CN109344729A - A kind of method of personnel's movement in identification road - Google Patents
A kind of method of personnel's movement in identification road Download PDFInfo
- Publication number
- CN109344729A CN109344729A CN201811044619.7A CN201811044619A CN109344729A CN 109344729 A CN109344729 A CN 109344729A CN 201811044619 A CN201811044619 A CN 201811044619A CN 109344729 A CN109344729 A CN 109344729A
- Authority
- CN
- China
- Prior art keywords
- window
- data
- point
- sliding
- points
- 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 44
- 238000000034 method Methods 0.000 title claims abstract description 41
- 230000002159 abnormal effect Effects 0.000 claims abstract description 37
- 230000002457 bidirectional effect Effects 0.000 claims description 39
- 230000003068 static effect Effects 0.000 claims description 14
- 238000004140 cleaning Methods 0.000 claims description 6
- 230000000284 resting effect Effects 0.000 claims description 2
- 238000005406 washing Methods 0.000 claims 1
- 238000012163 sequencing technique Methods 0.000 abstract 1
- 238000004364 calculation method Methods 0.000 description 4
- 238000013499 data model Methods 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000009087 cell motility Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000004576 sand Substances 0.000 description 1
- 239000000725 suspension Substances 0.000 description 1
- 230000001360 synchronised effect Effects 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
- G06V40/23—Recognition of whole body movements, e.g. for sport training
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/30—Noise filtering
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Health & Medical Sciences (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a kind of methods of personnel's movement in identification road, the road grid MR data of each user are converged in respective set based on MR location data and road raster data, all data are according to time-sequencing in each subscriber data set, single user data sequence after sequence forms the track in each user time sequence, pass through the two-way sliding window Trajectory Arithmetic of multiple alien frequencies, rest point and abnormal point in user trajectory are excluded, correctly identifies that on track motor point in any time sequence;The method of the present invention not only identifies that granularity is fine, but also higher to the recognition accuracy of noise, and has very strong fault-tolerant ability to the offset of noise and positioning.
Description
Technical Field
The invention relates to the technical field of wireless networks, in particular to a method for identifying the movement of people on a road.
Background
The road motion user identification is used as a key link for index evaluation of a road scene of a wireless network by an operator, and the network state of the road motion user is used as basic data for the index evaluation of the road scene. Therefore, the accuracy of the identification of the moving user is decisive for the accuracy of the index evaluation. The accuracy of the identification of the moving user includes two aspects of the amount of the identified sample and the identification accuracy, so that the user moving in any time period is identified as much as possible, and the tracks of all the tracks of the user are accurately identified, and the tracks are really in the moving state. The existing road motion user identification is mainly based on the road cell to identify the motion user. The identification of the moving user based on the road cell comprises the following steps: acquiring users under a road cell; the moving user is determined according to a certain rule, such as a predetermined number of cells passing within a predetermined time.
The existing method for identifying the moving users based on the road cells has the following defects:
the accuracy of the identified users has larger errors, the granularity of screening the road users according to the road cells is thicker, and roadside indoor users, corridor users or road static users can be calculated;
the accuracy of the identified moving user is poor, and if the cell is in circular cut or normal signal flutters, the static user is wrongly judged as a moving user; and eliminating abnormal signals in a series of continuous user movement tracks.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a method for identifying the movement of people on a road, which identifies the movement point, the static point and the abnormal point of a user on the road based on MR positioning data and road grid data.
The invention adopts the following technical scheme:
a method of identifying movement of a person in a roadway, comprising: moving, stationary, and outlier points identifying users on the road are identified based on the MR positioning data and the road raster data.
Preferably, identifying points of motion, resting points and outliers that identify users on the road based on the MR positioning data and the road grid data comprises:
extracting specified field data from the MR positioning data; the specified field data includes a terminal ID, longitude, latitude, time, and grid ID; the terminal ID uniquely identifies a user;
cleaning the extracted MR positioning data, and removing invalid data and repeated data in the MR positioning data;
combining the cleaned MR data with the road grid data to screen out the road grid MR data;
generating a trajectory over each user time series based on the road grid MR data;
based on the trajectory, a moving point, a stationary point, and an abnormal point on the user trajectory in any time series are identified.
Preferably, the cleaning of the extracted MR positioning data and the elimination of invalid data and duplicate data in the MR positioning data includes:
judging whether a field in the specified field data extracted from the MR positioning data is a null value, if so, removing the whole data;
judging whether the specified field data extracted from the MR positioning data has repeated data or not, and if so, rejecting the whole repeated data.
Preferably, generating a trajectory on each user time series based on the road grid MR data comprises:
and based on the road grid MR data, the data of each user are gathered in respective sets and are sorted according to time, and the sorted single-user data sequence forms a track on the time sequence of each user.
Preferably, based on the trajectory, identifying a moving point, a stationary point and an abnormal point on the user trajectory in any time series includes:
based on the track, identifying a moving point, a static point and an abnormal point on a user track in any time sequence by a multi-pilot-frequency bidirectional sliding window track algorithm;
the multiple pilot frequency bidirectional sliding window track algorithm is provided with two layers of sliding windows, wherein the number of points of the first layer of sliding windows is used as a sliding unit, and the time period of the second layer of sliding windows is used as a sliding unit.
Preferably, the method for identifying the moving point and the stationary point on the user trajectory in any time sequence by the multi-pilot-frequency bidirectional sliding window trajectory algorithm includes:
setting a first layer sliding window of the multi-pilot-frequency bidirectional sliding window track algorithm to comprise two points; wherein one of the two points is a target point;
acquiring the time difference and the distance between the two points; and calculating the speed between the two points based on the time difference and the distance, and judging whether the target point in the window is a moving point or a static point based on the speed between the two points.
Preferably, the method for identifying the moving point and the abnormal point on the user track in any time sequence by the multi-pilot-frequency bidirectional sliding window track algorithm comprises the following steps:
sliding a sliding unit forwards in a second layer sliding window of the multi-pilot-frequency bidirectional sliding window track algorithm, and calculating the average speed and the average distance of points in the forward window and/or a plurality of grids with the highest occurrence frequency in the window by a mean value clustering algorithm; wherein a target point is included within the forward window;
sliding a sliding unit backwards in a sliding window of a second layer of the multi-pilot-frequency bidirectional sliding window track algorithm, and calculating the average speed and the average distance of points in a backward window and/or a plurality of grids with the highest occurrence frequency in the window through a multi-dimensional density clustering algorithm; wherein the rearward window also includes a target point therein;
and judging whether the target point in the window is a moving point or an abnormal point based on the average speed and the average distance of the points in the forward window and/or the grids with the highest occurrence times in the window and the average speed and the average distance of the points in the backward window and/or the grids with the highest occurrence times in the window.
Preferably, the method for identifying the moving point and the abnormal point on the user track in any time sequence by the multi-pilot-frequency bidirectional sliding window track algorithm comprises the following steps:
sliding a sliding unit forwards in a second layer of sliding window of the multi-pilot-frequency bidirectional sliding window track algorithm, and calculating the average speed and the average distance of points in the forward window and/or a plurality of grids with the highest occurrence frequency in the window through a multi-dimensional density clustering algorithm; wherein a target point is included within the forward window;
sliding a sliding unit backwards in a sliding window of a second layer of the multi-pilot-frequency bidirectional sliding window track algorithm, and calculating the average speed and the average distance of points in a backward window and/or a plurality of grids with the highest occurrence frequency in the window by a mean value clustering algorithm; wherein the rearward window also includes a target point therein;
and judging whether the target point in the window is a moving point or an abnormal point based on the average speed and the average distance of the points in the forward window and/or the grids with the highest occurrence times in the window and the average speed and the average distance of the points in the backward window and/or the grids with the highest occurrence times in the window.
Preferably, the method for identifying the moving point and the abnormal point on the user track in any time sequence by the multi-pilot-frequency bidirectional sliding window track algorithm comprises the following steps:
sliding a sliding unit forwards in a second layer sliding window of the multi-pilot-frequency bidirectional sliding window track algorithm, and calculating the average speed and the average distance of points in the forward window and/or a plurality of grids with the highest occurrence frequency in the window by a mean value clustering algorithm; wherein a target point is included within the forward window;
sliding a sliding unit backwards in a sliding window of a second layer of the multi-pilot-frequency bidirectional sliding window track algorithm, and calculating the average speed and the average distance of points in a backward window and/or a plurality of grids with the highest occurrence frequency in the window by a mean value clustering algorithm; wherein the rearward window also includes a target point therein;
and judging whether the target point in the window is a moving point or an abnormal point based on the average speed and the average distance of the points in the forward window and/or the grids with the highest occurrence times in the window and the average speed and the average distance of the points in the backward window and/or the grids with the highest occurrence times in the window.
Preferably, the method for identifying the moving point and the abnormal point on the user track in any time sequence by the multi-pilot-frequency bidirectional sliding window track algorithm comprises the following steps:
sliding a sliding unit forwards in a second layer of sliding window of the multi-pilot-frequency bidirectional sliding window track algorithm, and calculating the average speed and the average distance of points in the forward window and/or a plurality of grids with the highest occurrence frequency in the window through a multi-dimensional density clustering algorithm; wherein a target point is included within the forward window;
sliding a sliding unit backwards in a sliding window of a second layer of the multi-pilot-frequency bidirectional sliding window track algorithm, and calculating the average speed and the average distance of points in a backward window and/or a plurality of grids with the highest occurrence frequency in the window through a multi-dimensional density clustering algorithm; wherein the rearward window also includes a target point therein;
and judging whether the target point in the window is a moving point or an abnormal point based on the average speed and the average distance of the points in the forward window and/or the grids with the highest occurrence times in the window and the average speed and the average distance of the points in the backward window and/or the grids with the highest occurrence times in the window.
Compared with the prior art, the invention has the following beneficial effects:
(1) the method for identifying the movement of the people in the road can distinguish the road users from the non-road users, and has high identification accuracy and fine distinguishing granularity; in the prior art, based on cell motion user identification, the differentiated granularity is a data set at a cell level, and each track point can be differentiated by the method;
(2) the invention relates to a method for identifying the movement of people on a road, which is characterized in that the data of each user are gathered in respective sets, all the data in each user data set are sorted according to time, a track on each user time sequence is formed by a sorted single-user data sequence, a stationary point and an abnormal point on the user track are eliminated by combining a multi-pilot-frequency bidirectional sliding window track algorithm, a mean value clustering algorithm and a multi-dimensional density clustering algorithm, and a moving point on the track in any time sequence is correctly identified; the method not only has fine identification granularity, but also has higher identification accuracy on the noise point, and has strong fault-tolerant capability on the noise point and the offset of positioning;
(3) according to the method for identifying the movement of the people on the road, the extracted MR positioning data is cleaned, invalid data and repeated data in the MR positioning data are removed, and the robustness of the method and the accuracy of the identification result are guaranteed.
The above description is only an overview of the technical solutions of the present invention, and the embodiments of the present invention are described below in order to make the technical means of the present invention more clearly understood and to make the description of the technical means more comprehensible.
The above and other objects, advantages and features of the present invention will become more apparent to those skilled in the art from the following detailed description of specific embodiments thereof, taken in conjunction with the accompanying drawings.
Drawings
FIG. 1 is a flow chart of a method of identifying movement of a person in a roadway according to the present invention;
FIG. 2 is a detailed flow chart of the method of identifying movement of a person in a roadway of the present invention;
FIG. 3 is a schematic diagram of a multiple pilot frequency bi-directional sliding window trajectory algorithm of the present invention;
FIG. 4 is a flow chart of the present invention for identifying two layers of sliding windows using a multiple pilot frequency bi-directional sliding window trajectory algorithm;
fig. 5 is a detailed flowchart of the present invention for identifying a target point.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
Referring to fig. 1, the present invention relates to a method for recognizing the movement of a person on a road, comprising:
in step 101, specified field data is extracted from the MR positioning data.
Step 102, cleaning the extracted MR positioning data, and removing invalid data and repeated data in the MR positioning data;
step 103, combining the cleaned MR data with road grid data to screen out road grid MR data;
step 104, generating a track on each user time sequence based on the road grid MR data;
and 105, identifying a moving point, a static point and an abnormal point on the user track in any time sequence based on the track.
Further, as shown in fig. 2, the method is based on a big data platform, screening of the road grid MR data is performed through MR positioning data and the road grid data by using a big data technology, then analysis of the travel distance and the travel speed of each terminal user is performed on the road grid MR data by using a distributed memory computing technology, identification of the motion track point is performed by using a multiple different-frequency bidirectional sliding window track algorithm and a mean value clustering algorithm, and noise point identification is performed by using a user-defined multidimensional density clustering algorithm. The problems of positioning precision deviation and data loss are solved in the identification process, and the moving point and the noise point are accurately identified to the maximum extent.
Specifically, the hadoop big data platform is used for storing original data, the platform can automatically divide the original data according to the size of a set data block when the original data is loaded according to a set data block value, for example, the original data is 1G, the platform sets the data block to be 128M, at the moment, the data can be automatically divided into 8 blocks on the platform, the distributed operation platform uses spark, when the spark loads the original data from the hadoop platform, the spark can default to use the number of the blocks of the data on the hadoop platform as data slicing data during calculation, then algorithm calculation is performed on each piece of data, and finally, the identification result of each record is output.
In this embodiment, the data extracted from the MR positioning data (MR raw data) in step 101 includes a terminal ID, a longitude, a latitude, a time, and a grid ID; wherein the terminal ID uniquely identifies a user; the data model for each record is as follows:
“imsi,lon,lat,time,grid_id”
wherein, imsi is used to represent a terminal ID, lon represents longitude, lat represents latitude, time represents time, grid _ ID represents grid ID, and specific data cases are as follows:
“460xxxx,116.54021,39.653052,2018-06-01 20:00:24,50XBJ52805050”
according to the data model, the output format of the recognition result of each record is as follows:
the output format is:
“imsi,lon,lat,time,grid_id,flag”
the method comprises the steps of obtaining a static mark, a static mark and a motion point, wherein the flag represents a dynamic and static mark, the flag is 0 and represents a static point, the flag is 1 and represents a motion point, and the flag is 2 and represents an abnormal point.
The following data cases represent motion points.
“460xxx,116.54021,39.653052,2018-06-01 20:00:24,50XBJ52805050,1”
And finally, storing the identified result set in a distributed file system of the hadoop platform in a text form, and establishing a hive intermediate table to provide query.
In this embodiment, step 102, the step of cleaning the extracted MR positioning data and eliminating invalid data and duplicate data in the MR positioning data includes:
the terminal ID field, the longitude field, the latitude field, the time field and the grid ID field are respectively checked, the terminal ID field is null or the part of data which cannot be identified must be removed, otherwise, serious data skew is caused during large data processing, and the program is crashed. Meanwhile, data with empty longitude field, latitude field, time field and grid ID field and repeated data in MR positioning data also need synchronous cleaning, and robustness and result accuracy of the program are guaranteed. The number anomaly data extracted from the MR positioning data is as follows:
", 116.5402, 9.6530, 2018-06-0120: 00:24, 50XBJ 52805050" or
"460 xxxx", "2018-06-0120: 00:24, 50XBJ 5280050",
the above data needs to be culled.
Because the original MR positioning data or the cleaned MR positioning data is a massive distributed data set, in order to reduce the data set and improve the calculation performance, and in order to improve the accuracy of the road motion point identification of a user, the cleaned MR positioning data and the road grid data are combined, building data and cell data are excluded, and road grid MR data are screened out.
In this embodiment, step 103 is to combine the cleaned MR data with the road grid data to screen out the road grid MR data, as follows.
Let the data model of the extracted road grid data be "grid _ id, city _ load", where: grid _ ID represents a road grid ID; the city _ load represents whether an urban road exists or not, the city _ load is an urban road, and the city _ load is a false road and does not represent the urban road; the cleaned MR positioning data and the road grid data are associated, and the road grid MR data is obtained by taking the data of the city _ load ═ true.
For example, the two pieces of MR data after extraction are as follows:
“460xxxx1,116.5402,39.6530,2018-06-01 20:00:00,50XBJ52805050”
“460xxxx2,116.5432,39.6531,2018-06-01 20:00:00,50XBJ52805151”
the two extracted road grid data are as follows:
50XBJ52805050,true;
50XBJ52805151,false;
combining the cleaned MR positioning data and the road grid data to generate the following two data:
“460xxx1,116.5402,39.6530,2018-06-01 20:00:00,50XBJ52805050,true”
“460xxx2,116.5432,39.6531,2018-06-01 20:00:00,50XBJ52805151,false”
the final screened road grid MR data is as follows:
“460xxx1,116.5402,39.6530,2018-06-01 20:00:00,50XBJ52805050,true”
after the road grid MR data are screened out, in order to enable the data to be suitable for an identification algorithm (such as a multiple pilot frequency two-way sliding window track algorithm), the data are formatted and standardized to a certain extent, time data are subjected to numerical conversion, the data are gathered according to users and then are arranged according to a time sequence, and the data are organized according to a model required by the algorithm.
For example, the road grid MR data of a selected user is as follows:
“460xxx1,116.5402,39.6530,2018-06-01 20:00:00,50XBJ52805050,true”
“460xxx1,116.5412,39.6531,2018-06-01 20:00:00,50XBJ52805049,true”
“460xxx1,116.5433,39.6536,2018-06-01 20:00:00,50XBJ52805051,true”
the normalized chronological data is as follows:
“460xxx1,116.5412,39.6531,1532262140000,50XBJ52805049”
“460xxx1,116.5402,39.6530,1532262144000,50XBJ52805050”
“460xxx1,116.5433,39.6536,1532262148000,50XBJ52805051”
the processing of steps 101 to 103 is to preprocess the original MR positioning data and road grid data, and the following steps will process the preprocessed data to identify moving points, stationary points and outliers (noise) on each user time series trajectory.
In this embodiment, the trajectory, the stationary point, the noise point, and the moving point are defined as follows.
(1) Trajectory definition
There is a connected set of spatio-temporal points in chronological order, the expression is as follows:
wherein,representing a point on the chronological trajectory that includes longitude, latitude and time attributes, i.e., the single piece of data after sorting. x is the number ofiIndicates longitude information, yiIndicating latitude information, tiRepresenting time, I represents a set of points that make up a trajectory;
(2) the rest point is defined as:
stationary point set, the expression is as follows:
wherein,denotes the kth point, (x)k,yk) Represents the k < th >Longitude and latitude of a point, CsIs the spatial range commonly defined by the trajectories making up S;mat the end of the dwell trajectory, t1Is the first point time, tthreshIs a predetermined time threshold; dmAt the end of the dwell trajectory, d1Is the initial position, dthreshIs a predetermined distance. The definition indicates that any point in S must fall on CsAnd the time difference of the first and the last points of S must be larger than tthreshOr the difference in distance between the head and the end points must be larger than dthresh。
(3) Noise (outlier) definition:
due to the problems of MR longitude and latitude positioning accuracy or data receiving quality and the like, the positioned longitude and latitude can drift out of a normal range D, or the speed exceeds V, and the expression is as follows:
wherein,the point of the k-th point is represented,representing the nth point, the noisy data cannot be simply judged by the speed and distance between two points, but the complex logic therein is represented by two functions respectively.
(4) The mobile point definition:
any trace point that is not stationary and not noisy belongs to a moving point. The expression is as follows:
wherein S represents a stationary point set, and DF represents a noise point set.
In this embodiment, the preprocessed road grid MR data of each user is gathered in the respective set and sorted according to time, and the sorted single-user data sequence forms a track on the time sequence of each user. The moving, stationary and abnormal points on the trajectory in any time series will be identified by the multiple inter-frequency bi-directional sliding window trajectory algorithm as follows.
Referring to fig. 3, in the multiple different-frequency bidirectional sliding window trajectory algorithm described in this embodiment, two layers of sliding windows are provided, and step length measurement units of the two layers of sliding windows are different, and the measurement of the first layer of sliding window is performed by taking the number of points as 1 sliding unit, in this example, by taking 1 point as a measurement unit; the second layer of sliding windows is measured by taking a time period T as 1 sliding unit, and the time period T is 10 minutes in this embodiment.
Specifically, referring to fig. 4 and 5, in the embodiment, in the first layer of sliding windows of the multiple different-frequency bidirectional sliding window trajectory algorithm, each sliding window includes two points, time subtraction between the two points is used as a time difference Δ T, longitude and latitude of the two points are converted into a distance D between the two points through a formula, and then a speed V between the two points is calculated to be D/Δ T, and based on the time Δ T and the speed V, it can be determined whether the target point is a moving point or a stationary point. If the target point is stationary, outputting a stationary flag 0; if the point is non-stationary, it is further determined whether the target point is a moving point or a noise point (outlier) in the second layer sliding window.
In this embodiment, the second layer of sliding windows are bidirectional sliding windows, and can be sliding windows forward or backward, and the specific implementation of determining whether the target point is a moving point or an abnormal point in the second layer of sliding windows includes:
a. in a second layer sliding window of the multi-pilot-frequency bidirectional sliding window track algorithm, a sliding unit is slid forwards, and the average speed V of the point in the forward window is calculated through a mean value clustering algorithm1Average distance S1And N with the highest occurrence number in the window1A grid; wherein, the forward window is a window which slides a sliding unit forwards; the forward window includes a mesh thereinMarking points; the average distance S1Equal to the average speed V1The product with time period T;
b. in a second layer sliding window of the multi-pilot-frequency bidirectional sliding window track algorithm, a sliding unit is slid backwards, and the average speed V of the points in the backward window is calculated through a multi-dimensional density clustering algorithm2Average distance S2And N with the highest occurrence number in the window2A grid; wherein, the backward window is a window which slides a sliding unit backward; a target point is also included within the rearward window; i.e. both the forward and backward windows comprise the target point; the average distance S2Equal to the average speed V2The product with time period T;
c. average velocity V based on points in the forward window1Average distance S1And N with the highest occurrence number in the window1A grid, and an average velocity V based on points in the backward window2Average distance S2And N with the highest occurrence number in the window2And the grid judges whether the target point in the window is a motion point or an abnormal point.
Specifically, the average velocity V can be judged1And an average velocity V2Whether the speed V between the two points calculated in the first layer sliding window is within a reasonable proportion to D/delta T or not, if not, the speed is an abnormal point, and an abnormal identifier is output; the average distance S can also be judged1And the average distance S2Whether the distance D between the two points calculated in the first layer sliding window is within a reasonable proportion is judged, if not, the distance D is an abnormal point, and an abnormal identifier is output; the frequency of all points in the front window and the rear window (in the forward window and the backward window) in each grid can be counted, and the N with the highest frequency of occurrence in the forward window is taken1N with highest occurrence in each grid and backward window2A grid, compare N1Whether the number of overlapping of the grids is less than N12, compare N2Whether the number of overlapping of the grids is less than N2If not, the data are abnormal points, and abnormal marks are output; otherwise, the target point is a motion point, and the motion is outputAnd (5) identifying.
It should be noted that the above is only an optional judgment example, and in specific implementation, the average speed, the average distance, and the number of grid overlaps may be subjected to combined judgment or single judgment according to actual conditions; the average speed V may be determined1And an average velocity V2The judgment after the combination is performed is similar when the average distance and the grid overlapping number are judged, and the embodiment of the invention is not particularly limited.
Further, in the step a, the average speed V of the point in the forward window is calculated1Average distance S1And N with the highest occurrence number in the window1The algorithm of each grid can also be a multidimensional density clustering algorithm; calculating the average speed V of the points in the backward window in the step b2Average distance S2And N with the highest occurrence number in the window2The algorithm of each grid can also be a mean value clustering algorithm; a mean value clustering algorithm can be used in the step a and the step b; alternatively, step a and step b both use a mean clustering algorithm. Of course, other clustering algorithms may be used, as long as the average speed of the points in the window can be calculated, and the present invention is not limited in particular.
The flow of the mean clustering algorithm is as follows:
a1, initializing. The speed of the input point is used as an object set X, a specified clustering class number N is input, wherein N is generally equal to 2, and N objects are randomly selected from X to serve as initial clustering centers. Iteration stop conditions are set, such as maximum loop times or cluster center convergence error margins.
a2, iteration is carried out. Assigning the data objects to the closest cluster centers according to a similarity criterion, thereby forming a class; and initializing a membership matrix.
a3, updating the clustering center. And taking the average vector of each class as a new clustering center, and reallocating the data objects.
Repeatedly executing the step a2 and the stepStep a3 until a suspension condition is satisfied. Then, a set with a large point set is selected as a normal signal set, and the average speed V of the set is calculated1Determining whether the speed is within a reasonable range, for the V1The rationality judgment needs to fully consider the characteristics of different motion scenes, mainly the walking speed, the riding speed of an electric vehicle or a bicycle, the high-speed motor speed of a private car, a bus or a subway and the like.
The multi-dimensional density clustering algorithm is realized by drawing a circle by taking each data point as a circle center and eps (field radius) as a radius, and then calculating how many points are in the circle, wherein the number is the density value of the point. And then selecting a density threshold MinPts, wherein the central points with the number of points in the circle less than MinPts are low-density points, and the central points with the number greater than or equal to the MinPts are high-density points. If there is a high density of dots within the circle of another high density of dots, the two dots are connected so that the dots can be continuously connected in series. Then, if there is a point of low density also within the circle of points of high density, it is also connected to the nearest point of high density, called the boundary point. Thus all points that can be connected together are in a cluster, and a low density point that is not within the circle of any high density point is an outlier, which if it is the current decision point is an outlier. If the current judging point is in the high-density cluster, calculating the average value V of the points in the cluster2This value is taken as the average speed V within the calculation window period2。
The above description is only an embodiment of the present invention, but the design concept of the present invention is not limited thereto, and any insubstantial modifications made by using the design concept should fall within the scope of infringing the present invention.
Claims (10)
1. A method of identifying movement of a person in a roadway, comprising: moving, stationary, and outlier points identifying users on the road are identified based on the MR positioning data and the road raster data.
2. Method of identifying movements of persons on roads according to claim 1, characterized in that identifying points of movement, resting points and anomalies that identify users on roads based on MR positioning data and road grid data comprises:
extracting specified field data from the MR positioning data;
cleaning the extracted MR positioning data, and removing invalid data and repeated data in the MR positioning data;
combining the cleaned MR data with the road grid data to screen out the road grid MR data;
generating a trajectory over each user time series based on the road grid MR data;
based on the trajectory, a moving point, a stationary point, and an abnormal point on the user trajectory in any time series are identified.
3. The method for identifying the movement of people on roads according to claim 2, wherein the washing of the extracted MR positioning data to remove invalid data and repeated data in the MR positioning data comprises:
judging whether a field in the specified field data extracted from the MR positioning data is a null value, if so, removing the whole data;
judging whether the specified field data extracted from the MR positioning data has repeated data or not, and if so, rejecting the whole repeated data.
4. The method of identifying motion of persons in a roadway as recited in claim 2, wherein generating a trajectory over each user time series based on the road grid MR data comprises:
and based on the road grid MR data, the data of each user are gathered in respective sets and are sorted according to time, and the sorted single-user data sequence forms a track on the time sequence of each user.
5. The method for identifying the movement of people on the road according to claim 2, wherein the step of identifying the movement point, the static point and the abnormal point on the user track in any time sequence based on the track comprises the following steps:
based on the track, identifying a moving point, a static point and an abnormal point on a user track in any time sequence by a multi-pilot-frequency bidirectional sliding window track algorithm;
the multiple pilot frequency bidirectional sliding window track algorithm is provided with two layers of sliding windows, wherein the number of points of the first layer of sliding windows is used as a sliding unit, and the time period of the second layer of sliding windows is used as a sliding unit.
6. The method for identifying the movement of people on the road as claimed in claim 5, wherein the method for identifying the moving point and the static point on the user track in any time sequence by a multi-pilot frequency bidirectional sliding window track algorithm comprises the following steps:
setting a first layer sliding window of the multi-pilot-frequency bidirectional sliding window track algorithm to comprise two points; wherein one of the two points is a target point;
acquiring the time difference and the distance between the two points; and calculating the speed between the two points based on the time difference and the distance, and judging whether the target point in the window is a moving point or a static point based on the speed between the two points.
7. The method for identifying the movement of people on the road according to claim 5 or 6, wherein the method for identifying the movement point and the abnormal point on the user track in any time sequence by a multiple different frequency bidirectional sliding window track algorithm comprises the following steps:
sliding a sliding unit forwards in a second layer sliding window of the multi-pilot-frequency bidirectional sliding window track algorithm, and calculating the average speed and the average distance of points in the forward window and/or a plurality of grids with the highest occurrence frequency in the window by a mean value clustering algorithm; wherein a target point is included within the forward window;
sliding a sliding unit backwards in a sliding window of a second layer of the multi-pilot-frequency bidirectional sliding window track algorithm, and calculating the average speed and the average distance of points in a backward window and/or a plurality of grids with the highest occurrence frequency in the window through a multi-dimensional density clustering algorithm; wherein the rearward window also includes a target point therein;
and judging whether the target point in the window is a moving point or an abnormal point based on the average speed and the average distance of the points in the forward window and/or the grids with the highest occurrence times in the window and the average speed and the average distance of the points in the backward window and/or the grids with the highest occurrence times in the window.
8. The method for identifying the movement of people on the road according to claim 5 or 6, wherein the method for identifying the movement point and the abnormal point on the user track in any time sequence by a multiple different frequency bidirectional sliding window track algorithm comprises the following steps:
sliding a sliding unit forwards in a second layer of sliding window of the multi-pilot-frequency bidirectional sliding window track algorithm, and calculating the average speed and the average distance of points in the forward window and/or a plurality of grids with the highest occurrence frequency in the window through a multi-dimensional density clustering algorithm; wherein a target point is included within the forward window;
sliding a sliding unit backwards in a sliding window of a second layer of the multi-pilot-frequency bidirectional sliding window track algorithm, and calculating the average speed and the average distance of points in a backward window and/or a plurality of grids with the highest occurrence frequency in the window by a mean value clustering algorithm; wherein the rearward window also includes a target point therein;
and judging whether the target point in the window is a moving point or an abnormal point based on the average speed and the average distance of the points in the forward window and/or the grids with the highest occurrence times in the window and the average speed and the average distance of the points in the backward window and/or the grids with the highest occurrence times in the window.
9. The method for identifying the movement of people on the road according to claim 5 or 6, wherein the method for identifying the movement point and the abnormal point on the user track in any time sequence by a multiple different frequency bidirectional sliding window track algorithm comprises the following steps:
sliding a sliding unit forwards in a second layer sliding window of the multi-pilot-frequency bidirectional sliding window track algorithm, and calculating the average speed and the average distance of points in the forward window and/or a plurality of grids with the highest occurrence frequency in the window by a mean value clustering algorithm; wherein a target point is included within the forward window;
sliding a sliding unit backwards in a sliding window of a second layer of the multi-pilot-frequency bidirectional sliding window track algorithm, and calculating the average speed and the average distance of points in a backward window and/or a plurality of grids with the highest occurrence frequency in the window by a mean value clustering algorithm; wherein the rearward window also includes a target point therein;
and judging whether the target point in the window is a moving point or an abnormal point based on the average speed and the average distance of the points in the forward window and/or the grids with the highest occurrence times in the window and the average speed and the average distance of the points in the backward window and/or the grids with the highest occurrence times in the window.
10. The method for identifying the movement of people on the road according to claim 5 or 6, wherein the method for identifying the movement point and the abnormal point on the user track in any time sequence by a multiple different frequency bidirectional sliding window track algorithm comprises the following steps:
sliding a sliding unit forwards in a second layer of sliding window of the multi-pilot-frequency bidirectional sliding window track algorithm, and calculating the average speed and the average distance of points in the forward window and/or a plurality of grids with the highest occurrence frequency in the window through a multi-dimensional density clustering algorithm; wherein a target point is included within the forward window;
sliding a sliding unit backwards in a sliding window of a second layer of the multi-pilot-frequency bidirectional sliding window track algorithm, and calculating the average speed and the average distance of points in a backward window and/or a plurality of grids with the highest occurrence frequency in the window through a multi-dimensional density clustering algorithm; wherein the rearward window also includes a target point therein;
and judging whether the target point in the window is a moving point or an abnormal point based on the average speed and the average distance of the points in the forward window and/or the grids with the highest occurrence times in the window and the average speed and the average distance of the points in the backward window and/or the grids with the highest occurrence times in the window.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811044619.7A CN109344729B (en) | 2018-09-07 | 2018-09-07 | Method for identifying movement of people on road |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811044619.7A CN109344729B (en) | 2018-09-07 | 2018-09-07 | Method for identifying movement of people on road |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109344729A true CN109344729A (en) | 2019-02-15 |
CN109344729B CN109344729B (en) | 2021-10-26 |
Family
ID=65304572
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811044619.7A Active CN109344729B (en) | 2018-09-07 | 2018-09-07 | Method for identifying movement of people on road |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109344729B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111080351A (en) * | 2019-12-05 | 2020-04-28 | 任子行网络技术股份有限公司 | Clustering method and system for multi-dimensional data set |
CN114530043A (en) * | 2022-03-03 | 2022-05-24 | 上海闪马智能科技有限公司 | Event detection method and device, storage medium and electronic device |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104239556A (en) * | 2014-09-25 | 2014-12-24 | 西安理工大学 | Density clustering-based self-adaptive trajectory prediction method |
CN105163321A (en) * | 2014-06-03 | 2015-12-16 | 中国移动通信集团公司 | Data flow positioning method and device |
US20150363508A1 (en) * | 2014-06-17 | 2015-12-17 | Naveen NANDAN | Grid-based analysis of geospatial trajectories |
CN106878951A (en) * | 2017-02-28 | 2017-06-20 | 上海讯飞瑞元信息技术有限公司 | User trajectory analysis method and system |
CN107133700A (en) * | 2017-05-12 | 2017-09-05 | 西南交通大学 | Mobile phone signaling data road network method based on R* tree indexes |
-
2018
- 2018-09-07 CN CN201811044619.7A patent/CN109344729B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105163321A (en) * | 2014-06-03 | 2015-12-16 | 中国移动通信集团公司 | Data flow positioning method and device |
US20150363508A1 (en) * | 2014-06-17 | 2015-12-17 | Naveen NANDAN | Grid-based analysis of geospatial trajectories |
CN104239556A (en) * | 2014-09-25 | 2014-12-24 | 西安理工大学 | Density clustering-based self-adaptive trajectory prediction method |
CN106878951A (en) * | 2017-02-28 | 2017-06-20 | 上海讯飞瑞元信息技术有限公司 | User trajectory analysis method and system |
CN107133700A (en) * | 2017-05-12 | 2017-09-05 | 西南交通大学 | Mobile phone signaling data road network method based on R* tree indexes |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111080351A (en) * | 2019-12-05 | 2020-04-28 | 任子行网络技术股份有限公司 | Clustering method and system for multi-dimensional data set |
CN114530043A (en) * | 2022-03-03 | 2022-05-24 | 上海闪马智能科技有限公司 | Event detection method and device, storage medium and electronic device |
Also Published As
Publication number | Publication date |
---|---|
CN109344729B (en) | 2021-10-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107682319B (en) | Enhanced angle anomaly factor-based data flow anomaly detection and multi-verification method | |
CN105160871B (en) | A kind of method of highway car upper servant's identification temporarily | |
CN108647649A (en) | The detection method of abnormal behaviour in a kind of video | |
CN108765961B (en) | Floating car data processing method based on improved amplitude limiting average filtering | |
CN111046968B (en) | Road network track clustering analysis method based on improved DPC algorithm | |
CN111104398B (en) | Detection method and elimination method for intelligent ship approximate repeated record | |
CN113516105B (en) | Lane detection method and device and computer readable storage medium | |
CN109344729B (en) | Method for identifying movement of people on road | |
CN103345842A (en) | Road vehicle classification system and method | |
CN112800115B (en) | Data processing method and data processing device | |
CN115311858B (en) | Urban road section hierarchical control method based on traffic flow toughness | |
CN111046303A (en) | Automatic detection method, device and system for hot spot area | |
CN108830882B (en) | Video abnormal behavior real-time detection method | |
CN103886609B (en) | Vehicle tracking method based on particle filtering and LBP features | |
CN106033643B (en) | A kind of data processing method and device | |
CN105049286A (en) | Cloud platform speed measurement data judging method based on hierarchical clustering | |
CN110851450A (en) | Accompanying vehicle instant discovery method based on incremental calculation | |
CN112801181B (en) | Urban signaling traffic flow user classification and prediction method, storage medium and system | |
CN111210634B (en) | Intelligent traffic information processing method and device, intelligent traffic system and server | |
CN101600115A (en) | A kind of method of eliminating periodic characteristic block of image stabilization system | |
CN111221819B (en) | Similar typhoon rapid matching method based on multidimensional data space segmentation | |
Muhammad et al. | Transportation mode detection from gps data: A data science benchmark study | |
CN109238271B (en) | Line fitting method based on time | |
CN113379334A (en) | Road section bicycle riding quality identification method based on noisy track data | |
CN112348265A (en) | Feasible path mining method and device under monitoring scene |
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 |