CN105788261B - A kind of road traffic spatial data compression method encoded based on PCA and LZW - Google Patents
A kind of road traffic spatial data compression method encoded based on PCA and LZW Download PDFInfo
- Publication number
- CN105788261B CN105788261B CN201610236010.4A CN201610236010A CN105788261B CN 105788261 B CN105788261 B CN 105788261B CN 201610236010 A CN201610236010 A CN 201610236010A CN 105788261 B CN105788261 B CN 105788261B
- Authority
- CN
- China
- Prior art keywords
- data
- section
- under
- road traffic
- sections
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
- G08G1/0129—Traffic data processing for creating historical data or processing based on historical data
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M7/00—Conversion of a code where information is represented by a given sequence or number of digits to a code where the same, similar or subset of information is represented by a different sequence or number of digits
- H03M7/30—Compression; Expansion; Suppression of unnecessary data, e.g. redundancy reduction
Landscapes
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Traffic Control Systems (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
A kind of road traffic spatial data compression method encoded based on PCA and LZW, based under same mode, spatially the highway traffic data of different sections of highway establishes road traffic features reference sequences;The road traffic section with correlation is chosen based on PCA to gather;Benchmark section is selected, using its data as road traffic reference data spatially;It extracts under same mode, the historical data in spatially other sections, the optimal threshold of space road traffic difference data is determined based on road traffic reference data under same mode, spatially as training data;The data for extracting spatially other sections, as real time data;Under same mode, based on road traffic reference data spatially, acquisition road traffic difference data;The compression for realizing road traffic spatial data is encoded based on LZW;Finally, the reconstruct of road traffic spatial data is realized based on LZW decoding techniques.The present invention, which simplifies, to be calculated, effectively improves processing speed.
Description
Technical field
The invention belongs to highway traffic data process fields, are related to the analysis and compression of highway traffic data, are a kind of roads
The compression method of road traffic data.
Background technology
With the continuous development of intelligent transportation system data acquisition technology, the intelligent transportation number obtained based on continuous acquisition
According to field of traffic will face mass data problem, it is necessary to carry out effective data compression to it, could be handled, be analyzed
And storage.
The internal characteristics of traffic flow data include mainly:Periodicity, similitude, correlation etc..The traffic flow of approach way
Between there is complicated spatial and temporal association, often similitude is higher, and same traffic flow shows extremely strong phase in time
Guan Xingyu is periodical.These similitudes show that there are a large amount of redundancies in data.
Based on the feature of traffic flow similitude, it is applied in highway traffic data compression field there are many method at present.
Include mainly:Principal Component Analysis (PCA), independent component analysis (ICA), predictive coding and dictionary encoding series process, based on small
The methods of wave (packet) transform method, artificial neural network.It mainly utilizes the thought of transform domain, highway traffic data is carried out more
Change of scale simultaneously carries out relevant treatment, realizes the compression of data, and obtain preferable effect.
Existing highway traffic data compression method Traffic Net data multipair greatly carry out data compression, to related road
Data compression method research in section time series is less.
Invention content
In order to overcome algorithm complexity, the lower deficiency of processing speed of existing highway traffic data compression method, the present invention
A kind of simplified road traffic spatial data compression side encoded based on PCA and LZW for calculating, effectively improving processing speed is provided
Method.
The technical solution adopted by the present invention to solve the technical problems is:
A kind of road traffic spatial data compression method encoded in PCA and LZW, includes the following steps:
1) based under same mode, spatially the highway traffic data of different sections of highway establish road traffic features refer to sequence
Row;Based on PCA methods, chooses the road traffic section with correlation and gather, process is as follows:
The road traffic historical data in s section, the acquisition in every section are extracted from reason traffic characteristic reference sequences
Data are r, and transform it into the matrix of s × r, are denoted as:Asⅹr;
Matrix AsⅹrJth row mean value be:
Based on aj, obtain AsⅹrNormalization matrix SAsⅹr:
The covariance matrix CSA of normalization matrix SA is:
The characteristic value D and feature vector V of covariance matrix CSA are obtained, then D=[λ1,λ2…λr];λ1≥λ2≥…≥λr;
Corresponding feature vector is:V=[v1,v2…vr];
Choose λ1, λ2The projection matrix VA that corresponding feature vector is constitutedr×2=[v1,v2], it is based on projection matrix and normalizing
The training matrix of change, seeks AsⅹrPrincipal component matrix A PCr×2:
APCs×2=SAsⅹr×VAr×2 (4)
Based on APCs×2, the distribution in s section is drawn on two dimensional surface, the distribution density of point indicates that corresponding road section is related
Property intensity, by correlation analysis, given threshold δ selects correlation to be more than the p+1 section of δ, and process is as follows:
Wherein, i, j indicate i-th, j section respectively, 0<i<S, 0<j<s;Indicate relevance function;
2) selection benchmark section, and using its data as road traffic reference data spatially;Extract under same mode,
The spatially historical data in other sections, as training data, based on road traffic base value under same mode, spatially
According to determining that the optimal threshold of space road traffic difference data, process are as follows:
Si(m*Δt,Mgh)=STi(m*Δt,Mgh)-SB(m*Δt,Mgh) (6)
ei(m,Mgh)=[Si(Δt,Mgh)Si(2*Δt,Mgh)...Si(m*Δt,Mgh)] (7)
pei(n,Mgh)=w (hei(m,Mgh)) (9)
pei(n,Mgh)=[Si'(1,Mgh)Si'(2,Mgh)...Si'(n,Mgh)] (10)
Wherein, Δ t is the collection period of road traffic state data;(m* Δs t) is that m-th of road traffic state data is adopted
Collect the period, 0≤m≤N, N indicate the quantity of the traffic information acquired daily;I (1≤i≤p) indicates i-th section;STi(m*Δ
T, Mgh) indicate mode MghUnder, (the highway traffic data in the moment i sections m* Δ t);SB (m* Δs t, Mgh) indicate mode MghUnder,
(the reference data in m* Δ t) moment benchmark section;Si(m* Δs t, Mgh) indicate mode MghUnder, (the training in the moment i sections m* Δ t)
The difference data of data and the reference data in benchmark section;ei(m, Mgh) indicate mode MghUnder, Δ t to (the period i sections m* Δ t)
Training data and benchmark section reference data difference data;hei(m, Mgh) indicate mode MghUnder, Δ t is to (when m* Δ t)
The difference data of the training data and the reference data in benchmark section in the sections i after section threshold process;Ei(m, Mgh) indicate mode Mgh
Under, Δ t to (the period i sections m* Δ t) choose threshold value;pei(n, Mgh) indicate mode MghUnder, Δ t to (the period i roads m* Δ t)
Result of the section with the difference data in benchmark section after LZW is encoded;Si' (n, Mgh) it is mode MghUnder, Δ t to (m* Δ t) period i
Nth data in result of the difference data in section and benchmark section after LZW is encoded;M is indicated in mode MghUnder, Δ t to (m*
The quantity in the sections i and the difference data in benchmark section before Δ t) duration compressions;N is indicated in mode MghUnder, Δ t to (m* Δs t)
Road traffic quantity after duration compression;W indicates LZW codings;Compression ratio is
3) data for extracting spatially other sections, as real time data;Mode MghUnder, based on road traffic spatially
Reference data, obtains road traffic difference data, and general expression is as follows:
MSj(m*Δt,Mgh)=SMj(m*Δt,Mgh)-SB(m*Δt,Mgh) (11)
errj(m,Mgh)=[MSj(Δt,Mgh)MSj(2*Δt,Mgh)...MSj(m*Δt,Mgh)] (12)
Wherein, j (1≤i≤p) indicates j-th strip section;SMj(m* Δs t, Mgh) indicate mode MghUnder, (the moment j roads m* Δ t)
The real time data of section;MSj(m* Δs t, Mgh) it is mode MghUnder, (the base of the real time data and benchmark section in the moment j sections m* Δ t)
The difference data of quasi- data;errj(m, Mgh) it is mode MghUnder, Δ t to (real time data in the period j sections m* Δ t) and benchmark road
The difference data of the reference data of section;
4) compression for realizing road traffic spatial data is encoded based on LZW, process is as follows:
The optimal threshold that the difference data in the sections i and benchmark section is trained is introduced into same mode Mgh, the sections j and benchmark
It in the difference data in section, is encoded in conjunction with LZW, realizes the compression in the sections j and benchmark section difference data, general expression is such as
Under:
herrsp(m*Δt,Mgh)=[herr1(m*Δt,Mgh)herr2(Δt,Mgh)...herrp'(m*Δt,Mgh)] (14)
perrp'(m*Δt,Mgh)=w (herrsp(m*Δt,Mgh)) (15)
perrp'(m*Δt,Mgh)=[MS1(m*Δt,Mgh)MS2(m*Δt,Mgh)...MSp'(m*Δt,Mgh)] (16)
Wherein, Eopt(Mgh) indicate trained optimal threshold;herrj(m* Δs t, Mgh) indicate mode MghUnder, (when m* Δ t)
Carve the difference data of the reference data in real time data and the benchmark section in the sections j after threshold process;M indicates mode MghUnder, Δ t arrives
(the quantity in the sections j and the difference data in benchmark section before m* Δ t) duration compressions;herrsp(m*Δt,Mgh) indicate mode Mgh
Under, (the magnitude-set of m* p section difference data of Δ t) moment;Perrp’(m*Δt,Mgh) indicate mode MghUnder, (m* Δs t)
The magnitude-set of difference data after p section compression of moment;MSj' (m* Δs t, Mgh) indicate mode MghUnder, (the moment j roads m* Δ t)
The compressed quantity of difference data of section;P ' indicates (quantity after m* Δ t) moment LZW coding;Compression ratio is:
Further, the compression method further includes following steps:
5) LZW decoding techniques are based on, realize that the reconstruct of road traffic spatial data, process are as follows:
The difference data in p section and benchmark section is reconstructed, in conjunction with reference data, realizes that p section counts in real time
According to decompression, general expression is as follows:
dperrp(m*Δt,Mgh)=w'(perrp'(m*Δt,Mgh)) (17)
dperrj(m,Mgh)=w'(perrj(Tn,Mgh)) (18)
CSMj(m,Mgh)=SB (m, Mgh)+dperrj(m,Mgh) (19)
Wherein, w ' indicates the decoding of LZW;dperrp(m* Δs t, Mgh) indicate mode MghUnder, (the m* Δ t) moment is decoded
The difference data in the sections p and benchmark section;CSMp(m* Δs t, Mgh) indicate mode MghUnder, (the p section that the m* Δ t) moment reconstructs
Highway traffic data.
The present invention technical concept be:Since the trip requirements of traffic participant, travel time and space have centainly
Regularity, therefore the road traffic state of relevant road segments has very strong correlation, i.e. relevant road segments on similar timing node
Road traffic state change curve have certain similitude.Therefore the present invention is based on the spatial coherences of road traffic system
Feature is compressed for relevant road segments road traffic spatial data.PCA methods are primarily based on, the road with correlation is selected
Duan Jihe;Secondly selection benchmark section and correlation section, and using its data as road traffic base value spatially
According to and training data;Then its difference data is obtained, determines the optimal threshold of road traffic space interpolation data;Further, it obtains
The real time data of other relevant road segments is taken, road traffic reference data is based on, obtains its difference data;Finally, LZW is based on to encode
And decoding technique, the compression and reconstruct of road traffic space interpolation data are realized respectively.
Beneficial effects of the present invention are mainly manifested in:Simplify and calculates, effectively improves processing speed.
Description of the drawings
Fig. 1 is the flow chart based on the PCA and LZW road traffic spatial data compression methods encoded.
Fig. 2 is the flow chart of reconstructing method.
Specific implementation mode
The invention will be further described below in conjunction with the accompanying drawings.
Referring to Figures 1 and 2, a kind of road traffic spatial data compression method based on LZW codings, includes the following steps:
1) based under same mode, spatially the highway traffic data of different sections of highway establish road traffic features refer to sequence
Row;Based on PCA methods, chooses the road traffic section with correlation and gather, process is as follows:
The road traffic historical data in s section, the acquisition in every section are extracted from reason traffic characteristic reference sequences
Data are r, and transform it into the matrix of s × r, are denoted as:Asⅹr。
Matrix AsⅹrJth row mean value be:
Based on aj, obtain AsⅹrNormalization matrix SAsⅹr:
The covariance matrix CSA of normalization matrix SA is:
The characteristic value D and feature vector V of covariance matrix CSA are obtained, then D=[λ1,λ2…λr];λ1≥λ2≥…≥λr;
Corresponding feature vector is:V=[v1,v2…vr];
Choose λ1, λ2The projection matrix VA that corresponding feature vector is constitutedr×2=[v1,v2], it is based on projection matrix and normalizing
The training matrix of change, seeks AsⅹrPrincipal component matrix A PCr×2:
APCs×2=SAsⅹr×VAr×2 (4)
Based on APCs×2, the distribution in s section is drawn on two dimensional surface, the distribution density of point indicates that corresponding road section is related
Property intensity, by correlation analysis, given threshold δ selects correlation to be more than the p+1 section of δ, and process is as follows:
Wherein, i, j indicate i-th, j section respectively, 0<i<S, 0<j<s;Indicate relevance function;
2) selection benchmark section, and using its data as road traffic reference data spatially;Extract under same mode,
The spatially historical data in other sections, as training data, based on road traffic base value under same mode, spatially
According to determining that the optimal threshold of space road traffic difference data, process are as follows:
Si(m*Δt,Mgh)=STi(m*Δt,Mgh)-SB(m*Δt,Mgh) (6)
ei(m,Mgh)=[Si(Δt,Mgh)Si(2*Δt,Mgh)...Si(m*Δt,Mgh)] (7)
pei(n,Mgh)=w (hei(m,Mgh)) (9)
pei(n,Mgh)=[Si'(1,Mgh)Si'(2,Mgh)...Si'(n,Mgh)] (10)
Wherein, Δ t is the collection period of road traffic state data;(m* Δs t) is that m-th of road traffic state data is adopted
Collect the period, 0≤m≤N, N indicate the quantity of the traffic information acquired daily;I (1≤i≤p) indicates i-th section;STi(m*Δ
T, Mgh) indicate mode MghUnder, (the highway traffic data in the moment i sections m* Δ t);SB (m* Δs t, Mgh) indicate mode MghUnder,
(the reference data in m* Δ t) moment benchmark section;Si(m* Δs t, Mgh) indicate mode MghUnder, (the training in the moment i sections m* Δ t)
The difference data of data and the reference data in benchmark section;ei(m, Mgh) indicate mode MghUnder, Δ t to (the period i sections m* Δ t)
Training data and benchmark section reference data difference data;hei(m, Mgh) indicate mode MghUnder, Δ t is to (when m* Δ t)
The difference data of the training data and the reference data in benchmark section in the sections i after section threshold process;Ei(m, Mgh) indicate mode Mgh
Under, Δ t to (the period i sections m* Δ t) choose threshold value;pei(n, Mgh) indicate mode MghUnder, Δ t to (the period i roads m* Δ t)
Result of the section with the difference data in benchmark section after LZW is encoded;Si' (n, Mgh) it is mode MghUnder, Δ t to (m* Δ t) period i
Nth data in result of the difference data in section and benchmark section after LZW is encoded;M is indicated in mode MghUnder, Δ t to (m*
The quantity in the sections i and the difference data in benchmark section before Δ t) duration compressions;N is indicated in mode MghUnder, Δ t to (m* Δs t)
Road traffic quantity after duration compression;W indicates LZW codings;Compression ratio is
3) data for extracting spatially other sections, as real time data;Mode MghUnder, based on road traffic spatially
Reference data, obtains road traffic difference data, and general expression is as follows:
MSj(m*Δt,Mgh)=SMj(m*Δt,Mgh)-SB(m*Δt,Mgh) (11)
errj(m,Mgh)=[MSj(Δt,Mgh)MSj(2*Δt,Mgh)...MSj(m*Δt,Mgh)] (12)
Wherein, j (1≤i≤p) indicates j-th strip section;SMj(m* Δs t, Mgh) indicate mode MghUnder, (the moment j roads m* Δ t)
The real time data of section;MSj(m* Δs t, Mgh) it is mode MghUnder, (the base of the real time data and benchmark section in the moment j sections m* Δ t)
The difference data of quasi- data;errj(m, Mgh) it is mode MghUnder, Δ t to (real time data in the period j sections m* Δ t) and benchmark road
The difference data of the reference data of section.
4) compression for realizing road traffic spatial data is encoded based on LZW, process is as follows:
The optimal threshold that the difference data in the sections i and benchmark section is trained is introduced into same mode Mgh, the sections j and benchmark
It in the difference data in section, is encoded in conjunction with LZW, realizes the compression in the sections j and benchmark section difference data, general expression is such as
Under:
herrsp(m*Δt,Mgh)=[herr1(m*Δt,Mgh)herr2(Δt,Mgh)...herrp'(m*Δt,Mgh)] (14)
perrp'(m*Δt,Mgh)=w (herrsp(m*Δt,Mgh)) (15)
perrp'(m*Δt,Mgh)=[MS1(m*Δt,Mgh)MS2(m*Δt,Mgh)...MSp'(m*Δt,Mgh)] (16)
Wherein, Eopt(Mgh) indicate trained optimal threshold;herrj(m* Δs t, Mgh) indicate mode MghUnder, (when m* Δ t)
Carve the difference data of the reference data in real time data and the benchmark section in the sections j after threshold process;M indicates mode MghUnder, Δ t arrives
(the quantity in the sections j and the difference data in benchmark section before m* Δ t) duration compressions;herrsp(m*Δt,Mgh) indicate mode Mgh
Under, (the magnitude-set of m* p section difference data of Δ t) moment;Perrp’(m*Δt,Mgh) indicate mode MghUnder, (m* Δs t)
The magnitude-set of difference data after p section compression of moment;MSj' (m* Δs t, Mgh) indicate mode MghUnder, (m* Δ t) moment, j
The compressed quantity of difference data in section;P ' indicates (quantity after m* Δ t) moment LZW coding;Compression ratio is:
5) LZW decoding techniques are based on, realize that the reconstruct of road traffic spatial data, process are as follows:
The difference data in p section and benchmark section is reconstructed, in conjunction with reference data, realizes that p section counts in real time
According to decompression, general expression is as follows:
dperrp(m*Δt,Mgh)=w'(perrp'(m*Δt,Mgh)) (17)
dperrj(m,Mgh)=w'(perrj(Tn,Mgh)) (18)
CSMj(m,Mgh)=SB (m, Mgh)+dperrj(m,Mgh) (19)
Wherein, w ' indicates the decoding of LZW;dperrp(m* Δs t, Mgh) indicate mode MghUnder, (the m* Δ t) moment is decoded
The difference data in the sections p and benchmark section;CSMp(m* Δs t, Mgh) indicate mode MghUnder, (the p section that the m* Δ t) moment reconstructs
Highway traffic data.
Example:A kind of road traffic spatial data compression method encoded based on PCA and LZW, is included the following steps:
1) based under same mode, spatially the highway traffic data of different sections of highway establish road traffic features refer to sequence
Row;Based on PCA methods, chooses the road traffic section with correlation and gather, process is as follows:
This experiment is with 6, Beijing section, four day June in 2011 weekend (18,19,25,26) same test point actual measurement stream
It measures data (sampling interval is 2 minutes) and is used as sample sequence, shown in road section information table 1.
Table 1
The road traffic historical data in 6 sections, the acquisition in every section are extracted from reason traffic characteristic reference sequences
Data are 720, and transform it into the matrix of s × r, i.e. s is 6, r 720, is denoted as:Asⅹr。
Matrix AsⅹrJth row mean value be:
Based on aj, obtain AsⅹrNormalization matrix SAsⅹr:
The covariance matrix CSA of normalization matrix SA is:
The characteristic value D and feature vector V of covariance matrix CSA are obtained, then D=[λ1,λ2…λr];λ1≥λ2≥…≥λr;
Corresponding feature vector is:V=[v1,v2…vr]。
Choose λ1, λ2The projection matrix VA that corresponding feature vector is constitutedr×2=[v1,v2], it is based on projection matrix and normalizing
The training matrix of change, seeks AsⅹrPrincipal component matrix A PCr×2:
APCs×2=SAsⅹr×VAr×2 (4)
Based on APCs×2, the distribution in s section is drawn on two dimensional surface, the distribution density of point indicates that corresponding road section is related
Property intensity.By correlation analysis, given threshold δ selects correlation to be more than the p+1 section of δ, and process is as follows:
Wherein, i, j indicate i-th, j section respectively, 0<i<S, 0<j<s;Indicate relevance function;
2) selection benchmark section, and using its data as road traffic reference data spatially;Extract under same mode,
The spatially historical data in other sections, as training data, based on road traffic base value under same mode, spatially
According to determining that the optimal threshold of space road traffic difference data, process are as follows:
Si(m*Δt,Mgh)=STi(m*Δt,Mgh)-SB(m*Δt,Mgh) (6)
ei(m,Mgh)=[Si(Δt,Mgh)Si(2*Δt,Mgh)...Si(m*Δt,Mgh)] (7)
pei(n,Mgh)=w (hei(m,Mgh)) (9)
pei(n,Mgh)=[Si'(1,Mgh)Si'(2,Mgh)...Si'(n,Mgh)] (10)
Wherein, Δ t is the collection period of road traffic state data;(m* Δs t) is that m-th of road traffic state data is adopted
Collect the period, 0≤m≤N, N indicate the quantity of the traffic information acquired daily;I (1≤i≤p) indicates i-th section;STi(m*Δ
T, Mgh) indicate mode MghUnder, (the highway traffic data in the moment i sections m* Δ t);SB (m* Δs t, Mgh) indicate mode MghUnder,
(the reference data in m* Δ t) moment benchmark section;Si(m* Δs t, Mgh) indicate mode MghUnder, (the training in the moment i sections m* Δ t)
The difference data of data and the reference data in benchmark section;ei(m, Mgh) indicate mode MghUnder, Δ t to (the period i sections m* Δ t)
Training data and benchmark section reference data difference data;hei(m, Mgh) indicate mode MghUnder, Δ t is to (when m* Δ t)
The difference data of the training data and the reference data in benchmark section in the sections i after section threshold process;Ei(m, Mgh) indicate mode Mgh
Under, Δ t to (the period i sections m* Δ t) choose threshold value;pei(n, Mgh) indicate mode MghUnder, Δ t to (the period i roads m* Δ t)
Result of the section with the difference data in benchmark section after LZW is encoded;Si' (n, Mgh) it is mode MghUnder, Δ t to (m* Δ t) period i
Nth data in result of the difference data in section and benchmark section after LZW is encoded;M is indicated in mode MghUnder, Δ t to (m*
The quantity in the sections i and the difference data in benchmark section before Δ t) duration compressions;N is indicated in mode MghUnder, Δ t to (m* Δs t)
Road traffic quantity after duration compression;W indicates LZW codings;Compression ratio is
3) data for extracting spatially other sections, as real time data;Mode MghUnder, based on road traffic spatially
Reference data, obtains road traffic difference data, and general expression is as follows:
MSj(m*Δt,Mgh)=SMj(m*Δt,Mgh)-SB(m*Δt,Mgh) (11)
errj(m,Mgh)=[MSj(Δt,Mgh)MSj(2*Δt,Mgh)...MSj(m*Δt,Mgh)] (12)
Wherein, j (1≤i≤p) indicates j-th strip section;SMj(m* Δs t, Mgh) indicate mode MghUnder, (the moment j roads m* Δ t)
The real time data of section;MSj(m* Δs t, Mgh) it is mode MghUnder, (the base of the real time data and benchmark section in the moment j sections m* Δ t)
The difference data of quasi- data;errj(m, Mgh) it is mode MghUnder, Δ t to (real time data in the period j sections m* Δ t) and benchmark road
The difference data of the reference data of section.
4) compression for realizing road traffic spatial data is encoded based on LZW, process is as follows:
The optimal threshold that the difference data in the sections i and benchmark section is trained is introduced into same mode Mgh, the sections j and benchmark
It in the difference data in section, is encoded in conjunction with LZW, realizes the compression in the sections j and benchmark section difference data, general expression is such as
Under:
herrsp(m*Δt,Mgh)=[herr1(m*Δt,Mgh)herr2(Δt,Mgh)...herrp'(m*Δt,Mgh)] (14)
perrp'(m*Δt,Mgh)=w (herrsp(m*Δt,Mgh)) (15)
perrp'(m*Δt,Mgh)=[MS1(m*Δt,Mgh)MS2(m*Δt,Mgh)...MSp'(m*Δt,Mgh)] (16)
Wherein, Eopt(Mgh) indicate trained optimal threshold;herrj(m* Δs t, Mgh) indicate mode MghUnder, (when m* Δ t)
Carve the difference data of the reference data in real time data and the benchmark section in the sections j after threshold process;M indicates mode MghUnder, Δ t arrives
(the quantity in the sections j and the difference data in benchmark section before m* Δ t) duration compressions;herrsp(m*Δt,Mgh) indicate mode Mgh
Under, (the magnitude-set of m* p section difference data of Δ t) moment;Perrp’(m*Δt,Mgh) indicate mode MghUnder, (m* Δs t)
The magnitude-set of difference data after p section compression of moment;MSj' (m* Δs t, Mgh) indicate mode MghUnder, (m* Δ t) moment, j
The compressed quantity of difference data in section;P ' indicates (quantity after m* Δ t) moment LZW coding;Compression ratio is:
5) LZW decoding techniques are based on, realize that the reconstruct of road traffic spatial data, process are as follows:
The difference data in p section and benchmark section is reconstructed, in conjunction with reference data, realizes that p section counts in real time
According to decompression, general expression is as follows:
dperrp(m*Δt,Mgh)=w'(perrp'(m*Δt,Mgh)) (17)
dperrj(m,Mgh)=w'(perrj(Tn,Mgh)) (18)
CSMj(m,Mgh)=SB (m, Mgh)+dperrj(m,Mgh) (19)
Wherein, w ' indicates the decoding of LZW;dperrp(m* Δs t, Mgh) indicate mode MghUnder, (the m* Δ t) moment is decoded
The difference data in the sections p and benchmark section;CSMp(m* Δs t, Mgh) indicate mode MghUnder, (the p section that the m* Δ t) moment reconstructs
Highway traffic data.
6) it is determined based on the parameter of the PCA and LZW road traffic spatial data compressions encoded
During the road traffic spatial data compression based on PCA and LZW codings, it is designed into following parameter:
SB(m*Δt)、STi(m*Δt)、Ei(m* Δs t), per, erri(m* Δs t), n, wherein Ei(m* Δs t) can by SB (m) and
Per is obtained, n and erri(m* Δs t) can be by SB (m* Δs t), STi(m*Δt),Ei(t) decisions of m* Δs, the parameter done here
The setting only general impact analysis to the road traffic spatial data compression encoded based on LZW.
Since these parameters respectively have an impact the precision of algorithm, influence of each parameter of independent analysis to arithmetic accuracy is not
It can ensure that the optimal of algorithm, therefore should consider that all parameters compress the highway traffic data simultaneously when carrying out Algorithm Analysis
Influence.
Influence of the compression alignment parameters to arithmetic accuracy is introduced to analyze:
Wherein, CRp(m,Mgh) indicate in mode MghUnder, the Δ t to (compression ratio in the p section of Δ t) periods m*;CMa(m,
Mgh) indicate in mode MghUnder, Δ t to (data amount check before the p section compression of m* Δ t) periods;CMb(m,Mgh) be expressed as in mould
State MghUnder, Δ t to (data amount check after the p section compression of m* Δ t) periods.
I.e. for different (SB (m* Δs t, Mgh), STj(m* Δs t, Mgh), Per), there are corresponding CRp(m,Mgh)。
Therefore there are following equatioies:
CRp(m* Δs t, Mgh)=f (SB (m* Δs t, Mgh), STj(m* Δs t, Mgh), Per) (21)
That is f (SB (m* Δs t, Mgh), STj(m* Δs t, Mgh), Per) and CRp(m* Δs t, Mgh) there are certain distribution relation f,
Find CRp(m* Δs t, Mgh) it is maximum when corresponding (SB (m, Mgh), STj(m, Mgh), Per), as optimized parameter sets process.Therefore
It can obtain such as drag:
Min f (SB (m* Δs t, Mgh), STj(m* Δs t, Mgh), Per) (22)
Finally (SB (m* Δs t, Mgh), STj(m* Δs t, Mgh), Per) value can by road traffic reference data and
The training of training data determines.
7) experimental result
Road traffic space reference data and training data based on same mode, acquisition optimized parameter (SB ((m* Δ t),
STj(m), Per).This experimental result is compressed mainly for the vehicle amount velocity amplitude in section.Extraction road traffic space counts in real time
According to based on LZW codings, the compression of realization road traffic space real time data.
It chooses compression ratio (CR), absolute error (AE), road is used as to percentage error (marerr), error to standard deviation (σ)
The index of road forecasting traffic flow precision, CR have been described in formula (20), remaining calculation formula difference is as follows:
Wherein
yp(m*Δt,Mgh)=CSMp(m*Δt,Mgh)-SMp(m*Δt,Mgh)
Wherein, yp(m*Δt,Mgh) indicate mode MghUnder, (p section real time data of m* Δ t) moment with reconstruct after reality
When data error amount,For mean error.
Based on above analysis, it is known that the road traffic section with correlation is HI3009b, HI3008b and HI7058b
Three sections.The data in this three sections are selected to carry out the research and application of compression algorithm.
The compression result statistical analysis of the section HI3009b, HI7058b is as shown in the following table 2,3.
Date | 18 | 19 | 25 | 26 | average |
CR | 11.80 | 10.91 | 11.08 | 9.00 | 10.70 |
AE | 11.24 | 10.94 | 12.43 | 12.05 | 11.67 |
marerr | 12.42 | 12.54 | 12.78 | 13.87 | 12.90 |
σ | 13.93 | 12.56 | 14.88 | 13.64 | 13.75 |
Table 2
Date | 18 | 19 | 25 | 26 | average |
CR | 16.74 | 16.00 | 15.65 | 11.80 | 15.05 |
AE | 6.65 | 6.93 | 6.83 | 7.41 | 6.96 |
marerr | 6.87 | 7.67 | 7.07 | 8.51 | 7.53 |
σ | 8.65 | 9.33 | 8.87 | 9.79 | 9.16 |
Table 3.
Claims (2)
1. a kind of road traffic spatial data compression method encoded based on PCA and LZW, it is characterised in that:The method includes
Following steps:
1) based under same mode, spatially the highway traffic data of different sections of highway establishes road traffic features reference sequences;Base
In PCA methods, chooses the road traffic section with correlation and gather, process is as follows:
The road traffic historical data in s section, the gathered data in every section are extracted from reason traffic characteristic reference sequences
For r, and the matrix of s × r is transformed it into, is denoted as:Asⅹr;
Matrix AsⅹrJth row mean value be:
Based on aj, obtain AsⅹrNormalization matrix SAsⅹr:
The covariance matrix CSA of normalization matrix SA is:
The characteristic value D and feature vector V of covariance matrix CSA are obtained, then D=[λ1,λ2…λr];λ1≥λ2≥…≥λr;It is corresponding
Feature vector be:V=[v1,v2…vr];
Choose λ1, λ2The projection matrix VA that corresponding feature vector is constitutedr×2=[v1,v2], based on projection matrix and normalized
Training matrix seeks AsⅹrPrincipal component matrix A PCr×2:
APCs×2=SAsⅹr×VAr×2 (4)
Based on APCs×2, the distribution in s section is drawn on two dimensional surface, the distribution density of point indicates that corresponding road section correlation is strong
Degree, by correlation analysis, given threshold δ selects correlation to be more than the p+1 section of δ, and process is as follows:
Wherein, i, j indicate i-th, j section respectively, 0<i<S, 0<j<s;Indicate relevance function;
2) selection benchmark section, and using its data as road traffic reference data spatially;It extracts under same mode, space
The historical data in upper other sections, as training data, based on road traffic reference data under same mode, spatially, really
Determine the optimal threshold of space road traffic difference data, process is as follows:
Si(m*Δt,Mgh)=STi(m*Δt,Mgh)-SB(m*Δt,Mgh) (6)
ei(m,Mgh)=[Si(Δt,Mgh)Si(2*Δt,Mgh)...Si(m*Δt,Mgh)] (7)
pei(n,Mgh)=w (hei(m,Mgh)) (9)
pei(n,Mgh)=[Si'(1,Mgh)Si'(2,Mgh)...Si'(n,Mgh)] (10)
Wherein, Δ t is the collection period of road traffic state data;(m* Δs t) is m-th of road traffic state data acquisition week
Phase, 0≤m≤N, N indicate the quantity of the traffic information acquired daily;I (1≤i≤p) indicates i-th section;STi(m* Δ t,
Mgh) indicate mode MghUnder, (the highway traffic data in the moment i sections m* Δ t);SB (m* Δs t, Mgh) indicate mode MghUnder, (m*
The reference data in Δ t) moment benchmark section;Si(m* Δs t, Mgh) indicate mode MghUnder, (the training number in the moment i sections m* Δ t)
According to the difference data of the reference data with benchmark section;ei(m, Mgh) indicate mode MghUnder, Δ t is to (the period i sections m* Δ t)
The difference data of training data and the reference data in benchmark section;hei(m, Mgh) indicate mode MghUnder, Δ t to (m* Δ t) periods
The difference data of the training data in the sections i and the reference data in benchmark section after threshold process;Ei(m, Mgh) indicate mode MghUnder,
Δ t to (the period i sections m* Δ t) choose threshold value;pei(n, Mgh) indicate mode MghUnder, Δ t to (the period i sections Δ t) m* with
Result of the difference data in benchmark section after LZW is encoded;Si' (n, Mgh) it is mode MghUnder, Δ t to (the period i sections m* Δ t)
With nth data in the result of the difference data in benchmark section after LZW is encoded;M is indicated in mode MghUnder, Δ t to (m* Δs t)
The quantity in the sections i and the difference data in benchmark section before duration compression;N is indicated in mode MghUnder, Δ t to (m* Δ t) periods
Compressed road traffic quantity;W indicates LZW codings;Compression ratio is
3) data for extracting spatially other sections, as real time data;Mode MghUnder, based on road traffic benchmark spatially
Data, obtain road traffic difference data, and general expression is as follows:
MSj(m*Δt,Mgh)=SMj(m*Δt,Mgh)-SB(m*Δt,Mgh) (11)
errj(m,Mgh)=[MSj(Δt,Mgh)MSj(2*Δt,Mgh)...MSj(m*Δt,Mgh)] (12)
Wherein, j (1≤i≤p) indicates j-th strip section;SMj(m* Δs t, Mgh) indicate mode MghUnder, (the moment j sections m* Δ t)
Real time data;MSj(m* Δs t, Mgh) it is mode MghUnder, (the base value of the real time data and benchmark section in the moment j sections m* Δ t)
According to difference data;errj(m, Mgh) it is mode MghUnder, Δ t to (real time data in the period j sections m* Δ t) and benchmark section
The difference data of reference data;
4) compression for realizing road traffic spatial data is encoded based on LZW, process is as follows:
The optimal threshold that the difference data in the sections i and benchmark section is trained is introduced into same mode Mgh, the sections j and benchmark section
Difference data in, encoded in conjunction with LZW, realize the compression in the sections j and benchmark section difference data, general expression is as follows:
herrsp(m*Δt,Mgh)=[herr1(m*Δt,Mgh)herr2(Δt,Mgh)...herrp'(m*Δt,Mgh)] (14)
perrp'(m*Δt,Mgh)=w (herrsp(m*Δt,Mgh)) (15)
perrp'(m*Δt,Mgh)=[MS1(m*Δt,Mgh)MS2(m*Δt,Mgh)...MSp'(m*Δt,Mgh)] (16)
Wherein, Eopt(Mgh) indicate trained optimal threshold;herrj(m* Δs t, Mgh) indicate mode MghUnder, (m* Δ t) moment thresholds
The difference data of the real time data and the reference data in benchmark section in the sections j after value processing;M indicates mode MghUnder, Δ t to (m*
The quantity in the sections j and the difference data in benchmark section before Δ t) duration compressions;herrsp(m*Δt,Mgh) indicate mode MghUnder,
(the magnitude-set of m* p section difference data of Δ t) moment;Perrp’(m*Δt,Mgh) indicate mode MghUnder, (the m* Δ t) moment
The magnitude-set of difference data after p section compression;MSj' (m* Δs t, Mgh) indicate mode MghUnder, (m* Δ t) moment, the sections j
The compressed quantity of difference data;P ' indicates (quantity after m* Δ t) moment LZW coding;Compression ratio is:
2. the road traffic spatial data compression method encoded as described in claim 1 based on PCA and LZW, it is characterised in that:
The compression method further includes following steps:
5) LZW decoding techniques are based on, realize that the reconstruct of road traffic spatial data, process are as follows:
The difference data in p section and benchmark section is reconstructed, in conjunction with reference data, realizes p section real time data
Decompression, general expression are as follows:
dperrp(m*Δt,Mgh)=w'(perrp'(m*Δt,Mgh)) (17)
dperrj(m,Mgh)=w'(perrj(Tn,Mgh)) (18)
CSMj(m,Mgh)=SB (m, Mgh)+dperrj(m,Mgh) (19)
Wherein, w ' indicates the decoding of LZW;dperrp(m* Δs t, Mgh) indicate mode MghUnder, (Δ t) the moment decoded roads p m*
The difference data of section and benchmark section;CSMp(m* Δs t, Mgh) indicate mode MghUnder, (the p section that the m* Δ t) moment reconstructs
Highway traffic data.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610236010.4A CN105788261B (en) | 2016-04-15 | 2016-04-15 | A kind of road traffic spatial data compression method encoded based on PCA and LZW |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610236010.4A CN105788261B (en) | 2016-04-15 | 2016-04-15 | A kind of road traffic spatial data compression method encoded based on PCA and LZW |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105788261A CN105788261A (en) | 2016-07-20 |
CN105788261B true CN105788261B (en) | 2018-09-21 |
Family
ID=56397714
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610236010.4A Active CN105788261B (en) | 2016-04-15 | 2016-04-15 | A kind of road traffic spatial data compression method encoded based on PCA and LZW |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105788261B (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108629410B (en) * | 2018-04-28 | 2021-01-22 | 中国科学院计算技术研究所 | Neural network processing method based on principal component analysis dimension reduction and/or dimension increase |
CN111369794B (en) * | 2020-02-28 | 2022-01-25 | 腾讯科技(深圳)有限公司 | Method, device and equipment for determining traffic participation information and storage medium |
CN112185122B (en) * | 2020-10-28 | 2022-05-20 | 中国地质大学(武汉) | Self-adaptive neighborhood determination method combining spatial structure and attribute information |
CN115964347B (en) * | 2023-03-16 | 2023-05-16 | 菏泽市产品检验检测研究院 | Intelligent storage method for data of market supervision and monitoring center |
CN116489378B (en) * | 2023-06-21 | 2023-08-25 | 昇显微电子(苏州)股份有限公司 | PCA-based 3D LUT data compression and decompression method and system |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6789052B1 (en) * | 2000-10-24 | 2004-09-07 | Advanced Micro Devices, Inc. | Method of using control models for data compression |
CN101848529A (en) * | 2010-03-31 | 2010-09-29 | 中国地质大学(武汉) | Method for compressing multiple principle component analysis data of wireless sensor network |
KR20110115281A (en) * | 2010-04-15 | 2011-10-21 | 성균관대학교산학협력단 | Partitioning method for high dimensional data |
CN103501438A (en) * | 2013-09-18 | 2014-01-08 | 浙江大学 | Content self-adaptation image compression method based on PCA |
CN105427583A (en) * | 2015-11-27 | 2016-03-23 | 浙江工业大学 | LZW-coding-based road traffic data compression method |
CN105469601A (en) * | 2015-12-09 | 2016-04-06 | 浙江工业大学 | A road traffic space data compression method based on LZW coding |
-
2016
- 2016-04-15 CN CN201610236010.4A patent/CN105788261B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6789052B1 (en) * | 2000-10-24 | 2004-09-07 | Advanced Micro Devices, Inc. | Method of using control models for data compression |
CN101848529A (en) * | 2010-03-31 | 2010-09-29 | 中国地质大学(武汉) | Method for compressing multiple principle component analysis data of wireless sensor network |
KR20110115281A (en) * | 2010-04-15 | 2011-10-21 | 성균관대학교산학협력단 | Partitioning method for high dimensional data |
CN103501438A (en) * | 2013-09-18 | 2014-01-08 | 浙江大学 | Content self-adaptation image compression method based on PCA |
CN105427583A (en) * | 2015-11-27 | 2016-03-23 | 浙江工业大学 | LZW-coding-based road traffic data compression method |
CN105469601A (en) * | 2015-12-09 | 2016-04-06 | 浙江工业大学 | A road traffic space data compression method based on LZW coding |
Non-Patent Citations (1)
Title |
---|
基于PCA和ICA的交通流量数据压缩方法比较研究;赵志强 等;《公路交通科技》;20081130;第25卷(第11期);第109-113、118页 * |
Also Published As
Publication number | Publication date |
---|---|
CN105788261A (en) | 2016-07-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105788261B (en) | A kind of road traffic spatial data compression method encoded based on PCA and LZW | |
CN103116148B (en) | Inversion method of nuclear magnetic resonance two-dimensional spectrum | |
CN108564129A (en) | A kind of track data sorting technique based on generation confrontation network | |
CN101493890B (en) | Dynamic vision caution region extracting method based on characteristic | |
CN112598438A (en) | Outdoor advertisement recommendation system and method based on large-scale user portrait | |
CN111899759B (en) | Method, device, equipment and medium for pre-training and model training of audio data | |
CN105260794A (en) | Load predicting method of cloud data center | |
CN105044794A (en) | Nuclear magnetic resonance echo data compression method and device | |
CN102148987A (en) | Compressed sensing image reconstructing method based on prior model and 10 norms | |
CN104298977A (en) | Low-order representing human body behavior identification method based on irrelevance constraint | |
CN112906911A (en) | Model training method for federal learning | |
CN110289081A (en) | The epilepsy detection method of depth network stack model adaptation weighted feature fusion | |
Wu et al. | Image denoising with rectified linear units | |
CN111144281A (en) | Urban rail transit OD passenger flow estimation method based on machine learning | |
CN106777159A (en) | A kind of video clip retrieval and localization method based on content | |
CN104657466A (en) | Method and device for identifying user interest based on forum post features | |
CN105469601B (en) | A kind of road traffic spatial data compression method based on LZW codings | |
CN106375754A (en) | No-reference video quality evaluation method based on visual stimulation attenuation characteristic | |
CN105893331A (en) | Principal component analysis algorithm based compression method for data on road traffic time series | |
CN105427583B (en) | A kind of highway traffic data compression method encoded based on LZW | |
CN117349657A (en) | Distributed data acquisition module and monitoring system for hydraulic engineering environment monitoring | |
CN114239757B (en) | Denoising method and system for electromagnetic time series data | |
Xu et al. | Relocated i-frames detection in h. 264 double compressed videos based on genetic-cnn | |
CN103778445B (en) | Cold-rolled strip steel surface defect reason analyzing method and system | |
CN102184210A (en) | Stratified decision tree constructing method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |