CN109214422A - Parking data method for repairing and mending, device, equipment and storage medium based on DCGAN - Google Patents
Parking data method for repairing and mending, device, equipment and storage medium based on DCGAN Download PDFInfo
- Publication number
- CN109214422A CN109214422A CN201810869067.7A CN201810869067A CN109214422A CN 109214422 A CN109214422 A CN 109214422A CN 201810869067 A CN201810869067 A CN 201810869067A CN 109214422 A CN109214422 A CN 109214422A
- Authority
- CN
- China
- Prior art keywords
- parking lot
- parking
- empty wagons
- data
- cluster
- 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
- G06F18/10—Pre-processing; Data cleansing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
-
- 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
- G08G1/0137—Measuring and analyzing of parameters relative to traffic conditions for specific applications
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Chemical & Material Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Analytical Chemistry (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Traffic Control Systems (AREA)
Abstract
Applicable data repairing technique of the present invention field, provide a kind of parking data method for repairing and mending based on DCGAN, device, equipment and storage medium, this method comprises: clustering in the parking lot in urban parking area set for parking lot cluster, parking lot includes sample parking lot and parking lot to be repaired, corresponding empty wagons rate curve image is generated according to the parking data in sample parking lot, according to the empty wagons rate curve image in sample parking lot in the cluster of parking lot, production of the training based on depth convolution fights network, to generate the empty wagons rate curve emulating image of each parking lot cluster, these empty wagons rate curve emulating images are handled and are mapped as corresponding one-dimensional empty wagons rate data, to carry out data modification to the parking lot to be repaired in each parking lot cluster according to these one-dimensional empty wagons rate data, to know in the priori for not depending on parking lot to be repaired More accurately data modification is carried out to parking lot to be repaired when knowledge, the parking data repairing of City-level is realized, has saved the data acquisition cost in parking lot.
Description
Technical field
The invention belongs to data mending technique field more particularly to a kind of parking data method for repairing and mending based on DCGAN, dress
It sets, equipment and storage medium.
Background technique
Pith of the communications and transportation as modern economy, but along with the development of social economy, growing traffic
Contradiction between demand and urban transportation load capacity is increasingly prominent.Wherein, than more prominent, intelligent parking system is parking difficulty problem
One of the most effectual way for solving the problems, such as this, reducing down time cost.
Any machine learning algorithm requires great amount of samples data as support, and the research and development of intelligent parking system also need
The parking data in a large amount of parking lots in certain area.Obtain parking data conventional method include search public data collection or
It is acquired by purchasing and installing sensor, these conventional method economic costs and time cost are higher, are difficult to obtain city
The parking data in all parking lots in city causes most of parking lot in city to lack parking data.In addition, construction damage, line
Road failure and processing mistake etc., but also part parking lot has that parking data is incomplete.Therefore, stop in face of on a large scale
The case where car data lacks is improved the integrality of parking data in city scope, is had how under the premise of economic cost is controllable
Effect property and predictability, provide effective support for relevant machine learning algorithm and traffic analysis model, are current intelligent parkings
Consider the problems of needed for algorithm design and training.
Model caused by the above problem can be expressed as because of shortage of data is distorted, and is generally enhanced using data in machine learning
Technology handles shortage of data problem, i.e., the data on conversational implication generate (or repairing) technology.The more recent application of this technology
Achievement is mainly reflected on the Style Transfer and reparation of image, and the research in traffic data repairing is relatively deficient.Traffic number at present
According to repairing field research be mainly some interpolation methods, such as the data modification method based on Lagrange's interpolation, be based on ox
The characteristics of data modification method of interpolation method and data modification method based on piecewise linear interpolation method, these methods be need it is certain
Priori knowledge, be typically used to have the traffic data repairing field of certain historical data.However, due to economic cost, installation
Reasons, the history parking data in most parking lots such as construction and economic property right are difficult to obtain.
Summary of the invention
The purpose of the present invention is to provide a kind of parking data method for repairing and mending, device, equipment and storage based on DCGAN
Medium, it is intended to solve the higher cost and available data that acquire all Parking data in city scope in the prior art
The method for repairing and mending problem larger to parking lot history parking data dependence.
On the one hand, the present invention provides a kind of parking data method for repairing and mending based on DCGAN, the method includes following
Step:
Urban parking area set and urban geography interest point set gathered in advance are obtained, in the urban parking area set
Parking lot include sample parking lot and parking lot to be repaired；
The parking lot is clustered for multiple parking lot clusters, and according to the sample according to the urban geography interest point set
The parking data in this parking lot generates the empty wagons rate curve image in the sample parking lot；
It constructs the production based on depth convolution and fights network, according to the sky in sample parking lot described in the parking lot cluster
Vehicle rate curve image is trained the production confrontation network based on depth convolution, to generate each parking lot
The corresponding empty wagons rate curve emulating image of cluster；
The corresponding empty wagons rate curve emulating image of each parking lot cluster is handled and is mapped as corresponding one-dimensional
Empty wagons rate data, according to the corresponding one-dimensional empty wagons rate data of each parking lot cluster, to the institute in each parking lot cluster
It states parking lot to be repaired and carries out parking data repairing.
On the other hand, the present invention provides a kind of parking data repair apparatus based on DCGAN, described device include:
Data capture unit is acquired, for obtaining urban parking area set gathered in advance and urban geography interest point set
It closes, the parking lot in the urban parking area set includes sample parking lot and parking lot to be repaired；
Data processing unit, for being clustered in the parking lot for multiple parkings according to the urban geography interest point set
Cluster, and generate according to the parking data in the sample parking lot empty wagons rate curve image in the sample parking lot；
Network struction training unit fights network for constructing the production based on depth convolution, according to the parking lot
The empty wagons rate curve image of sample parking lot described in cluster instructs the production confrontation network based on depth convolution
Practice, to generate the corresponding empty wagons rate curve emulating image of each parking lot cluster；And
Parking data repairs unit, to the corresponding empty wagons rate curve emulating image of each parking lot cluster
Corresponding one-dimensional empty wagons rate data are managed and are mapped as, according to the corresponding one-dimensional empty wagons rate data of each parking lot cluster, to institute
The parking lot to be repaired stated in each parking lot cluster carries out parking data repairing.
On the other hand, the present invention also provides a kind of calculating equipment, including memory, processor and it is stored in described deposit
In reservoir and the computer program that can run on the processor, the processor are realized such as when executing the computer program
Step described in the above-mentioned parking data method for repairing and mending based on DCGAN.
On the other hand, the present invention also provides a kind of computer readable storage medium, the computer readable storage mediums
It is stored with computer program, such as the above-mentioned parking data based on DCGAN is realized when the computer program is executed by processor
Step described in method for repairing and mending.
The present invention acquires urban parking area set and urban geography interest point set in advance, stops in urban parking area set
Parking lot includes sample parking lot and parking lot to be repaired, these parking lots cluster is stopped according to urban geography interest point set
Field cluster, and according to the empty wagons rate curve image in the parking data in sample parking lot generation sample parking lot, according in the cluster of parking lot
The empty wagons rate curve image in sample parking lot is trained the production confrontation network based on depth convolution, by training
The production confrontation network based on depth convolution generate the corresponding empty wagons rate curve emulating image of each parking lot cluster, to these
Empty wagons rate curve emulating image is handled and is mapped as corresponding one-dimensional empty wagons rate data, according to these one-dimensional space rate numbers
Parking data repairing is carried out according to the parking lot to be repaired in each parking lot cluster, thus in the elder generation for not depending on parking lot to be repaired
More accurately data modification is carried out to parking lot to be repaired when testing knowledge, realizes the parking data repairing of City-level, Jin Erjie
The about data acquisition cost in parking lot.
Detailed description of the invention
Fig. 1 is the implementation flow chart for the parking data method for repairing and mending based on DCGAN that the embodiment of the present invention one provides；
Fig. 2 is the structural schematic diagram of the parking data repair apparatus provided by Embodiment 2 of the present invention based on DCGAN；
Fig. 3 is the preferred structure signal of the parking data repair apparatus provided by Embodiment 2 of the present invention based on DCGAN
Figure；And
Fig. 4 is the structural schematic diagram for the calculating equipment that the embodiment of the present invention three provides.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
Specific implementation of the invention is described in detail below in conjunction with specific embodiment:
Embodiment one:
Fig. 1 shows the implementation process of the parking data method for repairing and mending based on DCGAN of the offer of the embodiment of the present invention one,
For ease of description, only parts related to embodiments of the present invention are shown, and details are as follows:
In step s101, it obtains urban parking area set gathered in advance and urban geography interest point set, city is stopped
Parking lot in the set of parking lot includes sample parking lot and parking lot to be repaired.
The embodiment of the present invention is suitable for data modification platform, system and equipment.In urban parking area set gathered in advance
Relevant information comprising multiple parking lots and each parking lot in city has the parking for lacking parking data in these parking lots
, there is the parking lot for having no lack of parking data, also in order to support the research and development of intelligent parking system to need to shortage parking data
Parking lot carries out parking data repairing, therefore for ease of description, the parking lot for lacking parking data is known as parking to be repaired
, the parking lot for having no lack of parking data is known as sample parking lot.The relevant information in parking lot to be repaired may include to be repaired
The scale in geographical location and to be repaired parking lot of the parking lot in city, the relevant information in sample parking lot may include that sample stops
Geographical location, the scale in sample parking lot and the parking data in sample parking lot of the parking lot in city.
It in embodiments of the present invention, include geographic interest point in city in urban geography interest point set gathered in advance
The geographical location of (Point of Interest, abbreviation POI, such as parking lot, market, station, school), can pass through crawler journey
Sequence obtains these information in map web site, and geographical location can be latitude and longitude information.
In step s 102, parking lot is clustered as multiple parking lot clusters according to urban geography interest point set, and according to
The parking data in sample parking lot generates the empty wagons rate curve image in sample parking lot.
In embodiments of the present invention, people often find the parking lot nearest from destination in life, if recently
Parking lot there is no parking stall just and can consider the parking lot of distance a little further, but the distance in the parking lot of final choice and destination
Will not infinity, that is, exist an acceptable maximum distance of people, the maximum distance is known as tolerance radius herein.Cause
This forms a circle using parking lot as the center of circle and by radius of circle of tolerance radius, and the POI in the circle will be to parking lot
Parking data have an impact, the influence that the POI outside circle generates Parking data is negligible, i.e., it is believed that around
The similar two Parking data of POI number of types are also likely to similar.
Preferably, it when being clustered in parking lot for multiple parking lot clusters according to urban geography interest point set, presets
The value of tolerance radius forms a circle using parking lot as the center of circle and by radius of circle of tolerance radius, from urban geography interest
Geographical location is obtained in point set appears in the POI in the circle, the POI around these POI, that is, parking lots, according to each parking lot
Around POI construct the corresponding space characteristics in each parking lot, if main POI has n kind (or urban geography in current city
N kind POI is acquired in interest point set), then the corresponding space characteristics in each parking lot are the feature vector of n dimension, according to
Each corresponding space characteristics in parking lot and preset density-based spatial clustering algorithm (Density-Based Spatial
Clustering of Applications with Noise, abbreviation DBSCAN), to all parkings in urban parking area set
Field is clustered, and multiple clusters is obtained, to effectively improve the classifying quality in parking lot.For ease of description, will gather herein
The cluster that class obtains is known as parking lot cluster.
It is further preferred that according to the corresponding space characteristics in each parking lot and DBSCAN in urban parking area set
When all parking lots are clustered, DBSCAN is determined according to the data scale of the formed data set of the space characteristics in all parking lots
In core point threshold value MinPts, according between the parking lot randomly selected in urban parking area set Euclidean distance it is equal
Value, determines the field radius ε of density in DBSCAN, to effectively improve the classifying quality in parking lot.
As illustratively, during specific experiment, 10 parking lots are randomly selected from 310 parking lots, 7 is calculated and obtains
The mean value of these parking lot Euclidean distances of (number of types of POI) dimension space is 22, the neck for being set as density in DBSCAN for 22
Domain radius ε, takes 49 for the core point threshold value MinPts in DBSCAN, by this 310 parking lots clusters is 6 classes by DBSCAN, i.e., and 6
A parking lot cluster.
In embodiments of the present invention, in the empty wagons rate curve for generating sample parking lot according to the parking data in sample parking lot
When image, the parking data in sample parking lot is the accurate time information in vehicles while passing sample parking lot, according to vehicles while passing sample
Sample parking lot can be obtained in the parking stall quantity of free time different moments in the accurate time information in this parking lot, it is contemplated that difference is stopped
The scale in parking lot may different (i.e. parking stall total quantity may be different), can to sample parking lot free time different moments parking
Bit quantity is normalized, and the idle parking stall rate for obtaining sample parking lot different moments herein will be empty in order to succinctly describe
Not busy parking stall rate is known as empty wagons rate.
In embodiments of the present invention, according to pre-set parameter of curve and sampled point number, sample parking lot is calculated not
The corresponding two-dimensional coordinate of empty wagons rate in the same time is made of the corresponding empty wagons rate curve in sample parking lot these two-dimensional coordinates, into
And obtain the empty wagons rate curve image in sample parking lot.
Preferably, sample parking lot moment tpEmpty wagons rate beSample parking lot moment tpEmpty wagons rate pair
The two-dimensional coordinate answered is Wherein, the time series in sample parking lot is expressed as It is sample parking lot in moment tpIdle parking stall quantity, Total be sample parking lot in
Parking stall sum, L are the length of curve in parameter of curve, and H is the height of curve in parameter of curve, and J is sampled point number.
Preferably due to which the POI in different parking lots must be similar in same parking lot cluster after clustering, but same after cluster
The parking data in different parking lots is not necessarily similar in one parking lot cluster, therefore passes through Pearson correlation coefficient after end of clustering
The similarity degree of different samples parking lot empty wagons rate curve in same parking lot cluster is calculated, it is different in same parking lot cluster to judge
Whether parking lot empty wagons rate curve is similar.For example, the Pearson correlation coefficient when two sample parking lot empty wagons rate curves is greater than
0.5, it is believed that the parking data in two sample parking lots has strong correlation, if stopping in a parking lot cluster with strong correlation
Parking lot permutation and combination quantity is more than 70% of all parking lot permutation and combination numbers in the parking lot cluster, then it is assumed that in the parking lot cluster
The parking data in different parking lots is similar.
In step s 103, production of the building based on depth convolution fights network, is stopped according to sample in the cluster of parking lot
The empty wagons rate curve image of field is trained the production confrontation network based on depth convolution, to generate each parking lot cluster
Corresponding empty wagons rate curve emulating image.
In embodiments of the present invention, according to preset convolutional neural networks (the Convolutional Neural for having supervision
Networks, abbreviation CNN) to the confrontation of preset production network (Generative Adversarial Networks, abbreviation
GAN it) is extended, fights network (Deep Convolutional Generative to obtain the production based on depth convolution
Adversarial Networks, abbreviation DCGAN) at this to have supervision convolutional neural networks and production confrontation network net
Network layers number and every layer of number of unit are without limiting.The DCGAN of building is compared to the GAN before extension, in model structure
Main change has: pond layer is substituted with convolutional layer, wherein the pond layer in discriminator is substituted with the convolutional layer with step-length, is generated
The convolutional layer of pond layer micro-stepping width in device substitutes；All using batch standardization on discriminator and generator；Remove full connection
Layer.Preferably, the output layer of generator is connected to the filter verified for normality in DCGAN, and only output meets normal state
Property verification empty wagons rate curve emulating image, to guarantee the effect of empty wagons rate curve emulating image that generator in DCGAN generates.
It is further preferred that the filter for normality verification uses up to Goss and mentions promise algorithm (D'Agostino-Pearson
Test), to improve the effect for carrying out normality verification to empty wagons rate curve emulating image.
In embodiments of the present invention, the empty wagons rate curve image in sample parking lot in the cluster of parking lot is defeated as training data
Enter in DCGAN, DCGAN is trained, and the corresponding empty wagons rate curve of the parking lot cluster is generated by trained DCGAN and is emulated
(the corresponding empty wagons rate curve emulating image in parking lot to be repaired i.e. in the parking lot cluster will here for convenient for distinguishing for image
The empty wagons rate curve image that DCGAN is generated is known as empty wagons rate curve emulating image), so that it is corresponding to produce each parking lot cluster
Empty wagons rate curve emulating image.
In step S104, the corresponding empty wagons rate curve emulating image of each parking lot cluster is handled and is mapped as phase
The one-dimensional empty wagons rate data answered, according to the corresponding one-dimensional empty wagons rate data of each parking lot cluster, in each parking lot cluster to
It repairs parking lot and carries out parking data repairing.
In embodiments of the present invention, after obtaining the corresponding empty wagons rate curve emulating image of each parking lot cluster, to empty wagons
Rate curve image carries out the image processing operations such as gray processing, binaryzation and noise reduction process, will treated empty wagons rate curve graph
As being mapped as corresponding one-dimensional empty wagons rate data (time series of empty wagons rate).According to the corresponding one-dimensional empty wagons rate number of parking lot cluster
According to the scale with parking lot to be repaired in the parking lot cluster, parking lot different moments to be repaired in the parking lot cluster can be calculated
Free parking space quantity, and then realize in each parking lot cluster each parking lot to be repaired parking data repair.It is inciting somebody to action
When empty wagons rate curve image is mapped as corresponding one-dimensional empty wagons rate data that treated, when above-mentioned sample parking lot difference can be used
The empty wagons rate at quarter to empty wagons rate curve image mapping relations reverse-power realize.
Preferably, the formula of gray processing is carried out to empty wagons rate curve emulating image are as follows:
R2=G2=B2=R1*a1+G1*a2+B1*a3, wherein R2、G2And B2For empty wagons rate curve emulating image gray processing it
Rgb value afterwards, R1、G1And B1For the rgb value before empty wagons rate curve emulating image gray processing, a1、a2And a3For preset gray scale
Change parameter.
Preferably, when carrying out binaryzation to the empty wagons rate curve emulating image after gray processing, a threshold value is preset,
The each pixel for traversing empty wagons rate curve emulating image sets white for the pixel that pixel value is more than the threshold value, otherwise
It is set as black, to improve the treatment effect of empty wagons rate curve emulating image.
Preferably, include that the removal of burr point and outlier remove to the noise reduction process of empty wagons rate curve emulating image, consider
Into empty wagons rate curve emulating image the burr point of curve be as pixel it is overstocked caused by, need to reduce pixel point density,
Therefore it is made a return journey flash removed point using mean filter, to improve the treatment effect of burr point on empty wagons rate curve emulating image.
It is further preferred that removing the formula of flash removed point using mean filter are as follows:
Wherein, M indicates preset filtering window size, and f (w, e) is to remove flash removed
Empty wagons rate curve emulating image before point, f ' (x, y) is the empty wagons rate curve emulating image after removing flash removed point.
Preferably, when carrying out outlier removal to empty wagons rate curve emulating image, using the abnormality detection based on distance
Algorithm detects the outlier on empty wagons rate curve emulating image, to improve outlier detection effect.It is further preferred that based on away from
From Outlier Detection Algorithm be local outlier factor LOF algorithm (Local Outlier Factor), to improve outlier detection
Effect.
In embodiments of the present invention, parking lot is clustered as parking lot cluster, is generated according to the parking data in sample parking lot
The empty wagons rate curve image in sample parking lot, according to the empty wagons rate curve image in sample parking lot in the cluster of parking lot to based on depth
The production confrontation network of convolution is trained, and is generated often with fighting network by the trained production based on depth convolution
The corresponding empty wagons rate curve emulating image of a parking lot cluster handles these empty wagons rate curve emulating images and is mapped as pair
The one-dimensional empty wagons rate data answered, to be carried out according to these one-dimensional space rate data to the parking lot to be repaired in each parking lot cluster
Parking data repairing, to be carried out more accurately when not depending on the priori knowledge in parking lot to be repaired to parking lot to be repaired
Data modification, realizes the parking data repairing of City-level, and then has saved the data acquisition cost in parking lot.
Embodiment two:
Fig. 2 shows the embodiment of the present invention three provide the parking data repair apparatus based on DCGAN structure, in order to
Convenient for explanation, only parts related to embodiments of the present invention are shown, including:
Data capture unit 21 is acquired, for obtaining urban parking area set gathered in advance and urban geography interest point set
It closes, the parking lot in urban parking area set includes sample parking lot and parking lot to be repaired.
In embodiments of the present invention, comprising multiple parking lots in city and every in urban parking area set gathered in advance
The relevant information in a parking lot, these parking lots include parking lot to be repaired and sample parking lot.The correlation in parking lot to be repaired
Information may include the scale in geographical location and to be repaired parking lot of the parking lot to be repaired in city, the correlation in sample parking lot
Information may include geographical location, the scale in sample parking lot and the parking number in sample parking lot of the sample parking lot in city
According to.
It in embodiments of the present invention, include geographic interest point in city in urban geography interest point set gathered in advance
(POI) geographical location, can obtain these information by crawlers in map web site, and geographical location can believe for longitude and latitude
Breath.
Data processing unit 22, for parking lot to be clustered as multiple parking lot clusters according to urban geography interest point set,
And the empty wagons rate curve image in sample parking lot is generated according to the parking data in sample parking lot.
In embodiments of the present invention, people often find the parking lot nearest from destination in life, if recently
Parking lot there is no parking stall just and can consider the parking lot of distance a little further, but the distance in the parking lot of final choice and destination
Will not infinity, that is, exist an acceptable maximum distance of people, the maximum distance is known as tolerance radius herein.Cause
This forms a circle using parking lot as the center of circle and by radius of circle of tolerance radius, and the POI in the circle will be to parking lot
Parking data have an impact, the influence that the POI outside circle generates Parking data is negligible, i.e., it is believed that around
The similar two Parking data of POI number of types are also likely to similar.
Preferably, it when being clustered in parking lot for multiple parking lot clusters according to urban geography interest point set, presets
The value of tolerance radius forms a circle using parking lot as the center of circle and by radius of circle of tolerance radius, from urban geography interest
Geographical location is obtained in point set appears in the POI in the circle, the POI around these POI, that is, parking lots, according to each parking lot
The POI of surrounding constructs the corresponding space characteristics in each parking lot, if main POI has n kind, each parking lot in current city
Corresponding space characteristics are the feature vector of n dimension, according to corresponding space characteristics in each parking lot and preset based on close
The Spatial Clustering (DBSCAN) of degree, clusters all parking lots in urban parking area set, obtains multiple clusters, thus
Effectively improve the classifying quality in parking lot.For ease of description, the cluster that cluster obtains is known as parking lot cluster herein.
It is further preferred that according to the corresponding space characteristics in each parking lot and DBSCAN in urban parking area set
When all parking lots are clustered, DBSCAN is determined according to the data scale of the formed data set of the space characteristics in all parking lots
In core point threshold value MinPts, according between the parking lot randomly selected in urban parking area set Euclidean distance it is equal
Value, determines the field radius ε of density in DBSCAN, to effectively improve the classifying quality in parking lot.
In embodiments of the present invention, in the empty wagons rate curve for generating sample parking lot according to the parking data in sample parking lot
When image, the parking data in sample parking lot is the accurate time information in vehicles while passing sample parking lot, according to vehicles while passing sample
Sample parking lot can be obtained in the parking stall quantity of free time different moments in the accurate time information in this parking lot, it is contemplated that difference is stopped
The scale in parking lot may be different, can be normalized, obtain in the parking stall quantity of free time different moments to sample parking lot
To the empty wagons rate of sample parking lot different moments.
In embodiments of the present invention, according to pre-set parameter of curve and sampled point number, sample parking lot is calculated not
The corresponding two-dimensional coordinate of empty wagons rate in the same time is made of the corresponding empty wagons rate curve in sample parking lot these two-dimensional coordinates, into
And obtain the empty wagons rate curve image in sample parking lot.
Preferably, sample parking lot moment tpEmpty wagons rate beSample parking lot moment tpEmpty wagons rate pair
The two-dimensional coordinate answered is Wherein, the time series in sample parking lot is expressed as It is sample parking lot in moment tpIdle parking stall quantity, Total be sample parking lot in
Parking stall sum, L are the length of curve in parameter of curve, and H is the height of curve in parameter of curve, and J is sampled point number.
Preferably due to which the POI in different parking lots must be similar in same parking lot cluster after clustering, but same after cluster
The parking data in different parking lots is not necessarily similar in one parking lot cluster, therefore passes through Pearson correlation coefficient after end of clustering
The similarity degree of different samples parking lot empty wagons rate curve in same parking lot cluster is calculated, it is different in same parking lot cluster to judge
Whether parking lot empty wagons rate curve is similar.
Network struction training unit 23 fights network for constructing the production based on depth convolution, according to parking lot cluster
The empty wagons rate curve image in middle sample parking lot is trained the production confrontation network based on depth convolution, every to generate
The corresponding empty wagons rate curve emulating image of a parking lot cluster.
In embodiments of the present invention, according to the preset convolutional neural networks (CNN) for having supervision to preset production pair
Anti- network (GAN) is extended, to obtain production confrontation network (DCGAN) based on depth convolution at this to the volume for having supervision
The network number of plies and every layer of number of unit of product neural network and production confrontation network are without limiting.The DCGAN of building
Compared to the GAN before extension, the Main change in model structure has: pond layer is substituted with convolutional layer, wherein the pond in discriminator
Change layer to be substituted with the convolutional layer with step-length, the convolutional layer substitution of the pond layer micro-stepping width in generator；In discriminator and generation
All using batch standardization on device；Remove full articulamentum.Preferably, the output layer of generator is connected to for normal state in DCGAN
Property verification filter, only output meets the empty wagons rate curve emulating image of normality verification, to guarantee in DCGAN that generator is raw
At empty wagons rate curve emulating image effect.It is further preferred that the filter for normality verification is used and is mentioned up to Goss
Promise algorithm (D'Agostino-Pearson test), to improve the effect for carrying out normality verification to empty wagons rate curve emulating image
Fruit.
In embodiments of the present invention, the empty wagons rate curve image in sample parking lot in the cluster of parking lot is defeated as training data
Enter in DCGAN, DCGAN is trained, and the corresponding empty wagons rate curve of the parking lot cluster is generated by trained DCGAN and is emulated
Image, to produce the corresponding empty wagons rate curve emulating image of each parking lot cluster.
Parking data repairs unit 24, for handling the corresponding empty wagons rate curve emulating image of each parking lot cluster
And corresponding one-dimensional empty wagons rate data are mapped as, according to the corresponding one-dimensional empty wagons rate data of each parking lot cluster, to each parking
Parking lot to be repaired in the cluster of field carries out parking data repairing.
In embodiments of the present invention, after obtaining the corresponding empty wagons rate curve emulating image of each parking lot cluster, to empty wagons
Rate curve image carries out the image processing operations such as gray processing, binaryzation and noise reduction process, will treated empty wagons rate curve graph
As being mapped as corresponding one-dimensional empty wagons rate data.According in the corresponding one-dimensional empty wagons rate data of parking lot cluster and the parking lot cluster to
The free parking space quantity of parking lot different moments to be repaired in the parking lot cluster can be calculated in the scale for repairing parking lot,
And then it realizes and the parking data in each parking lot to be repaired in each parking lot cluster is repaired.Will treated empty wagons rate curve
When image is mapped as corresponding one-dimensional empty wagons rate data, the empty wagons rates of above-mentioned sample parking lot different moments can be used to empty wagons rate
The reverse-power of the mapping relations of curve image is realized.
Preferably, the formula of gray processing is carried out to empty wagons rate curve emulating image are as follows:
R2=G2=B2=R1*a1+G1*a2+B1*a3, wherein R2、G2And B2For empty wagons rate curve emulating image gray processing it
Rgb value afterwards, R1、G1And B1For the rgb value before empty wagons rate curve emulating image gray processing, a1、a2And a3For preset gray scale
Change parameter.
Preferably, when carrying out binaryzation to the empty wagons rate curve emulating image after gray processing, a threshold value is preset,
The each pixel for traversing empty wagons rate curve emulating image sets white for the pixel that pixel value is more than the threshold value, otherwise
It is set as black, to improve the treatment effect of empty wagons rate curve emulating image.
Preferably, include that the removal of burr point and outlier remove to the noise reduction process of empty wagons rate curve emulating image, consider
Into empty wagons rate curve emulating image the burr point of curve be as pixel it is overstocked caused by, need to reduce pixel point density,
Therefore it is made a return journey flash removed point using mean filter, to improve the treatment effect of burr point on empty wagons rate curve emulating image.
It is further preferred that removing the formula of flash removed point using mean filter are as follows:
Wherein, M indicates preset filtering window size, and f (w, e) is to remove flash removed
Empty wagons rate curve emulating image before point, f ' (x, y) is the empty wagons rate curve emulating image after removing flash removed point.
Preferably, when carrying out outlier removal to empty wagons rate curve emulating image, using the abnormality detection based on distance
Algorithm detects the outlier on empty wagons rate curve emulating image, to improve outlier detection effect.It is further preferred that based on away from
From Outlier Detection Algorithm be local outlier factor LOF algorithm, to improve outlier detection effect.
Preferably, data processing unit 22 includes:
Geographic interest point determination unit 321 is used for according to preset tolerance radius in urban geography interest point set
Determine the geographic interest point in urban parking area set around each parking lot；And
Parking lot cluster cell 322, for constructing each parking lot pair according to the geographic interest point around each parking lot
The space characteristics answered, according to the corresponding space characteristics in each parking lot and preset density-based spatial clustering algorithm, to stopping
Parking lot carries out High Dimensional Clustering Analysis.
Preferably, network struction training unit 23 includes:
Network struction unit 331, for being fought according to the preset convolutional neural networks for having supervision to preset production
Network is extended, and fights network to construct the production based on depth convolution, the production based on depth convolution fights network
In output layer be connected to for normality verify filter.
In embodiments of the present invention, parking lot is clustered as parking lot cluster, is generated according to the parking data in sample parking lot
The empty wagons rate curve image in sample parking lot, according to the empty wagons rate curve image in sample parking lot in the cluster of parking lot to based on depth
The production confrontation network of convolution is trained, and is generated often with fighting network by the trained production based on depth convolution
The corresponding empty wagons rate curve emulating image of a parking lot cluster handles these empty wagons rate curve emulating images and is mapped as pair
The one-dimensional empty wagons rate data answered, to be carried out according to these one-dimensional space rate data to the parking lot to be repaired in each parking lot cluster
Parking data repairing, to be carried out more accurately when not depending on the priori knowledge in parking lot to be repaired to parking lot to be repaired
Data modification, realizes the parking data repairing of City-level, and then has saved the data acquisition cost in parking lot.
In embodiments of the present invention, each unit of the parking data repair apparatus based on DCGAN can be by corresponding hardware
Or software unit realizes that each unit can be independent soft and hardware unit, also can integrate as a soft and hardware unit, herein
Not to limit the present invention.
Embodiment three:
Fig. 4 shows the structure of the calculating equipment of the offer of the embodiment of the present invention three, for ease of description, illustrates only and this
The relevant part of inventive embodiments.
The calculating equipment 4 of the embodiment of the present invention includes processor 40, memory 41 and is stored in memory 41 and can
The computer program 42 run on processor 40.The processor 40 realizes above method embodiment when executing computer program 42
In step, such as step S101 to S104 shown in FIG. 1.Alternatively, being realized when the execution computer program 42 of processor 40 above-mentioned
The function of each unit in Installation practice, such as the function of unit 21 to 24 shown in Fig. 2.
In embodiments of the present invention, parking lot is clustered as parking lot cluster, is generated according to the parking data in sample parking lot
The empty wagons rate curve image in sample parking lot, according to the empty wagons rate curve image in sample parking lot in the cluster of parking lot to based on depth
The production confrontation network of convolution is trained, and is generated often with fighting network by the trained production based on depth convolution
The corresponding empty wagons rate curve emulating image of a parking lot cluster handles these empty wagons rate curve emulating images and is mapped as pair
The one-dimensional empty wagons rate data answered, to be carried out according to these one-dimensional space rate data to the parking lot to be repaired in each parking lot cluster
Parking data repairing, to be carried out more accurately when not depending on the priori knowledge in parking lot to be repaired to parking lot to be repaired
Data modification, realizes the parking data repairing of City-level, and then has saved the data acquisition cost in parking lot.
Example IV:
In embodiments of the present invention, a kind of computer readable storage medium is provided, which deposits
Computer program is contained, the step in above method embodiment is realized when which is executed by processor, for example, Fig. 1
Shown step S101 to S104.Alternatively, realizing each list in above-mentioned apparatus embodiment when the computer program is executed by processor
The function of member, such as the function of unit 21 to 24 shown in Fig. 2.
In embodiments of the present invention, parking lot is clustered as parking lot cluster, is generated according to the parking data in sample parking lot
The empty wagons rate curve image in sample parking lot, according to the empty wagons rate curve image in sample parking lot in the cluster of parking lot to based on depth
The production confrontation network of convolution is trained, and is generated often with fighting network by the trained production based on depth convolution
The corresponding empty wagons rate curve emulating image of a parking lot cluster handles these empty wagons rate curve emulating images and is mapped as pair
The one-dimensional empty wagons rate data answered, to be carried out according to these one-dimensional space rate data to the parking lot to be repaired in each parking lot cluster
Parking data repairing, to be carried out more accurately when not depending on the priori knowledge in parking lot to be repaired to parking lot to be repaired
Data modification, realizes the parking data repairing of City-level, and then has saved the data acquisition cost in parking lot.
The computer readable storage medium of the embodiment of the present invention may include can carry computer program code any
Entity or device, recording medium, for example, the memories such as ROM/RAM, disk, CD, flash memory.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Made any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.
Claims (10)
1. a kind of parking data method for repairing and mending based on DCGAN, which is characterized in that the method includes the following steps:
Urban parking area set and urban geography interest point set gathered in advance are obtained, is stopped in the urban parking area set
Parking lot includes sample parking lot and parking lot to be repaired；
The parking lot is clustered as multiple parking lot clusters according to the urban geography interest point set, and is stopped according to the sample
The parking data in parking lot generates the empty wagons rate curve image in the sample parking lot；
It constructs the production based on depth convolution and fights network, according to the empty wagons rate in sample parking lot described in the parking lot cluster
Curve image is trained the production confrontation network based on depth convolution, to generate each parking lot cluster pair
The empty wagons rate curve emulating image answered；
The corresponding empty wagons rate curve emulating image of each parking lot cluster is handled and is mapped as corresponding one-dimensional empty wagons
Rate data, according to the corresponding one-dimensional empty wagons rate data of each parking lot cluster, to described in each parking lot cluster to
It repairs parking lot and carries out parking data repairing.
2. the method as described in claim 1, which is characterized in that according to the urban geography interest point set by the parking lot
The step of cluster is multiple parking lot clusters, comprising:
It is determined in the urban geography interest point set according to preset tolerance radius every in the urban parking area set
Geographic interest point around a parking lot；
The corresponding space characteristics in each parking lot are constructed according to the geographic interest point around each parking lot, according to institute
The corresponding space characteristics in each parking lot and preset density-based spatial clustering algorithm are stated, higher-dimension is carried out to the parking lot
Cluster.
3. the method as described in claim 1, which is characterized in that generate the sample according to the parking data in the sample parking lot
The step of empty wagons rate curve image in this parking lot, comprising:
The empty wagons rate of sample parking lot different moments is calculated according to the parking data in the sample parking lot；
According to preset parameter of curve and sampled point number, calculate corresponding to the empty wagons rate of sample parking lot different moments
Two-dimensional coordinate.
4. the method as described in claim 1, which is characterized in that the step of production confrontation network of the building based on depth convolution
Suddenly, comprising:
Preset production confrontation network is extended according to the preset convolutional neural networks for having supervision, to construct the base
Network is fought in the production of depth convolution, the output layer of generator connects in the production confrontation network based on depth convolution
It is then used to the filter of normality verification.
5. the method as described in claim 1, which is characterized in that emulated to the corresponding empty wagons rate curve of each parking lot cluster
The step of image is handled and is mapped as corresponding one-dimensional empty wagons rate data, comprising:
Carrying out gray processing, binaryzation and noise reduction process, the noise reduction process to the empty wagons rate curve emulating image includes hair
Thorn point removal and outlier removal；
By treated, the empty wagons rate curve emulating image is mapped as the corresponding one-dimensional empty wagons rate data.
6. a kind of parking data repair apparatus based on DCGAN, which is characterized in that described device includes:
Data capture unit is acquired, for obtaining urban parking area set and urban geography interest point set gathered in advance, institute
Stating the parking lot in urban parking area set includes sample parking lot and parking lot to be repaired；
Data processing unit, for being clustered in the parking lot for multiple parking lots according to the urban geography interest point set
Cluster, and generate according to the parking data in the sample parking lot empty wagons rate curve image in the sample parking lot；
Network struction training unit fights network for constructing the production based on depth convolution, according in the parking lot cluster
The empty wagons rate curve image in the sample parking lot is trained the production confrontation network based on depth convolution, with
Generate the corresponding empty wagons rate curve emulating image of each parking lot cluster；And
Parking data repairs unit, for being handled simultaneously the corresponding empty wagons rate curve emulating image of each parking lot cluster
Corresponding one-dimensional empty wagons rate data are mapped as, according to the corresponding one-dimensional empty wagons rate data of each parking lot cluster, to described every
The parking lot to be repaired in a parking lot cluster carries out parking data repairing.
7. device as claimed in claim 6, which is characterized in that the data processing unit includes:
Geographic interest point determination unit, for being determined in the urban geography interest point set according to preset tolerance radius
Geographic interest point in the urban parking area set around each parking lot；And
Parking lot cluster cell, for constructing each parking lot pair according to the geographic interest point around each parking lot
The space characteristics answered, according to each corresponding space characteristics in parking lot and preset density-based spatial clustering algorithm,
High Dimensional Clustering Analysis is carried out to the parking lot.
8. device as claimed in claim 6, which is characterized in that the network struction training unit includes:
Network struction unit, for being carried out according to the preset convolutional neural networks for having supervision to preset production confrontation network
Extension fights network to construct the production based on depth convolution, and the production based on depth convolution fights network
The output layer of middle generator is connected to the filter verified for normality.
9. a kind of calculating equipment, including memory, processor and storage are in the memory and can be on the processor
The computer program of operation, which is characterized in that the processor realizes such as claim 1 to 5 when executing the computer program
The step of any one the method.
10. a kind of computer readable storage medium, the computer-readable recording medium storage has computer program, and feature exists
In when the computer program is executed by processor the step of any one of such as claim 1 to 5 of realization the method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810869067.7A CN109214422B (en) | 2018-08-02 | 2018-08-02 | Parking data repairing method, device, equipment and storage medium based on DCGAN |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810869067.7A CN109214422B (en) | 2018-08-02 | 2018-08-02 | Parking data repairing method, device, equipment and storage medium based on DCGAN |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109214422A true CN109214422A (en) | 2019-01-15 |
| CN109214422B CN109214422B (en) | 2020-05-22 |
Family
ID=64987991
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810869067.7A Active CN109214422B (en) | 2018-08-02 | 2018-08-02 | Parking data repairing method, device, equipment and storage medium based on DCGAN |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109214422B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109978075A (en) * | 2019-04-04 | 2019-07-05 | 江苏满运软件科技有限公司 | Vehicle dummy location information identifying method, device, electronic equipment, storage medium |
| CN110084158A (en) * | 2019-04-15 | 2019-08-02 | 杭州拓深科技有限公司 | A kind of electrical equipment recognition methods based on intelligent algorithm |
| CN110097060A (en) * | 2019-03-28 | 2019-08-06 | 浙江工业大学 | A kind of opener recognition methods towards trunk image |
| CN110503834A (en) * | 2019-09-10 | 2019-11-26 | 上海科技大学 | The intelligent traffic administration system decision-making technique of Multiple Intersections collaboration is realized based on GAN |
| CN110751853A (en) * | 2019-10-25 | 2020-02-04 | 百度在线网络技术（北京）有限公司 | Parking space data validity identification method and device |
| CN113643564A (en) * | 2021-07-27 | 2021-11-12 | 中国科学院深圳先进技术研究院 | Parking data restoration method and device, computer equipment and storage medium |
| WO2023004595A1 (en) * | 2021-07-27 | 2023-02-02 | 中国科学院深圳先进技术研究院 | Parking data recovery method and apparatus, and computer device and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106483842A (en) * | 2015-08-28 | 2017-03-08 | 易良碧 | A kind of high-precision intelligent is imitated nuclear signal and system and its method of work is occurred |
| CN106997380A (en) * | 2017-03-21 | 2017-08-01 | 北京工业大学 | Imaging spectrum safe retrieving method based on DCGAN depth networks |
| CN108053454A (en) * | 2017-12-04 | 2018-05-18 | 华中科技大学 | A kind of graph structure data creation method that confrontation network is generated based on depth convolution |
| CN108334941A (en) * | 2018-03-06 | 2018-07-27 | 陕西师范大学 | A kind of Bridge Crack image generation model fighting network based on production |
-
2018
- 2018-08-02 CN CN201810869067.7A patent/CN109214422B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106483842A (en) * | 2015-08-28 | 2017-03-08 | 易良碧 | A kind of high-precision intelligent is imitated nuclear signal and system and its method of work is occurred |
| CN106997380A (en) * | 2017-03-21 | 2017-08-01 | 北京工业大学 | Imaging spectrum safe retrieving method based on DCGAN depth networks |
| CN108053454A (en) * | 2017-12-04 | 2018-05-18 | 华中科技大学 | A kind of graph structure data creation method that confrontation network is generated based on depth convolution |
| CN108334941A (en) * | 2018-03-06 | 2018-07-27 | 陕西师范大学 | A kind of Bridge Crack image generation model fighting network based on production |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110097060A (en) * | 2019-03-28 | 2019-08-06 | 浙江工业大学 | A kind of opener recognition methods towards trunk image |
| CN110097060B (en) * | 2019-03-28 | 2021-04-06 | 浙江工业大学 | Open set identification method for trunk image |
| CN109978075A (en) * | 2019-04-04 | 2019-07-05 | 江苏满运软件科技有限公司 | Vehicle dummy location information identifying method, device, electronic equipment, storage medium |
| CN110084158A (en) * | 2019-04-15 | 2019-08-02 | 杭州拓深科技有限公司 | A kind of electrical equipment recognition methods based on intelligent algorithm |
| CN110503834A (en) * | 2019-09-10 | 2019-11-26 | 上海科技大学 | The intelligent traffic administration system decision-making technique of Multiple Intersections collaboration is realized based on GAN |
| CN110751853A (en) * | 2019-10-25 | 2020-02-04 | 百度在线网络技术（北京）有限公司 | Parking space data validity identification method and device |
| CN110751853B (en) * | 2019-10-25 | 2021-05-18 | 百度在线网络技术（北京）有限公司 | Parking space data validity identification method and device |
| CN113643564A (en) * | 2021-07-27 | 2021-11-12 | 中国科学院深圳先进技术研究院 | Parking data restoration method and device, computer equipment and storage medium |
| WO2023004595A1 (en) * | 2021-07-27 | 2023-02-02 | 中国科学院深圳先进技术研究院 | Parking data recovery method and apparatus, and computer device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109214422B (en) | 2020-05-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109214422A (en) | Parking data method for repairing and mending, device, equipment and storage medium based on DCGAN | |
| CN109117987B (en) | Personalized traffic accident risk prediction recommendation method based on deep learning | |
| CN107918753B (en) | Processing Method of Point-clouds and device | |
| CN110909666B (en) | Night vehicle detection method based on improved YOLOv3 convolutional neural network | |
| CN105241524B (en) | A kind of city flood warning system and method based on radial basis function neural network model | |
| CN108921875A (en) | A kind of real-time traffic flow detection and method for tracing based on data of taking photo by plane | |
| CN107463940A (en) | Vehicle type recognition method and apparatus based on data in mobile phone | |
| CN104077613A (en) | Crowd density estimation method based on cascaded multilevel convolution neural network | |
| CN107705551A (en) | The appraisal procedure and device of traffic | |
| CN114450557B (en) | Route deviation quantification and vehicle route learning based thereon | |
| CN105574888B (en) | Crack location lookup method based on gray scale peak value | |
| US11842549B2 (en) | Method and system for muck truck management in smart city based on internet of things | |
| CN114614525A (en) | Intelligent charging pile management system | |
| CN111460866A (en) | Lane line detection and driving control method and device and electronic equipment | |
| CN111476075B (en) | CNN-based object detection method and device by using 1x1 convolution | |
| CN112926425A (en) | Road state detection method, device and equipment | |
| CN115527270A (en) | Method for identifying specific behaviors in intensive crowd environment | |
| CN114999154B (en) | Road service level prediction method based on graph rolling network | |
| CN110057981A (en) | The recognition methods of road network discharge of pollutant sources and system | |
| WO2020024206A1 (en) | Dcgan-based parking data repairing method and apparatus, and device and storage medium | |
| Zou et al. | Application of DCGAN in Data Repair of Parking Lots | |
| CN116824277B (en) | Visual target detection model for road disease detection, construction method and application | |
| CN115131961B (en) | Intelligent traffic perception method and device | |
| CN112613375B (en) | Tire damage detection and identification method and equipment | |
| CN115376025B (en) | Unmanned aerial vehicle target detection method, unmanned aerial vehicle target detection system, unmanned aerial vehicle target detection equipment and storage medium |
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 |