tries to extract behavior information from a track, and the behavior information is further used as a support means for important research such as human behavior patterns and emergency prediction. The PIR sensor data does not have dominant behavior semantics at all, the huge requirements of the trajectory behavior semantics cannot be met, great difficulties are caused on analysis and application, and scientific and effective solutions are urgently needed.
Therefore, from the perspective of information integration, introducing internal semantic information of a motion behavior and combining with a connectivity topology of a sensor are possible ways to solve the above problems and further extract an effective track. And the track data does not have semantic information per se, and the semantic information is the result of data mining. If the track analysis model includes semantic features and structural features, information mining and track reduction are synchronous, so that the information integrity can be ensured to the maximum extent, and the data analysis complexity is simplified.
Disclosure of Invention
The purpose of the invention is as follows: aiming at the problems of the existing track analysis method, the invention aims to provide an indoor behavior track motion semantic analysis method based on a passive infrared sensor.
The technical scheme is as follows: the invention relates to a semantic analysis method for indoor behavior track motion based on a passive infrared sensor, which comprises the following specific steps:
(1) according to the indoor area plan and the distribution coordinate position of the infrared sensor, combining the adjacency relation of the sensors, and taking the sensors as nodes, constructing a sensor network diagram without direction and weight;
(2) the method comprises the steps of taking a selected analysis interest point as a search center, outwards expanding the most adjacent nodes in four directions of up, down, left and right to form a space analysis window, screening the selected interest point and sensor nodes in four adjacent domain directions from the space analysis window to form a connected subgraph, establishing a five-adjacent-domain model containing the center node and the four adjacent-domain-direction nodes as a minimum space unit, and performing adjacent-domain coding on the connected subgraph by using a blank object in geometric algebra to realize the conversion from the connected subgraph to algebraic coding;
(3) within the spatial analysis window, a time period of interest is selected, and a temporal analysis window is established in accordance with the sensor response time. Taking the response data of the central node as a core, and dividing the data into a front response sequence and a rear response sequence by the rest neighborhood nodes according to the response sequence;
(4) establishing a minimum semantic unit by the five-neighborhood model in the step (2), converting different space states presented by nodes in the five-neighborhood model into monomer motion codes or group motion codes, wherein each code corresponds to a specific motion semantic;
(5) according to the defined semantic extraction rule, different combination modes of the front and back response sequences are judged, the single motion code or the group motion code is obtained through neighborhood coding calculation, and the motion semantic type predefined corresponding to the minimum semantic unit is converted into the final motion semantic.
Further, the method for calculating the azimuth angle interval of the neighborhood in step 2 and the blade coding mode of the five neighborhood models are respectively as follows:
(2.1) dividing the space window according to the azimuth angle, wherein the space window is divided into four neighborhood directions including an upper neighborhood direction, a lower neighborhood direction, a left neighborhood direction and a right neighborhood direction, and the azimuth angle intervals of the four neighborhoods are respectively as follows:
wherein theta is an azimuth angle interval, and the value of theta is in the clockwise direction taking the north of the interest point as a starting point.
(2.2) constructing a coding mode suitable for calculation on a connected subgraph formed by the self node, the upper node, the right node, the lower node and the left node, and utilizing a blank object group in geometric algebra<eC，eT，eB，eL，eR>Algebraically coding the nodes, eC，eT，eB，eL，eRRespectively corresponding to five nodes including a center, an upper node, a lower node, a left node and a right node.
Further, the calculation method of the time window T and the constraint judgment of the classification and calculation order of the node response sequence in step 3 are as follows:
(3.1) time window implementation formula
Wherein v is the normal walking speed of the human body, s is the detection radius of the sensor, and d is the feasible path distance for installing the sensor.
(3.2) the node response sequences are classified according to response time, and the classification rule can be represented by the following formula:
wherein preNodeList and nextnodeList represent the pre-and post-response sequences, T, respectivelyNeighborhood zoneAnd TCenter of a shipRespectively representing the response time of the neighborhood node and the central node, T is a time window, eFIELDeCenter of a shipOr eCenter of a shipeNeighborhood zoneRepresenting node neighborhood codingThe calculation sequence of the geometric product results is different according to the response condition.
Further, the specific method for monomer motion coding and group motion coding in the minimum semantic unit establishment in step 4 is as follows:
the monomer motion coding is a geometric product result of neighborhood node and center node neighborhood coding, and the contained motion semantics comprise standing, crossing, turning, entering, leaving and returning; the group motion coding is formed by connecting monomer motion expressions with different characteristics through a plus sign to form a multi-vector structure in a geometric algebraic space, and the motion semantics of the multi-vector structure are divided and aggregated.
Further, the semantic conversion rule in step 5 is specifically:
wherein, two columns of preNodeList and nextneList represent the number of response nodes in the front and back response sequences, the front and back sequence relation series represents whether the front and back sequence response nodes are the same, the motion coding and motion semantic column represent monomer or group motion coding obtained by neighborhood coding geometric product calculation and the corresponding semantic classification, eCRepresenting the center node code, eiAnd ejAnd respectively representing neighborhood codes of the nodes, wherein subscripts meet i, j epsilon { R, L, T, B } and i is not equal to j, R, L, T, B respectively represent the right direction, the left direction, the upper direction and the lower direction.
Has the advantages that: the invention utilizes the superior space-time expression capability of geometric algebra to convert the sensor response into the incidence relation between nodes, utilizes the space and time analysis window to constrain the track semantic analysis condition, establishes the minimum semantic analysis unit, completes the one-to-one conversion rule of the response sequence to the motion semantic, and realizes the automatic semantic analysis of the indoor track.
Drawings
FIG. 1 is a flow chart illustrating core steps of an embodiment of the present invention;
FIG. 2 is a schematic diagram of blade encoding of a five-neighborhood model according to an embodiment of the present invention;
FIG. 3 is a diagram illustrating a motion segmentation time window time calculation according to an embodiment of the present invention;
FIG. 4 is a diagram illustrating a minimum semantic unit and motion coding according to an embodiment of the present invention;
fig. 5 is a schematic diagram of a case result of motion trajectory semantic extraction in the embodiment of the present invention.
Detailed Description
The technical solution of the present invention is described in detail below, but the scope of the present invention is not limited to the embodiments.
The embodiment of the invention provides a passive infrared sensor-based indoor behavior track motion semantic analysis method, and the core step flow of the method is shown in figure 1. Firstly, according to an indoor area plan and the distribution coordinate position of an infrared sensor, combining the adjacency relation of the sensor, and taking the sensor as a node, constructing a sensor network diagram without direction and weight; then, with the selected analysis interest point as a search center, expanding the most adjacent nodes in the upper, lower, left and right directions outwards to form a space analysis window, screening the selected interest point and the sensor nodes in the four adjacent domain directions from the space analysis window to form a connected subgraph, establishing a five-neighborhood model containing the center node and the four adjacent domain nodes as a minimum space unit, and performing neighborhood coding on the connected subgraph by using a blade object in geometric algebra to realize the conversion from the connected subgraph to algebraic coding; secondly, selecting an interest time period in a space analysis window, establishing a time analysis window according to the response time of the sensor, taking the response data of the central node as a core, and dividing the data into a front response sequence and a rear response sequence by the rest neighborhood nodes according to the response sequence; then, establishing a minimum semantic unit by using a five-neighborhood model, converting different space states presented by nodes in the five-neighborhood model into monomer motion codes or group motion codes, wherein each motion code corresponds to a specific motion semantic; and finally, according to a defined semantic extraction rule, judging different combination modes of the front response sequence and the rear response sequence, calculating by using neighborhood coding to obtain a single motion code or a group motion code, and converting a predefined motion semantic type corresponding to a minimum semantic unit into a final motion semantic. The specific implementation comprises the following key steps:
the first process is as follows: selecting and preprocessing experimental data to construct a sensor network diagram
Step 1: and (4) selecting required laboratory data, and acquiring basic information data of the data, such as the number of sensors, the sensor ID, the recording start/stop time, the total number of records and the like.
Step 2: according to an indoor plan, a nondirectional weightless network diagram of the sensors in the space is established by referring to the distribution positions and the adjacent relations of the sensors.
And a second process: point of interest based spatial analysis window determination
Step 1: selecting important sensor nodes in a test area as interest points, establishing a spatial analysis window with the interest points as central points, and expanding adjacent nodes in four directions, namely, searching the nearest sensor nodes in the four directions directly connected with the central nodes in a sensor network graph.
Step 2: and screening out sensor nodes including a central node (selected interest point) and four neighborhood (upper, lower, left and right) directions from the space analysis window, wherein the division of the space distribution interval is determined according to the following formula.
The value of theta is clockwise with the north of the interest point as a starting point. And then, a sensor communication subgraph containing all the selected nodes is screened out from the original sensor network.
And step 3: performing neighborhood coding on the sensor connected subgraph, and defining a five-neighborhood model by a central node and four neighborhood directions (upper, right, lower and left)<C，T，R，B，L>The embodiment of the invention uses a blade object group in a geometric algebraic space to construct a coding mode suitable for calculation for a connected subgraph consisting of a central node, an upper node, a right node, a lower node and a left node<eC，eT，eB，eL，eR>And (6) coding is carried out. As particularly shown in fig. 2And coding the centers of the five neighborhood models and the nodes of the upper, lower, left and right sensors.
And thirdly, determining based on the time analysis window of the interest moment:
step 1: according to the acquisition time span of the experimental data, selecting a time period meeting the analysis requirement as an interest moment, and determining a time analysis window of track mining in the interest moment.
Step 2: the time analysis window is used for capturing the response states of the central node and the neighborhood nodes, and continuous capturing of the movement of the central node by the adjacent nodes needs to be ensured, and the continuous capturing is different according to the distribution distance of the sensors. Assuming that v is the normal walking speed of the human body, s is the detection radius of the sensor, d is the feasible path distance of the sensor installation, the time analysis window T can be expressed by the formula
And (4) determining. As shown in fig. 3, if there is a response from a node to a neighboring node within a limited response time, the node is considered to be moving continuously and the two nodes are related, and if there is no response, the motion is considered to be terminated.
And step 3: the nodes are sorted according to the sequence of the node responses in the time analysis window, the calculation sequence of the motion coding is determined according to the response sequence, and the specific constraint rule can be expressed by the following formula
According to the sequence of response time compared with the central node, the response sequence is divided into a pre-response sequence (preNodeList) and a post-response sequence (nextNodeList). The calculation order of the pre-response sequence motion coding is eNeighborhood zoneeCenter of a shipThen the order of computation of the response sequence is eCenter of a shipeNeighborhood zone。
And fourthly, semantic basic unit and motion coding:
step 1: as the human body track motion is a process of transmitting from the node to the neighborhood node, the motion state of the human body track motion presents different space structures in five nodes, and the space state is converted into monomer motion coding or group motion coding which respectively corresponds to different motion semantics.
Step 2: as shown in fig. 4, the single motion coding is a k-blade (k is 1, 2, 3) geometric product coding form obtained by calculating a neighborhood coding geometric product of a neighborhood node and a central node, and motion semantics included in the single motion coding are 1 standing type, 4 crossing types, 8 turning types, 4 entering types, 4 leaving types and 8 returning types. For example, when the motion state is eL-eC-eRWhen it is, it represents a simple traversing movement from left to middle to right, and its single motion code is denoted as eLeCeR. Classifying and specifically coding according to motion semantics, wherein the standing semantics is { O }, and represents no motion; the traversal semantics are { eBeCeT，eTeCeB，eLeCeR，eReCeL}, e.g. eBeCeTIs a through from bottom to top; the semantic of turning is { eTeCeR，eReCeT，eReCeB，eBeCeR，eBeCeL，eLeCeB，eLeCeT，eTeCeLRepresents steering action through an intermediate node, e.g. eTeCeRFrom top to middle to right; entry semantics of { eLeC，eTeC，eReC，eBeC}, e.g. eLeCTo enter the center from the left; the exit semantics are { eCeL，eCeT，eCeR，eCeB}, e.g. eCeLTo move away from the center to the left; return semantics as { eReCeR，eLeCeL，eBeCeB，eTeCeT，eCeLeC，eCeReC，eCeBeC，eCeTeC}, e.g. eReCeRTo go from the right side to the middle and back to the right side.
And step 3: as shown in FIG. 4, group motion coding connects monomer motion expressions of different features with "+" signs, forming a multi-vector structure in geometric algebraic space, which contains 6 separate and 6 aggregate motion semantics. For example, when the motion state is eC-eL，eC-eBWhen it is, it represents the separate movement from the central node to the left node and the lower node at the same time, and its group movement code is marked as eCeL+eCeB. Classifying specific codes according to motion semantics, the separation semantics being { eCeL+eCeB，eCeL+eCeT，eCeL+eCeR，eCeT+eCeB，eCeT+eCeB，eCeB+eCeR}, e.g. eCeL+eCeBTo the left and lower nodes from the central node; aggregate semantics as { eBeC+eLeC，eTeC+eLeC，eLeC+eReC，eTeC+eReC，eBeC+eTeC，eReC+eBeCIn which eLeC+eBeCFrom the left node and the lower node toward the central node.
Fifth, semantic extraction rule and movement semantic conversion
Step 1: according to the combination relation of the front response sequence and the rear response sequence, semantic extraction rules are defined, each semantic type corresponds to an information rule, one-to-one semantic information conversion is realized, and the specific rules are shown in the following table:
wherein, two columns of preNodeList and nextneList represent the number of response nodes in the front and back response sequences, the front and back sequence relation series represents whether the front and back sequence response nodes are the same, the motion coding and motion semantic column represent monomer or group motion coding obtained by neighborhood coding geometric product calculation and the corresponding semantic classification, eCRepresenting the center node code, eiAnd ejRespectively representing the neighborhood coding of the node, the subscript satisfies i, j e to { R, L, T, B } and i is not equal to j, R, L, T, B respectively represents the right, left, upper and lower four directions
Step 2: and solving the motion code according to a semantic extraction rule by using a calculation mode of a response node neighborhood coding geometric product according to a calculation sequence with response time as constraint and the sequence of the divided front and rear response sequences in the time analysis window, and converting the semantic type predefined corresponding to the minimum semantic unit into final semantic.
And step 3: and repeating the semantic analysis process to generate semantic information of different position points in different time, and performing related statistics and visual analysis.
The following is an example of the analysis of sensor data in an electrical research laboratory, which illustrates the implementation of the method according to an embodiment of the present invention.
The first process is as follows: selecting and preprocessing experimental data, constructing sensor connectivity graph and coding
Step 1: an experimental target area is selected, the data of the embodiment is a certain electrical research laboratory, the total number of the electrical research laboratory is 213, the data is recorded for 1 year, and the time distribution is from 2006, 3, 21 days to 2007, 3, 24 days. The recorded data are divided into four parts, namely sensor id, target detection starting time, target detection ending time and detection effectiveness, and the data have 30239000 records.
Step 2: and establishing a undirected weightless network diagram of the sensors in the space according to the plan view and the distribution positions of the sensors in the laboratory.
And a second process: point of interest based spatial analysis window determination
Step 1: because the experimental data is detection data of a certain laboratory, important sensor nodes in the area are selected as interest points (such as important offices, elevator entrances, meeting rooms and the like), a spatial analysis window with the interest points as central points is established, and the most adjacent nodes in the upper, lower, left and right directions are outwards expanded according to a sensor network diagram.
Step 2: screening out sensor nodes including a central node (selected interest point) and four neighborhood (upper, lower, left and right) directions thereof from a space analysis window to serve as sensor subgraphs, and using blank object groups in a geometric algebra space for the central node and the four neighborhood direction nodes<eC，eT，eB，eL，eR>And (6) coding is carried out.
And thirdly, determining based on the time analysis window of the interest moment:
step 1: the embodiment data comprises a data record with the total length of one year, a time period meeting the analysis requirement is selected as an interest moment (such as morning work, lunch, evening work, work day, weekend and the like), and a time analysis window of track mining is determined in the interest moment.
Step 2: the time analysis window is used for capturing the response states of the central node and the neighborhood nodes, in the embodiment, it is assumed that the normal walking speed of a person is 1.2 m/s, each sensor is tightly distributed, no gap or small gap exists between the detection ranges of the sensors, and the range of each sensor is 2 m, so that the induction time of the person passing through one sensor and entering the adjacent sensor is 2-4 s, and the time window can be set to be 3 s by taking the intermediate value. If the adjacent node responds within 3 seconds from a certain node, the node is considered to move continuously and the two nodes are related, and if no response exists, the movement is considered to be terminated.
And step 3: the nodes are sorted according to the sequence of the node responses in the time window, and the response sequence is divided into a front response sequence (preNodeList) and a rear response sequence (nextNodeList).
And fourthly, semantic basic unit and motion coding:
using five neighborhood models as basic space units, coding the propagation process caused by human motion between nodes and neighborhood nodes, converting the propagation process into monomer motion coding or group motion coding according to different space states,
fifth, semantic extraction rule and movement semantic conversion
Step 1: and defining semantic extraction rules according to the combination relation of the front response sequence and the rear response sequence, wherein each semantic type corresponds to an information rule, and one-to-one semantic information conversion is realized.
Step 2: and solving the motion code according to the semantic extraction rule by using a calculation mode of the neighborhood coding geometric product of the response node according to the calculation sequence of the target front and back response sequences under the constraint of response time, and converting the semantic type predefined corresponding to the minimum semantic unit into final semantics.
And step 3: and repeating the semantic analysis process, extracting track semantic information in different time, and carrying out relevant statistics. Taking the 'entry' semantic as an example, the 'entry' semantic of Monday to Monday in a selected time period is extracted and counted, visual expression is carried out according to the occurrence frequency and spatial distribution of the 'entry' semantic, and the darker the color is, the higher the occurrence frequency of the 'entry' behavior at the position is. As shown in fig. 5, three subgraphs, a, b and c, respectively show the semantic frequencies from monday to wednesday.
Claims (4)
1. A semantic analysis method for indoor behavior track motion based on a passive infrared sensor is characterized by comprising the following steps:
(1) according to the indoor area plan and the distribution coordinate position of the infrared sensor, combining the adjacency relation of the sensors, and taking the sensors as nodes, constructing a sensor network diagram without direction and weight;
(2) the method comprises the steps of taking a selected analysis interest point as a search center, outwards expanding the most adjacent nodes in four directions of up, down, left and right to form a space analysis window, screening the selected interest point and sensor nodes in four adjacent domain directions from the space analysis window to form a connected subgraph, establishing a five-adjacent-domain model containing the center node and the four adjacent-domain-direction nodes as a minimum space unit, and performing adjacent-domain coding on the connected subgraph by using a blank object in geometric algebra to realize the conversion from the connected subgraph to algebraic coding; the space analysis window is divided into four neighborhood directions including an upper neighborhood direction, a lower neighborhood direction, a left neighborhood direction and a right neighborhood direction, and the azimuth angle intervals of the four neighborhoods are respectively as follows:
wherein theta is an azimuth angle interval, and the value of theta is in the clockwise direction taking the north of the interest point as a starting point;
(3) in the space analysis window, selecting an interest time period, establishing a time analysis window according to the response time of the sensor, taking the response data of the central node as a core, and dividing the data into a front response sequence and a rear response sequence by the rest neighborhood nodes according to the response sequence; the time analysis window T is determined according to the following formula:
wherein v is the normal walking speed of the human body, s is the detection radius of the sensor, and d is the feasible path distance for installing the sensor; if the response time of the adjacent node is limited from a certain node, the node is considered to move continuously and the two nodes are related, and if no response is generated, the movement is considered to be terminated;
the classification rule of the pre-response sequence and the post-response sequence is represented by the following formula:
wherein prenodeList and nextnodeList represent the pre-and post-response sequences, T, respectivelyNeighborhood zoneAnd TCenter of a shipRespectively representing the response time of the neighborhood node and the center node, T is an analysis time window, eNeighborhood zoneeCenter of a shipOr eCenter of a shipeNeighborhood zoneThe geometric product results of the representative node neighborhood codes are calculated in different orders according to response conditions;
(4) establishing a minimum semantic unit by the five-neighborhood model in the step (2), converting different space states presented by nodes in the five-neighborhood model into monomer motion codes or group motion codes, wherein each code corresponds to a specific motion semantic;
(5) according to the defined semantic extraction rule, different combination modes of the front and back response sequences are judged, the single motion code or the group motion code is obtained through neighborhood coding calculation, and the motion semantic type predefined corresponding to the minimum semantic unit is converted into the final motion semantic.
2. The method as claimed in claim 1, wherein the semantic analysis of indoor behavior trajectory motion based on passive infrared sensor is performed in step (2) by using blade object group in geometric algebra<eC,eT,eB,eL,eR>Respectively algebraically coding the central node and the upper, lower, left and right nodes, eC,eT,eB,eL,eRRespectively corresponding to five nodes including a center, an upper node, a lower node, a left node and a right node.
3. The indoor behavior track motion semantic analysis method based on the passive infrared sensor, according to claim 1, characterized in that in the step (4), the monomer motion codes are geometric product results of neighborhood node and center node neighborhood codes, and the included motion semantics are six types of standing, crossing, turning, entering, leaving and returning; the group motion coding is formed by connecting monomer motion expressions with different characteristics through a plus sign to form a multi-vector structure in a geometric algebraic space, and the motion semantics of the multi-vector structure are divided and aggregated.
4. The method for semantic analysis of indoor behavior trajectory motion based on a passive infrared sensor as claimed in claim 1, wherein the semantic extraction rule in step (5) specifically comprises:
wherein the content of the first and second substances,two columns of preNodeList and nextnodeList represent the number of response nodes in the front and back response sequences, the front and back sequence relation series represents whether the front and back sequence response nodes are the same, the motion coding and motion semantic column represent monomer or group motion coding obtained by calculating the geometric product of neighborhood coding and the corresponding semantic classification, eCRepresenting the center node code, eiAnd ejAnd respectively representing neighborhood codes of the nodes, wherein subscripts meet i, j epsilon { R, L, T, B } and i is not equal to j, R, L, T, B respectively represent the right direction, the left direction, the upper direction and the lower 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 CN107657215A (en) | 2018-02-02 |
| CN107657215B true 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) |
Families Citing this family (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 |
| CN109668094B (en) * | 2019-01-29 | 2021-09-07 | 深圳中电桑飞智能照明科技有限公司 | Intelligent street lamp system |
| 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 |
| 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 |
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 |
|---|
| 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;冯琳耀;《中国优秀硕士学位论文全文数据库》;20160215;第14-16页，33-40页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107657215A (en) | 2018-02-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107657215B (en) | Indoor behavior track motion semantic analysis method based on passive infrared sensor | |
| Teknomo | Microscopic pedestrian flow characteristics: Development of an image processing data collection and simulation model | |
| Sundhari et al. | IoT assisted Hierarchical Computation Strategic Making (HCSM) and Dynamic Stochastic Optimization Technique (DSOT) for energy optimization in wireless sensor networks for smart city monitoring | |
| Xie et al. | Video crowd detection and abnormal behavior model detection based on machine learning method | |
| CN108898520A (en) | Safety of student monitoring method and system based on track data | |
| Li et al. | A top–bottom clustering algorithm based on crowd trajectories for small group classification | |
| CN111783295B (en) | Dynamic identification and prediction evaluation method and system for urban community specific human behavior chain | |
| CN114372503A (en) | Cluster vehicle motion trail prediction method | |
| Połap | Neuro-heuristic analysis of surveillance video in a centralized IoT system | |
| Weng et al. | Review of analyses on crowd-gathering risk and its evaluation methods | |
| Brščić et al. | Changes in usage of an indoor public space: Analysis of one year of person tracking | |
| Alolaiwy et al. | Crowd path trajectory prediction using least action principle | |
| Chauhan et al. | Study of moving object detection and tracking for video surveillance | |
| CN115688952A (en) | Big data-based campus congestion area prediction method | |
| Brax et al. | Finding behavioural anomalies in public areas using video surveillance data | |
| Jiang et al. | Behavior pattern mining based on spatiotemporal trajectory multidimensional information fusion | |
| Khan et al. | Integrated analysis and synthesis of pedestrian dynamics: First results in a real world case study | |
| Yu et al. | Network security monitoring method based on deep learning | |
| CN112183824B (en) | Online and offline correlated urban passenger flow prediction method | |
| Hao et al. | Group abnormal behaviour detection algorithm based on global optical flow | |
| Yao et al. | Using machine learning approach to construct the people flow tracking system for smart cities | |
| Wan et al. | Dynamic scene modelling and anomaly detection based on trajectory analysis | |
| Fujita et al. | Human crowd detection for physical sensing assisted geo-social multimedia mining | |
| Hao et al. | A Detection Method of Abnormal Event in Crowds Based on Image Entropy | |
| CN113220675A (en) | College student behavior analysis system and method based on WiFi positioning data |
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 |