CN110288044A - A kind of trajectory simplification method divided based on track with Priority Queues - Google Patents
A kind of trajectory simplification method divided based on track with Priority Queues Download PDFInfo
- Publication number
- CN110288044A CN110288044A CN201910586296.2A CN201910586296A CN110288044A CN 110288044 A CN110288044 A CN 110288044A CN 201910586296 A CN201910586296 A CN 201910586296A CN 110288044 A CN110288044 A CN 110288044A
- Authority
- CN
- China
- Prior art keywords
- track
- point
- trajectory
- priority
- priority queues
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Train Traffic Observation, Control, And Security (AREA)
Abstract
The present invention discloses a kind of trajectory simplification method divided based on track with Priority Queues, includes: S1, calculating the average speed of the orbit segment of initial trace, obtain the average speed of entire track;S2, clustering is carried out to the average speed of entire track, generates class name and velocity interval table;S3, the orbit segment in initial trace is demarcated, forms the orbit segment set being made of label;S4, the division that calibrated orbit segment set is carried out to sub-trajectory, until result meets preset threshold value;S5, it is based on ready-portioned sub-trajectory, generates the rate of simplification and tolerance;S6, parameter is carried out to simplify processing in the incoming trajectory simplification algorithm based on Priority Queues together with sub-trajectory section;S7, the simplification result of each sub-trajectory is merged.The invention avoids primal algorithms the performance degradation on biggish track collection the problem of, while algorithm also has preferable performance in terms of space error, synchronous Euclidean distance error, has very high practical value.
Description
Technical field
It is the present invention relates to field of mobile computing, in particular to a kind of to be divided and the letter of the track of Priority Queues based on track
Change method.
Background technique
The Emirates Telecommunications (Etisalat) of the United Arab Emirates in Barcelona conference in 2019, it is straight that person of outstanding talent throws a thousand pieces of gold
It connects and has signed the national big list of 5G deployment.Japan is soft silver-colored (Softbank)) laboratory Future Stride realizes three vehicle groups volume
Group is single to be driven, and the advantage of 5G communication is sufficiently illustrated.5G communication has the characteristic of high speed low delay.These characteristics are at nobody
Drive etc. has important role.But either 5G or conventional satellite communication, their cost are all very high.And it is general
The method of sampling is sensor to be installed in mobile object, and cooperate electronic map skill using existing satellite or ground base station
Art etc., within a certain period of time send target where location information and temporal information.Since technical conditions limit, position obtained
It is discrete for setting with temporal information.And accurate track, sample rate are often higher (one sampled point of every five seconds), cause track number
It is very huge according to collecting, great difficulty is brought to the transmission of track data, storage, analysis mining effective information etc..Therefore,
It is necessary to simplify under given error condition to initial trace data.
The storage of mobile object track data and mining analysis are the important topics in mobile computing field.In automatic Pilot
And it is widely used in intelligent transportation.Wherein, mobile object trajectory simplification is the important component in the technical field.Although
People have been presented for various trajectory simplification algorithms, but these algorithms all have some limitations.It is one of classical
Trajectory simplification algorithm is SQUISH algorithm.Algorithm SQUISH is as follows:
Input: initial trace Traw, Priority Queues size size;
Output: simplified track Tsim;
1, it initializes length and is the Priority Queues Buffer of size, while recording a little and weight;
2、forPiin TrawThe entire initial trace of: // traversal;
3, by PiBeing pressed into Buffer, the last one is put and sets π (Pi)=∞ // π (Pj) tracing point redundancy, weight.
4、π(Pi-1)=sed (Pi-2Pi-1Pi) // penultimate the point in buffer area is assigned initial value is the point and its forerunner
It is subsequent
// synchronization the Euclidean distance formed;
5, if len (Buffer) > size: // buffer area has been expired, and len () is the length of sequence;
6, π (P is found outj) the 0 the smallest point of < j < size;
7、del(Pj);
8, by π (Pj) value of 0 < j < size is applied directly to π (pred (Pj))andπ(succ(Pj));
9、returnTsim=Buffer;
SQUISH algorithm is an on-line Algorithm, and the point in track is inputted in real time one by one.Additionally need input is slow
Area (Priority Queues size) size is rushed, this buffer area is referred to as Priority Queues.Initially, the tracing point of all acquisitions is all stored in slow
It rushes in area, until buffer area is full of.Once data volume exceeds buffer area, into buffer area, any one point of deposit requires to delete
An existing point in buffer area.The deleted point selected is seldom comprising " information " or does not include the point of " information ".Believe
Cease the point of high redundancy.The original trace compression algorithm based on Priority Queues increases when a point is removed from Priority Queues
Add the priority of its neighbor point.This method is a kind of good approximation.Motion profile can be considered a horse for statistically
Er Kefu chain only will affect the subsequent of it so a point is only influenced by its forerunner.But actual conditions complexity is more than hypothesis
Situation is serious, as more and more points are removed from buffer area, error propagation can inevitably occurs.So the algorithm lacks
It is also fairly obvious for falling into.The excessively coarse redundancy method of salary distribution leads to systematic error persistent accumulation.In actual use
Higher simplification than when the algorithm performance may mutually be on duty.A problem more outstanding is that general simplified algorithm all can
Allow user oneself define range of tolerable variance, come to final one guiding error range of result and this algorithm does not have.It is not inconsistent
Common use habit is closed, so that user's degrees of freedom are relatively low.
In order to solve in the original trace compression algorithm based on Priority Queues, in fact it could happen that the accumulation of error, higher
The rate of simplification or when facing biggish track, in fact it could happen that performance sharp-decay, and lack the tolerance with directive significance and refer to
Fixed, the freedom degree of user is lower, researches and develops a kind of actually necessary with the trajectory simplification algorithm of Priority Queues based on track division.
Summary of the invention
The purpose of the present invention is providing a kind of trajectory simplification method divided based on track with Priority Queues, it is to be directed to
Deficiency existing for SQUISH algorithm is improved, and is carried out firstly for the track of input according to the average speed of each of which orbit segment
Cluster, the velocity characteristic in initial trace intensively reflects, and carries out rough classification to the motion state of initial trace,
Obtain velocity interval table.The orbit segment in initial trace is demarcated according to this table again, what formation one had been demarcated
Track carries out sub-trajectory division according to post-class processing then to the track foundation that this has been demarcated.For ready-portioned sub- rail
Mark is compressed, then becomes compression result after merging.Present invention employs k-means clustering algorithm, post-class processing algorithm,
Auto-adaptive parameter generates, and obtains the more effectively trace compression algorithm based on Priority Queues.
In order to achieve the above object, the invention is realized by the following technical scheme:
A kind of trajectory simplification method divided based on track with Priority Queues, the method includes the steps of:
S1, input initial trace after pretreatment, seek average speed to each orbit segment in initial trace
The operation of degree;
S2, clustering is carried out to the average speed of the entire track of step S1, generates class name and velocity interval table;
S3, it is based on the class name and velocity interval table, the average speed of all orbit segments in initial trace is marked
It is fixed, form the orbit segment set being made of label;
S4, the division that calibrated orbit segment set is carried out to sub-trajectory, until the gini index of each subsegment divided
Meet preset threshold value;
S5, it is based on ready-portioned sub-trajectory in the step S4, generates the rate of simplification and corresponding tolerance;
S6, the simplification rate of adaptive generation in the step S5 and Tolerance Parameters are passed to together with sub-trajectory section based on excellent
It carries out simplifying processing in the trajectory simplification algorithm of first queue;
S7, the simplification result of each sub-trajectory is merged, obtains final result.
Preferably, in the step S1, pretreated process includes:
The data of initial trace are used into Pi=(xi,yi,ti) Unified Form, wherein i is indicated i-th of initial trace
Tracing point, PiIndicate the spatial position coordinate of i-th of tracing point, xi,yiRespectively indicate PiBy projecting on two-dimensional surface
Horizontal, ordinate, tiIndicate the acquisition moment i.e. timestamp of the coordinate points;
The initial trace of input indicates are as follows:
Wherein, n is the number of tracing point contained by track, i.e. len (Traw),Refer to an orbit segment.
Preferably, in the step S2, the average speed of every section of track are as follows:
Wherein,For orbit segmentLength,For orbit segmentThe difference of the timestamp of first and last two o'clock
Value;
The average speed velocity_tab of all orbit segments of entire track is denoted as:
Preferably, in the step S2, by the average speed of all orbit segments of entire track
Velocity_tab carries out clustering using K-means algorithm, includes:
Silhouette coefficient is calculated separately as cluster coefficients k value by traversing the numerical value in certain threshold range, it will wherein
The k value used when the smallest k value of silhouette coefficient is as actual division is incited somebody to action in the result of the amendment with k value, K-means algorithm
It constantly corrects to optimal solution.
Preferably, in the step S2, according to the class name and velocity interval table, to all orbit segments of entire track
Average speed velocity_tab, which is marked, generates lable_tab, forms the orbit segment set being made of label.
Preferably, in the step S4, the calibrated orbit segment set is carried out using the method for post-class processing
The division of sub-trajectory includes: by traversing entire track, bipartition point is successively carried out in everywhere based on orbit segment set, and
The variation for calculating the gini index before and after division, selecting gini index to reduce the maximum division mode of degree is optimal division
Mode;
Wherein, gini index calculation formula is as follows:
In formula, TsampleIndicate example track;Split (i) indicates to divide at i point;T1-iIndicate that the 1st point arrives in track
The track of i-th point of composition, Ti-nIndicate the track of i-th point to n-th point of composition;Gini(T1-i) and Gini (Ti-n) be
The gini index of corresponding track;
Formula (4) is a recursion, recursively carries out the division of sub-trajectory, until the gini index of each sub-trajectory section
When less than the preset threshold value, terminate recursive procedure.
Preferably, former according to minimum description length based on ready-portioned sub-trajectory in the step S4 in the step S5
Then, the parameter predigesting rate size of adaptation is generatedεWith tolerance errorε;Wherein, sizeεIndicate that the track for simplifying track is counted,
errorεIndicate maximum synchronous Euclidean distance.
Preferably, it in the step S5, further includes:
S51, simplified track is constructed, wherein initial simplification track is the straightway that the starting point of track forms to terminal;
S52, the searching point that there is maximum synchronous Euclidean distance in the initial trace of input with simplified track, the maximum are found
Synchronous Euclidean distance is denoted as errorε, wherein the synchronous Euclidean distance notation of the maximum of initial value is errormax;
S53, the searching point in step S52 is added in simplified track, and calculated equilibrium parameter fix:
In formula, sizeεFor the track points for simplifying track, sizemaxFor the track points of the initial trace of input;
S54, the entire initial trace of traversal, record is balanced the minimum value min (fix) of parameter fix, corresponding at this time
sizeεWith errorεFor the track optimal parameter of generation.
Preferably, in the step S6, the trajectory simplification algorithm based on Priority Queues includes following procedure:
S61, initialization, construction length are input parameter sizeεPriority Queues, while storage track point and priority,
The priority of Priority Queues starting point and terminal is always infinitely great;
S62, tracing point is constantly pressed into Priority Queues, one point of every addition, current point becomes Priority Queues
New terminal, original terminal is by calculating it with the synchronous Euclidean distance of the orbit segment of former and later two adjacent point compositions as former
First terminal priority;
Whether the priority of the original terminal calculated in S63, judgment step S62 is less than the parameter error of inputε, if so,
Original terminal is then directly deleted in Priority Queues, otherwise, retains original terminal;
S64, after adding new point, if the length of Priority Queues is greater than size at this time, sought in the Priority Queues
Look for the point of minimum priority and be deleted, at the same by the priority of the point according to its forerunner's point with it is anti-at a distance from subsequent point
Than the priority for distributing to its forerunner's point and subsequent point;
The process of S65, the S63 and S64 addition point that repeats the above steps, until the ending of the initial trace of input is arrived in processing;
S66, the synchronization Euclidean distance for calculating obtained new Priority Queues each point again and synchronous with maximum European
Distance errorεCompare, deletes and be less than errorεPoint, left point composition track be final simplification track.
In the step S64, further include:
For tracing point Pi、Pj、PkIf delPjAnd i < j < k:
Wherein, π (Pi) it is PiPriority, π (Pj) it is PjPriority, π (Pk) it is PkPriority;Respectively orbit segmentLength.
Compared with prior art, the invention has the benefit that the initial parameter and mathematical information of original simplification algorithm
Closely coupled is the size of Priority Queues, needs user that could accurately propose after certain experience accumulation;And
Initial parameter needed for the new algorithm of the present invention be gini index threshold value this be that the perception for dividing scattered degree to track is retouched
It states;As long as the present invention has preliminary understanding that can directly give track;Furthermore of the invention in terms of compression result
ATNS algorithm avoid to a certain extent primal algorithm the performance degradation on biggish track collection the problem of, while algorithm is in sky
Between also have preferable performance in terms of error, synchronous Euclidean distance error;In general, proposed by the invention to be divided based on track
It is preferable with the simplification algorithm using effect of Priority Queues, there is very high practical value.
Detailed description of the invention
Fig. 1 is the trajectory simplification method ATNS flow chart of the invention divided based on track with Priority Queues;
Fig. 2 is the comparison schematic diagram of the runing time of innovatory algorithm (ATNS) and primal algorithm (SQUISH) of the invention;
Fig. 3 is that innovatory algorithm (ATNS) of the invention and the synchronous Euclidean distance of primal algorithm (SQUISH) mistake (SED) are poor
Comparison schematic diagram;
Fig. 4 is that innovatory algorithm (ATNS) of the invention shows compared with the space error (PED) of primal algorithm (SQUISH)
It is intended to;
Fig. 5 is a kind of class name and velocity interval table that the present invention enumerates.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described.
The invention discloses it is a kind of based on track divide with the trajectory simplification method ATNS of Priority Queues, this method include with
Lower step:
S1, input initial trace and gini index threshold value, and orbit segment is carried out to by pretreated initial trace
The calculating of average speed;
S2, the processing of the clustering based on K-means algorithm is carried out for above-mentioned calculated result, generate class name and speed model
Enclose table;
S3, based on the class name and velocity interval table generated in step S2, to the average speed of the orbit segment in initial trace
It is demarcated, forms one group of orbit segment set being made of label;
S4, for calibrated orbit segment set, the division of sub-trajectory is carried out using the method for post-class processing, until drawing
The gini index for each subsegment divided meets the threshold value provided in advance;
S5, suitable parameter is generated according to Minimal Description Length Criterion for ready-portioned sub-trajectory in step S4
sizeεWith errorε;Wherein, sizeεFor the track points (also referred to as simplifying rate) for simplifying track, errorεIt is synchronized for maximum European
Distance (also referred to as tolerance).
S6, the parameter of step S5 adaptive generation is passed to the trajectory simplification calculation based on Priority Queues together with sub-trajectory section
In method, carry out simplifying processing;
S7, the simplification result of each sub-trajectory is merged, forms result.
In the step S1, data prediction includes following:
For initial trace data often without a kind of unified form, but for convenience calculate, this algorithm recommend
Using shaped like Pi=(xi,yi,ti) form indicate, i indicate initial trace i-th of key point, PiIndicate i-th of key point
Spatial position coordinate, xi,yiRespectively indicate PiBy projecting to the cross on two-dimensional surface, ordinate;Location information xi,yiIt can be according to
It is directly obtained according to geographical knowledge or defined electronic map;Last tiIndicate the acquisition moment i.e. timestamp of the coordinate points;
As zero when being acquired often due to data derived from 1980 year January 6 day zero of this algorithm of GPS sensor recommendation using universal time
Point, timestamp later are to add up to form in the zero point, since then could be as the input of algorithm using unified format.
In the step S2, the K-means algorithm (k means clustering algorithm) is specifically comprising following: certain by traversing
Numerical value in threshold range (motion mode that may include in initial trace such as: walking, running, number by bus), when
Make cluster coefficients k value, calculate separately silhouette coefficient, using the smallest k value of wherein silhouette coefficient as the k value used when actual division
(i.e. best cluster coefficients);And in the amendment with k value, the result of K-means algorithm also will be corrected constantly to optimal solution.
For the initial trace T of inputrawIt can state are as follows:
Wherein, n is number, that is, len (T of tracing point contained by trackraw),It is an orbit segment;Each section of track
Average speed be are as follows:
Wherein,For orbit segmentLength,For orbit segmentThe difference of the timestamp of first and last two o'clock
Value.
Then have for the average speed of entire track:
In the present embodiment, clustering method, either K- are used for the average speed velocity_tab of entire track
The calculating of means algorithm or silhouette coefficient, select it is excellent, can simply directly use TensorFlow and Python Bao Laishi
It is existing, the table of similar velocity interval-class name, velocity amplitude as shown in Figure 5 and error threshold are generated after handling by the step
The table of value;Velocity_tab is marked according to this table and generates lable_tab, shaped like (000111333...),
The combination of i.e. each orbit segment label, number therein is class name.
The algorithm of the division sub-trajectory of the S4 specifically:
By traversing entire track, the rail being made of label after calibration (is i.e. passed through based on lable_tab in step S3
Mark Duan Jihe), bipartition point successively is carried out in everywhere, and calculate the variation of the gini index before and after division, selects Geordie
It is optimal division mode that index, which reduces the maximum division mode of degree,.
Gini index calculation formula is as follows:
In formula, TsampleIndicate example track;Split (i) is to divide at i point;T1-iIt indicates in track 1st o'clock to the
The track of i point composition, Ti-nSimilarly, the track of i-th point to n-th point of composition is indicated;Gini(T1-i) and Gini (Ti-n)
For the gini index of corresponding track;
Above-mentioned formula (4) is a recursion, then recursively carries out the division of sub-trajectory, until the base of each sub-trajectory section
Buddhist nun's index is less than defined threshold value, and (the gini index threshold value inputted when i.e. algorithm starts, the threshold value is by wanting initial trace
Cutting degree choose 0-1 within the scope of number) when, terminate recursive procedure.
In the step S5, the generation step of auto-adaptive parameter is as follows:
S51, simplified track is constructed;Wherein, the straightway that initial simplification track forms to terminal for the starting point of track;
S52, the point for having maximum synchronous Euclidean distance in the initial trace of input with simplified track, this most Datong District are found
Step Euclidean distance is denoted as errorε(also referred to as tolerance), wherein the synchronous Euclidean distance notation of the maximum of initial value is errormax;
S65, this point found in step S52 is added in simplified track, calculated equilibrium parameter fix:
Wherein, sizeεFor the track points for simplifying track, sizemaxFor the track points of the initial trace of input.
S54, the entire initial trace of traversal, record is balanced the minimum value min (fix) of parameter fix, corresponding at this time
sizeεWith errorεFor the track optimal parameter of generation.
In the step S6, the trajectory simplification algorithm based on Priority Queues is as follows:
S61, initialization, construction length are input parameter sizeεPriority Queues, can storage track point and preferential simultaneously
The priority of degree, the Priority Queues starting point and terminal is always infinitely great;
S62, tracing point is constantly pressed into Priority Queues, one point of every addition, which becomes Priority Queues
New terminal, original terminal by calculate it with it is adjacent former and later two point form orbit segment synchronous Euclidean distance conduct
Original terminal priority;
Whether the priority of the original terminal calculated in S63, judgment step S62 is less than the parameter error of inputε, if so,
Original terminal is then directly deleted in Priority Queues, otherwise, retains this point.
If S64, the length of Priority Queues is greater than parameter size after adding new pointε, then found in Priority Queues
The point of minimum priority deletes it, while by the priority of the point according to the inverse ratio minute at a distance from its forerunner's point and subsequent point
The priority of dispensing its forerunner's point and subsequent point.
In the step S64, for tracing point Pi、Pj、PkIf delPjAnd i < j < k:
Wherein, π (P hereini) it is PiPriority, π (Pj) it is PjPriority, π (Pk) it is PkPriority;Respectively orbit segmentLength.
The process of S65, the S63 and S64 addition point that repeats the above steps, until the ending of the initial trace of input is arrived in processing;
S66, the synchronization Euclidean distance that each point is calculated again to obtained new Priority Queues it is synchronous with maximum it is European away from
From errorεCompare, deletes and be less than errorεPoint, left point composition track be final simplification track.
Innovatory algorithm of the invention (ATNS) is illustrated in figure 2 to show compared with the runing time of primal algorithm (SQUISH)
It is intended to, abscissa is track, and ordinate is runing time.As seen from the figure, at runtime between the order of magnitude on both difference not
It is big and innovatory algorithm is relatively advantageous.
It is illustrated in figure 3 innovatory algorithm of the invention (ATNS) Euclidean distance error synchronous with primal algorithm (SQUISH)
(SED) comparison schematic diagram, abscissa are track, and ordinate is synchronous Euclidean distance error.As seen from the figure, two algorithms are with number
According to the increase of collection, the accumulation of error also shows in increase but innovatory algorithm more excellent.
It is illustrated in figure 4 the space error (PED) of innovatory algorithm of the invention (ATNS) and primal algorithm (SQUISH)
Comparison schematic diagram, abscissa are track, and ordinate is space error.As seen from the figure, increase of two algorithms with data set, error
Accumulation also shows in increase but innovatory algorithm more excellent.
In conclusion ATNS algorithm of the invention avoids the property on biggish track collection of primal algorithm to a certain extent
The problem of capable of decaying, while algorithm also has preferable performance in terms of space error, synchronous Euclidean distance error;In general,
Proposed by the invention being divided based on track is preferable with the simplification algorithm using effect of Priority Queues, has very high practical valence
Value.
It is discussed in detail although the contents of the present invention have passed through above preferred embodiment, but it should be appreciated that above-mentioned
Description is not considered as limitation of the present invention.After those skilled in the art have read above content, for of the invention
A variety of modifications and substitutions all will be apparent.Therefore, protection scope of the present invention should be limited to the appended claims.
Claims (10)
1. a kind of trajectory simplification method divided based on track with Priority Queues, which is characterized in that the method includes the steps of:
S1, input initial trace by pretreatment after, each orbit segment in initial trace is carried out seeking average speed
Operation;
S2, clustering is carried out to the average speed of the entire track of step S1, generates class name and velocity interval table;
S3, it is based on the class name and velocity interval table, the average speed of all orbit segments in initial trace is demarcated, shape
At the orbit segment set being made of label;
S4, the division that calibrated orbit segment set is carried out to sub-trajectory, until the gini index of each subsegment divided meets
Preset threshold value;
S5, it is based on ready-portioned sub-trajectory in the step S4, generates the rate of simplification and corresponding tolerance;
S6, the simplification rate of adaptive generation in the step S5 and Tolerance Parameters are passed to together with sub-trajectory section based on preferential team
It carries out simplifying processing in the trajectory simplification algorithm of column;
S7, the simplification result of each sub-trajectory is merged, obtains final result.
2. the trajectory simplification method with Priority Queues is divided based on track as described in claim 1, which is characterized in that
In the step S1, pretreated process includes:
The data of initial trace are used into Pi=(xi,yi,ti) Unified Form, wherein i indicate initial trace i-th of track
Point, PiIndicate the spatial position coordinate of i-th of tracing point, xi,yiRespectively indicate PiIt is horizontal, vertical on two-dimensional surface by projecting to
Coordinate, tiIndicate the acquisition moment i.e. timestamp of the coordinate points;
The initial trace of input indicates are as follows:
Wherein, n is the number of tracing point contained by track, i.e. len (Traw),Refer to an orbit segment.
3. the trajectory simplification method with Priority Queues is divided based on track as claimed in claim 2, which is characterized in that
In the step S2, the average speed of every section of track are as follows:
Wherein,For orbit segmentLength,For orbit segmentThe difference of the timestamp of first and last two o'clock;
The average speed velocity_tab of all orbit segments of entire track is denoted as:
4. the trajectory simplification method with Priority Queues is divided based on track as claimed in claim 3, which is characterized in that
In the step S2, the average speed velocity_tab of all orbit segments of entire track is used into K-means algorithm
Clustering is carried out, includes:
Silhouette coefficient is calculated separately as cluster coefficients k value by traversing the numerical value in certain threshold range, it will wherein profile
The k value used when the smallest k value of coefficient is as actual division, the result of the amendment with k value, K-means algorithm will be continuous
It corrects to optimal solution.
5. the trajectory simplification method divided based on track with Priority Queues as described in claim 3 or 4, which is characterized in that
In the step S2, according to the class name and velocity interval table, to the average speed of all orbit segments of entire track
Velocity_tab, which is marked, generates lable_tab, forms the orbit segment set being made of label.
6. the trajectory simplification method with Priority Queues is divided based on track as described in claim 1, which is characterized in that
In the step S4, the calibrated orbit segment set is subjected to drawing for sub-trajectory using the method for post-class processing
Point, include: by traversing entire track, bipartition point successively being carried out in everywhere based on orbit segment set, and calculate and dividing
The variation of the gini index of front and back, selecting gini index to reduce the maximum division mode of degree is optimal division mode;
Wherein, gini index calculation formula is as follows:
In formula, TsampleIndicate example track;Split (i) indicates to divide at i point;T1-iIt indicates in track 1st o'clock to i-th
The track of a point composition, Ti-nIndicate the track of i-th point to n-th point of composition;Gini(T1-i) and Gini (Ti-n) it is to correspond to
The gini index of track;
Formula (4) is a recursion, recursively carries out the division of sub-trajectory, until the gini index of each sub-trajectory section is less than
When the preset threshold value, terminate recursive procedure.
7. the trajectory simplification method with Priority Queues is divided based on track as described in claim 1, which is characterized in that
In the step S5, adaptation is generated according to Minimal Description Length Criterion based on ready-portioned sub-trajectory in the step S4
Parameter predigesting rate sizeεWith tolerance errorε;Wherein, sizeεIndicate that the track for simplifying track is counted, errorεIndicate maximum
Synchronous Euclidean distance.
8. the trajectory simplification method with Priority Queues is divided based on track as claimed in claim 7, which is characterized in that
In the step S5, further include:
S51, simplified track is constructed, wherein initial simplification track is the straightway that the starting point of track forms to terminal;
S52, the searching point that there is maximum synchronous Euclidean distance in the initial trace of input with simplified track is found, the maximum is synchronous
Euclidean distance is denoted as errorε, wherein the synchronous Euclidean distance notation of the maximum of initial value is errormax;
S53, the searching point in step S52 is added in simplified track, and calculated equilibrium parameter fix:
In formula, sizeεFor the track points for simplifying track, sizemaxFor the track points of the initial trace of input;
S54, the entire initial trace of traversal, record are balanced the minimum value min (fix) of parameter fix, at this time corresponding sizeεWith
errorεFor the track optimal parameter of generation.
9. the trajectory simplification method with Priority Queues is divided based on track as claimed in claim 8, which is characterized in that
In the step S6, the trajectory simplification algorithm based on Priority Queues includes following procedure:
S61, initialization, construction length are input parameter sizeεPriority Queues, while storage track point and priority, preferential team
The priority of column starting point and terminal is always infinitely great;
S62, tracing point is constantly pressed into Priority Queues, one point of every addition, current point becomes the new of Priority Queues
Terminal, original terminal by calculate its with the synchronous Euclidean distance of the orbit segment of former and later two adjacent point compositions as it is original eventually
Point priority;
Whether the priority of the original terminal calculated in S63, judgment step S62 is less than the parameter error of inputε, if so, directly
It connects and deletes original terminal in Priority Queues, otherwise, retain original terminal;
S64, after adding new point, if the length of Priority Queues is greater than size at this time, found most in the Priority Queues
The point of small priority is simultaneously deleted, while by the priority of the point according to the inverse ratio minute at a distance from its forerunner's point and subsequent point
The priority of dispensing its forerunner's point and subsequent point;
The process of S65, the S63 and S64 addition point that repeats the above steps, until the ending of the initial trace of input is arrived in processing;
S66, the synchronization Euclidean distance for calculating obtained new Priority Queues each point again and Euclidean distance synchronous with maximum
errorεCompare, deletes and be less than errorεPoint, left point composition track be final simplification track.
10. the trajectory simplification method with Priority Queues is divided based on track as claimed in claim 9, which is characterized in that
In the step S64, further include:
For tracing point Pi、Pj、PkIf delPjAnd i < j < k:
Wherein, π (Pi) it is PiPriority, π (Pj) it is PjPriority, π (Pk) it is PkPriority;
Respectively orbit segmentLength.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910586296.2A CN110288044B (en) | 2019-07-01 | 2019-07-01 | Trajectory simplification method based on trajectory division and priority queue |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910586296.2A CN110288044B (en) | 2019-07-01 | 2019-07-01 | Trajectory simplification method based on trajectory division and priority queue |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110288044A true CN110288044A (en) | 2019-09-27 |
CN110288044B CN110288044B (en) | 2021-06-08 |
Family
ID=68021537
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910586296.2A Active CN110288044B (en) | 2019-07-01 | 2019-07-01 | Trajectory simplification method based on trajectory division and priority queue |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110288044B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112668239A (en) * | 2020-12-30 | 2021-04-16 | 山东交通学院 | Hybrid power truck fleet experience teaching method based on counterstudy |
CN113052265A (en) * | 2021-04-26 | 2021-06-29 | 上海海事大学 | Moving object track simplification algorithm based on feature selection |
CN115775285A (en) * | 2022-12-09 | 2023-03-10 | 杭州滨电信息技术有限公司 | Client-side adaptive map vector line simplification method and system based on priority queue |
CN117194591A (en) * | 2023-06-20 | 2023-12-08 | 北京和德宇航技术有限公司 | Method, device, equipment and medium for displaying sphere track to plane map |
CN117874155A (en) * | 2024-01-15 | 2024-04-12 | 中国共产党四川省委员会党校、四川行政学院(四川长征干部学院) | Moving object track simplifying method based on speed immature random value |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104794494A (en) * | 2015-04-30 | 2015-07-22 | 中国地质大学(武汉) | Extensible quick trajectory clustering method |
CN104794295A (en) * | 2015-04-28 | 2015-07-22 | 上海海事大学 | Real-time ship trajectory simplifying method |
CN104951655A (en) * | 2015-06-08 | 2015-09-30 | 浙江大学 | Vehicle movement pattern mining method based on frequent pattern tree |
CN108931795A (en) * | 2018-05-21 | 2018-12-04 | 千寻位置网络有限公司 | Positioning equipment track optimization and boundary extraction method and device |
CN109238271A (en) * | 2018-08-23 | 2019-01-18 | 中国科学院电子学研究所苏州研究院 | A kind of time-based line fitting method |
CN109739585A (en) * | 2018-12-29 | 2019-05-10 | 广西交通科学研究院有限公司 | The traffic congestion point discovery method calculated based on spark cluster parallelization |
-
2019
- 2019-07-01 CN CN201910586296.2A patent/CN110288044B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104794295A (en) * | 2015-04-28 | 2015-07-22 | 上海海事大学 | Real-time ship trajectory simplifying method |
CN104794494A (en) * | 2015-04-30 | 2015-07-22 | 中国地质大学(武汉) | Extensible quick trajectory clustering method |
CN104951655A (en) * | 2015-06-08 | 2015-09-30 | 浙江大学 | Vehicle movement pattern mining method based on frequent pattern tree |
CN108931795A (en) * | 2018-05-21 | 2018-12-04 | 千寻位置网络有限公司 | Positioning equipment track optimization and boundary extraction method and device |
CN109238271A (en) * | 2018-08-23 | 2019-01-18 | 中国科学院电子学研究所苏州研究院 | A kind of time-based line fitting method |
CN109739585A (en) * | 2018-12-29 | 2019-05-10 | 广西交通科学研究院有限公司 | The traffic congestion point discovery method calculated based on spark cluster parallelization |
Non-Patent Citations (6)
Title |
---|
JONATHAN MUCKELL等: "Compression of Trajectory Data: A Comprehensive Evaluation and New Approach", 《GEOINFORMATICA》 * |
JONATHAN MUCKELL等: "SQUISH: An Online Approach for GPS Trajectory Compression", 《PROCEEDINGS OF THE 2ND INTERNATIONAL CONFERENCE ON COMPUTING FOR GEOSPATIAL RESEARCH & APPLICATIONS》 * |
NIRVANA MERATNIA等: "Spatiotemporal Compression Techniques for Moving Point Objects", 《ADVANCES IN DATABASE TECHNOLOGY-EDBT 2004》 * |
张甜等: "基于分段的移动对象轨迹简化算法", 《计算机应用研究》 * |
杨彪等: "一种基于速度的移动对象轨迹简化算法", 《计算机应用研究》 * |
王敏芬: "基于速度约束的轨迹划分方法", 《科技创新与应用》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112668239A (en) * | 2020-12-30 | 2021-04-16 | 山东交通学院 | Hybrid power truck fleet experience teaching method based on counterstudy |
CN113052265A (en) * | 2021-04-26 | 2021-06-29 | 上海海事大学 | Moving object track simplification algorithm based on feature selection |
CN113052265B (en) * | 2021-04-26 | 2024-04-05 | 上海海事大学 | Moving object track simplifying method based on feature selection |
CN115775285A (en) * | 2022-12-09 | 2023-03-10 | 杭州滨电信息技术有限公司 | Client-side adaptive map vector line simplification method and system based on priority queue |
CN115775285B (en) * | 2022-12-09 | 2023-09-15 | 杭州滨电信息技术有限公司 | Client self-adaptive map vector line simplifying method and system based on priority queue |
CN117194591A (en) * | 2023-06-20 | 2023-12-08 | 北京和德宇航技术有限公司 | Method, device, equipment and medium for displaying sphere track to plane map |
CN117194591B (en) * | 2023-06-20 | 2024-05-31 | 北京和德宇航技术有限公司 | Method, device, equipment and medium for displaying sphere track to plane map |
CN117874155A (en) * | 2024-01-15 | 2024-04-12 | 中国共产党四川省委员会党校、四川行政学院(四川长征干部学院) | Moving object track simplifying method based on speed immature random value |
CN117874155B (en) * | 2024-01-15 | 2024-06-11 | 中国共产党四川省委员会党校、四川行政学院(四川长征干部学院) | Moving object track simplifying method based on speed immature random value |
Also Published As
Publication number | Publication date |
---|---|
CN110288044B (en) | 2021-06-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110288044A (en) | A kind of trajectory simplification method divided based on track with Priority Queues | |
CN110502596B (en) | Track online sliding window compression method based on pedestrian track characteristics | |
CN106772685B (en) | Similar typhoon matching algorithm and software support system based on Web-GIS | |
CN108645421A (en) | Adaptive Online Map matching process based on Hidden Markov Model | |
CN106877875A (en) | A kind of vehicle running orbit compression method based on threshold value combination algorithm | |
CN110020178A (en) | Point of interest recognition methods, device, equipment and storage medium | |
CN113188544B (en) | Unmanned aerial vehicle base station path planning method based on cache | |
CN111429744A (en) | Road condition information fusion analysis method, device and storage medium | |
CN109724619A (en) | Information-pushing method, device and server | |
CN115358412A (en) | Asynchronous federal optimization method based on edge auxiliary vehicle network | |
Willis et al. | Forecasting distribution system loads using curve shape clustering | |
CN110059795A (en) | A kind of mobile subscriber's node networking method merging geographical location and temporal characteristics | |
Rainer et al. | Optimized diffuse scattering selection for large area real-time geometry-based stochastic modeling of vehicular communication links | |
CN107121661B (en) | Positioning method, device and system and server | |
CN109683180A (en) | A kind of GPS motion profile optimization method and system | |
CN107195192B (en) | A kind of recognition methods of vehicle-mounted orientation operating status | |
CN116414934A (en) | Space-time grid-based target operation safety airspace planning method | |
CN116055553A (en) | Method, system and device for processing vehicle push information | |
CN111199646B (en) | Urban signal control main road vehicle track reconstruction method based on sparse detection data | |
CN113421444A (en) | Urban road network signal control method and device based on vehicle path information | |
CN113536493A (en) | Effective path generation method based on clustering reverse thrust and section passenger flow estimation method | |
Hönle et al. | Preprocessing position data of mobile objects | |
US20240019260A1 (en) | Stochastic Route Planning for Electric Vehicles | |
Han et al. | A map matching algorithm for complex road conditions based on base station data | |
CN110322078A (en) | The flight flow control methods and computer storage medium of sector borders |
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 |