JP2019191936A - Resolution enhancing device, resolution enhancing method, spatial data resolution enhancing method, and program - Google Patents

Abstract

To provide a resolution enhancing device, a resolution enhancing method, a spatial data resolution enhancing method, and a program capable of accurately resolution-enhancing spatial data using auxiliary spatial data having different granularities.SOLUTION: A spatial interpolation unit 150 interpolates, for each of a plurality of domains that indicate a type of value, auxiliary spatial data of a corresponding domain into the auxiliary spatial data of a granularity targeted using an arbitrary spatial interpolation method, and a parameter estimation unit 160 estimates, based on spatial data to be converted and each of the auxiliary spatial data of the granularity targeted for each of the plurality of domains, a parameter of a regression model for converting the spatial data to be converted into the spatial data to be converted with the granularity targeted.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a high resolution device, a high resolution method, and a program, and more particularly, to a high resolution device, a high resolution method, and a program for increasing the resolution of spatial data.

  In recent years, organizations such as governments and companies have collected and disclosed various types of spatial data (poverty level, air pollution level, crime count, population, traffic volume, etc.) for the purpose of improving the urban environment and business. Yes. Spatial data refers to data given as a pair of an area (latitude / longitude, address, region, etc.) and some value associated therewith.

  Since such spatial data is expensive to collect and it is difficult to collect spatially dense data, it is averaged and provided in an area where the unit for dividing the space is wide, that is, an area with a coarse degree of granularity. Often.

  However, high-resolution spatial data is desirable for better urban environment improvement. For example, by narrowing down in detail areas with high poverty and high levels of air pollution, it becomes possible to intervene in determining appropriate policies rather than coarse-grained spatial data.

  Therefore, it is conceivable to convert low resolution spatial data collected with coarse granularity into high resolution spatial data represented with fine granularity.

  For example, apart from the target low-resolution spatial data, other types of high-resolution spatial data are prepared, and by using a regression model that is learned by using them auxiliary to increase the granularity. There is a technique for predicting a value at a fine granularity of target spatial data (Non-patent Document 1).

  In addition, for example, if the spatial correlation (the property that the sample value associated with the position takes a close value when the position is close) and the effect of the regression model described above are taken into account, the spatial data to be targeted is detailed. There is a technique for predicting a value in particle size (Non-Patent Document 2).

C. Smith-Clarke, A. Mashhadi and L. Capra, “Poverty on the Cheap: Estimating Poverty Maps Using Aggregated Mobile Communication Networks”, Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, 2014, pp. 511-520. N. W. Park, “Spatial Downscaling of TRMM Precipitation Using Geostatistics and Fine Scale Environmental Variables”, Advances in Meteorology, volume 2013, article ID 237126, 2013.

  The above prior art (Non-Patent Documents 1 and 2) is based on a regression model learned by using other types of high-resolution spatial data as auxiliary data while considering the spatial correlation of the target spatial data. Based on this, it is possible to predict values at a fine granularity of the target spatial data.

  Here, in the prior art, it is assumed that “the granularity of auxiliary spatial data (aggregation units such as addresses and regions) is equal to each other”.

  However, this assumption is not always appropriate when more types of spatial data are used as supplementary data. This is because the spatial data collected by governments and companies is aggregated with a unique granularity based on the policies of each organization.

  For example, in the national census, population density data with fine granularity for each address can be obtained, but crime occurrence rate data can be obtained for each police jurisdiction area.

  Thus, the conventional technique has a problem that it is difficult to apply when the granularity differs for each spatial data.

  The present invention has been made in view of the above points, and uses a supplementary spatial data having a different granularity to increase the resolution of spatial data with high accuracy and a method for increasing the resolution. And to provide a program.

  The high resolution device according to the present invention uses the auxiliary spatial data with different types of values for the spatial data to be converted consisting of values associated with each area with a granularity indicating the unit of the area into which the space is divided. A resolution increasing device that learns a regression model for conversion to the target spatial data of the granularity to be a target, and for each of a plurality of domains indicating the type of value, the auxiliary of the domain Spatial data is interpolated into the target spatial data of the granularity using an arbitrary spatial interpolation method, the spatial data to be converted, and the domain for each of the plurality of domains A parameter estimation unit configured to estimate a parameter of the regression model based on each of the auxiliary spatial data of the granularity targeted. That.

  In addition, the high resolution method according to the present invention converts the spatial data to be converted, which is a value associated with each area with the granularity indicating the unit of the area into which the space is divided, into auxiliary spatial data having different types of values. It is a high resolution method for learning a regression model for converting to the conversion target spatial data of the granularity to be used, the spatial interpolation unit for each of a plurality of domains indicating the type of the value, The auxiliary spatial data of the domain is interpolated to the auxiliary spatial data of the target granularity using an arbitrary spatial interpolation method, and a parameter estimation unit includes the spatial data to be converted, The regression model parameters are estimated based on each of the target spatial data of the target granularity for each of a plurality of domains.

  According to the resolution increasing device and the resolution increasing method according to the present invention, each of the plurality of domains indicating the type of value associated with each area in the granularity indicating the unit of the area into which the space interpolation unit divides the space is provided. The auxiliary spatial data of the domain is interpolated into the auxiliary spatial data of the target granularity using an arbitrary spatial interpolation method.

  Then, the parameter estimation unit, based on the conversion target spatial data and each of the auxiliary spatial data of the target granularity for each of a plurality of domains, the conversion target space of the granularity Estimate regression model parameters to convert to data.

  In this way, for each of a plurality of domains indicating value types, the auxiliary spatial data of the domain is interpolated into the auxiliary spatial data of the target granularity using an arbitrary spatial interpolation method, and converted. Based on the target spatial data and each of the auxiliary spatial data of the granularity targeted for each of a plurality of domains, the conversion target space of the granularity targeting the spatial data to be converted By estimating the parameters of the regression model for conversion to data, it is possible to increase the resolution of the spatial data with high accuracy using auxiliary spatial data having different granularities.

  Further, the spatial interpolation unit of the resolution increasing apparatus according to the present invention uses the Gaussian process regression as the spatial interpolation method for each of the plurality of domains, and the auxiliary spatial data of the target granularity. As an average and variance of the domain values for each area with the granularity as the target, the parameter estimation unit, for each of the plurality of domains by the spatial data to be converted and the spatial interpolation unit The parameters of the regression model can be estimated based on the obtained average and variance of the domain values for each area with the target granularity.

  Further, the spatial interpolation unit of the resolution increasing method according to the present invention uses the Gaussian process regression as the spatial interpolation method for each of the plurality of domains, and the auxiliary spatial data of the target granularity. As an average and variance of the domain values for each area with the granularity as the target, the parameter estimation unit, for each of the plurality of domains by the spatial data to be converted and the spatial interpolation unit The parameters of the regression model can be estimated based on the obtained average and variance of the domain values for each area with the target granularity.

  Further, the high resolution device according to the present invention includes the spatial data to be converted, each of the auxiliary spatial data of the granularity to be the target for each of the plurality of domains, and the regression model. Based on this, it is possible to provide a high-resolution data calculation unit that calculates the conversion target spatial data of the granularity as the target.

  Further, in the high resolution method according to the present invention, the high resolution data calculation unit includes each of the spatial data to be converted and the auxiliary spatial data of the granularity to be the target for each of the plurality of domains. Based on the regression model, the spatial data to be converted having the granularity as the target can be calculated.

  A program according to the present invention is a program for causing each unit of the above-described high resolution device to function.

  According to the high resolution device, high resolution method, and program of the present invention, it is possible to increase the resolution of spatial data with high accuracy using auxiliary spatial data having different granularities.

It is a block diagram which shows the structure of the resolution increasing apparatus which concerns on embodiment of this invention. It is the example (upper figure) and output example (lower figure) of the particle size in embodiment of this invention. It is a flowchart which shows the spatial data high resolution process routine of the high resolution apparatus which concerns on embodiment of this invention. It is a flowchart which shows the search processing routine of the resolution increasing apparatus which concerns on embodiment of this invention.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

<Outline of High Resolution Apparatus According to Embodiment of the Present Invention>
First, the outline | summary of embodiment of this invention is demonstrated.

  In the present embodiment, spatial data of various domains (types) having various granularities is supplementarily used, and a statistical relationship between the spatial data having coarse granularity to be converted is learned by a regression model.

Specifically, by applying spatial interpolation methods such as Kriging (reference 1) and Gaussian process regression (reference 2) to auxiliary spatial data with various granularities, these auxiliary spatial data Make the particle size uniform. Then, using these as inputs, a statistical relationship with the spatial data to be converted is learned by a regression model.
[Reference 1] T. Hengl, GBM Heuvelink and A. Stein, “A Generic Framework for Spatial Prediction of Soil Variables based on Regression-Kriging”, Geoderma, volume 120, issues 11-2, 2004, pages 75-93.
[Reference 2] CE Rasmussen and CKI Williams, Gaussian Processes for Machine Learning, MIT Press, 2006.

  With such a configuration, the regression model can be learned even when auxiliary spatial data having different granularities are given.

  For example, even when the aggregated units are different, such as population density data for each address and crime occurrence rate data for each police jurisdiction area, the regression model is learned and used as the target. It can be used to increase the resolution of coarse-grained spatial data.

  In addition, since the spatial interpolation method is applied to the auxiliary spatial data and the sample value at an arbitrary point can be predicted, the target granularity can be arbitrarily set.

  For example, it is possible to set the granularity targeting an area set or the like divided by a 100 m × 100 m grid. As a result, high-resolution data can be calculated for an area set based on the granularity required by the government or companies.

  In this embodiment, a case where spatial interpolation using Gaussian process regression is performed on auxiliary spatial data having various granularities will be described. By using the Gaussian process regression, the predicted value and variance (reliability with respect to the predicted value) at an arbitrary point are calculated, and the regression model is effectively learned by adding the reliability with respect to the predicted value.

  In the learning of the regression model, the importance for each of the auxiliary spatial data is learned. Here, the degree of importance represents the degree to which each of the auxiliary spatial data can explain the target high-resolution spatial data. This importance is not determined solely by the correlation with the target spatial data, but also depends on the coarseness of the granularity of the spatial data used as an auxiliary.

  For example, it is assumed that there is spatial data A and auxiliary spatial data B and C to be converted, and the correlations between A and B and A and C are exactly the same. At this time, if the spatial granularity of B is fine and the spatial granularity of C is coarse, it is considered that the spatial data B with fine granularity can explain the high-resolution spatial data more.

  Thus, there is a merit that the regression model can be estimated while taking into account the difference in spatial granularity of the importance of each spatial data used as an auxiliary.

  Therefore, the regression model can be learned effectively while taking into account the difference in the granularity of the spatial data used as ancillary, and the coarse-grained spatial data to be converted can be converted to high resolution by using the learned regression model. Can do.

<Configuration of High Resolution Apparatus According to Embodiment of the Present Invention>
With reference to FIG. 1, the structure of the high-resolution apparatus 10 which concerns on embodiment of this invention is demonstrated. FIG. 1 is a block diagram showing a configuration of a resolution increasing apparatus 10 according to an embodiment of the present invention.

  The high resolution device 10 is composed of a computer including a CPU, a RAM, and a ROM that stores a program for executing a high resolution processing routine and a search processing routine, which will be described later. It is configured as follows.

  As shown in FIG. 1, the resolution increasing apparatus 10 according to this embodiment includes an auxiliary spatial data storage unit 100, a target granularity area center storage unit 110, a conversion target spatial data storage unit 120, and a granularity area correspondence table. Storage unit 130, operation unit 140, spatial interpolation unit 150, parameter estimation unit 160, high resolution data calculation unit 170, high resolution data storage unit 180, input unit 190, search unit 200, output unit 210.

  The auxiliary spatial data storage unit 100, the target granularity area center storage unit 110, the conversion target spatial data storage unit 120, the granularity area correspondence table storage unit 130, and the high resolution data storage unit 180 are a database that includes a Web server or a database. A server or the like reads data according to the request and transmits the corresponding data to the request source.

  The auxiliary spatial data storage unit 100 stores auxiliary spatial data that can be used auxiliary to increase the resolution of the spatial data to be converted for each domain indicating the type of value.

Specifically, a set representing the entire input space

And here,

Is a subset of the Euclidean space, for example, representing the entire target city.

Also, the set representing the domain

And the domain

The area set that the spatial data of

And here,

Is

Division of

The union of areas that are elements of

Matches.

Also area

The value of the spatial data of domain s in

It expresses.

And put together the spatial data of domains

Dimension vector

And here,

Is

This represents the number of areas included in the.

Domains area

Center coordinates at

And collectively

And

That is, the auxiliary spatial data storage unit 100

Are stored as auxiliary spatial data.

The target granularity area center storage unit 110 is a set of central coordinates of each area j for an area set of target granularity.

Are stored for each granularity.

  Here, the granularity is a unit of an area that divides a space. For example, as shown in the upper diagram of FIG. 2, a plurality of target particle size candidates that are particle sizes that can be targets are prepared, and the target particle size candidate A and the target particle size candidate B have a coarser particle size. is there.

Specifically, the target granularity area set

And Fine area

Center coordinates at

And collectively

And

here,

Is

This represents the number of areas included in the.

The conversion target spatial data storage unit 120 is a coarse granularity spatial data to be converted.

Is stored.

Here, the area set of coarse granularity of the spatial data to be converted

And Also rough area

The value of the target spatial data at

And collectively

Dimension vector

And

here,

Is

This represents the number of areas included in the.

The granularity area correspondence table storage unit 130 is a correspondence table that indicates which area of coarse granularity each area of the target granularity is included in.

Is stored.

here,

Represents a set of target granularity areas j included in the coarse granularity area i.

  The operation unit 140 receives various operations from the user for the data stored in the auxiliary spatial data storage unit 100, the target granularity area center storage unit 110, the conversion target spatial data storage unit 120, and the granularity area correspondence table storage unit 130.

  The various operations are, for example, operations for registering, correcting, and deleting stored information. The input unit of the operation unit 140 may be anything such as a keyboard, a mouse, a menu screen, or a touch panel. The operation unit 140 is realized by a device driver of an input unit such as a mouse or menu screen control software.

  For each of the plurality of domains, the spatial interpolation unit 150 interpolates the auxiliary spatial data of the domain into the auxiliary spatial data having a target granularity using an arbitrary spatial interpolation method.

  In this embodiment, for each of a plurality of domains, using Gaussian process regression as a spatial interpolation method, as an auxiliary spatial data of the target granularity, the average value of the domain for each area at the target granularity The case of obtaining the variance will be described.

Specifically, the spatial interpolation unit 150 performs auxiliary spatial data of the domain s stored in the auxiliary spatial data storage unit 100 as described below.

Is obtained as a continuous expression (predicted distribution) in each domain s for each granularity, and then the target granularity area center storage unit 110 is obtained. The average (predicted value) and variance at the center of each area stored in the are estimated.

First,

Is some coordinates,

Be a potential function for auxiliary spatial data in domain s.

Is assumed to follow a Gaussian process, the average is 0, and the correlation function is the following equation (1).

here,

There is a point

Is a scale parameter that determines the extent of the correlation to a point around

Is a dispersion parameter that determines the magnitude of the correlation. Formula (1) is called a squared-exponential kernel and is one of the correlation functions that are most often used to measure the similarity of certain data in spatial coordinates.

Center coordinates of each area in the area set of domain s

Given training data representing

The joint distribution can be expressed by a multidimensional Gaussian distribution and can be expressed as the following equation (2).

here,

Is

Matrix, each element is

It is.

Next, the spatial data of domain s

Assumes that Gaussian noise is added. At this time, the spatial data of domain s

Follows the following equation (3), which is a conditional probability.

here,

Is the identity matrix,

Is a dispersion parameter for noise.

Here, due to the conjugate properties of the above formulas (2) and (3),

Can be analytically eliminated. For this reason, the spatial data of domain s

Is defined by the following equation (4).

Then, the space interpolation unit 150 is a hyperparameter for Expression (4) based on the marginal likelihood maximization.

Is estimated. The marginalized log likelihood function that takes the logarithm of both sides of Equation (4) can be expressed by Equation (5) below.

That is, the space interpolation unit 150 is configured to make the hyperparameter that maximizes the expression (5).

Ask for. Here, any optimization method may be used. For example, the BFGS method (reference document 3) can be used.
[Reference 3] DC Liu and J. Nocedal, “On the Limited Memory BFGS Method for Large Scale Optimization”, Mathematical Programming, vol. 45, 1989, pp. 503-528.

Next, the space interpolation unit 150 calculates the estimated hyperparameter.

For each granularity, the center coordinates of each area j stored in the target granularity area center storage unit 110

Calculate the predicted distribution at.

, Each test sample

Predicted value at

The combined prediction distribution is expressed by the following equation (6).

here,

Is an average (predicted average) and can be calculated by the following equation (7):

Is a variance (correlation matrix) and can be calculated by the following equation (8).

here,

Is

Matrix

Each element of training data

Each training sample

And test data

Each test sample

Is the value of the correlation function (equation (1)). However, the estimated hyperparameter

Is used.

Also,

Is

Matrix

Each element of the test data

Each test sample

Is the value of the correlation function (formula (1)). However, the estimated hyperparameter

Is used.

Thus, the space interpolation unit 150 performs the above processing for all domains s as hyperparameters.

For each target granularity

And dispersion

Is obtained by interpolating the auxiliary spatial data of the domain s into the auxiliary spatial data of the target granularity.

Then, the space interpolation unit 150 calculates the average over all the obtained domains s.

And dispersion

To the parameter estimation unit 160 and the high resolution data calculation unit 170.

  The parameter estimation unit 160 calculates the space to be converted with the target granularity based on the spatial data to be converted and the average and variance of the domain values for each area with the target granularity for each of the plurality of domains. Estimate regression model parameters to convert to data.

Specifically, the parameter estimation unit 160 firstly

Is a potential function for spatial data to be transformed,

Follows a Gaussian process, and the average function is represented by the following equation (9).

here,

Is a weight parameter for the spatial data of domain s,

Is a bias parameter. The correlation function is the same as that in the above equation (1). However, the hyperparameter of the correlation function

And

Next, the parameter estimation unit 160 determines the center coordinates of each area j of the area set at the target granularity.

The sample value of the spatial data to be converted for

And

here,

Is the spatial data obtained by converting the spatial data to be converted into the spatial data to be converted with the target granularity.

Call it.

Given an auxiliary spatial data prediction for each domain s,

The conditional probability of can be written as the following equation (10).

here,

And

Is

Is a correlation matrix. Also,

Is a column vector in which numerical values 1 are arranged,

It was.

Spatial data to be converted

Is

High resolution data

Assuming that

Can be considered to be generated from the conditional distribution of the following equation (11).

here,

Is the dispersion parameter for noise,

Is

Is the averaging matrix.

The parameter estimation unit 160 is stored in the granularity area correspondence table storage unit 130.

Using,

Elements of

Is

If you meet

And 0 if not satisfied.

Next, the parameter estimation unit 160 is a weight parameter that is a model parameter.

And parameters controlling spatial correlation

Is estimated.

Specifically, the conversion target spatial data stored in the conversion target spatial data storage unit 120

Can be expressed by the following equation (12).

here,

It was. Also,

Is

Covariance matrix of each element

Can be represented by the following formula (13).

here,

Is a Kronecker delta function which takes a value of 1 when A = B and takes a value of 0 otherwise. In addition, the second term in the second row of the equation (13) is

Represents the variance for the regression residual. here,

Note that is added. The average obtained by the spatial interpolation unit 150 using auxiliary spatial data in each domain s

Variance for

But

Affects the variance of the regression residual. As a result, the weight parameter is added while taking into account the reliability of prediction estimated using auxiliary spatial data, that is, the reliability of spatial interpolation.

Can be estimated.

The parameter estimation unit 160 then weights the parameter so as to maximize the above equation (12).

And parameters controlling spatial correlation

Is estimated.

  Here, the marginalized log likelihood function having the logarithm on both sides of the equation (12) can be written as the following equation (14).

Therefore, the parameter estimation unit 160 uses a weight parameter that maximizes the above equation (14).

And parameters controlling spatial correlation

Is estimated. Any optimization method may be used. For example, the BFGS method (reference document 3) can be used.

The parameter estimation unit 160 then calculates the estimated weight parameter.

And parameters controlling spatial correlation

Is transferred to the high-resolution data calculation unit 170.

  The high-resolution data calculation unit 170 converts the target granularity based on the spatial data to be converted, each auxiliary spatial data having the target granularity for each of the plurality of domains, and the regression model. The spatial data of is calculated.

Specifically, the high resolution data calculation unit 170 converts the spatial data to be converted stored in the conversion target spatial data storage unit 120.

The average of auxiliary spatial data in each domain obtained by the spatial interpolation unit 150

And dispersion

, And a set of center coordinates of each area j with respect to the target granularity area set stored in the target granularity area center storage unit 110

And the correspondence table stored in the granularity area correspondence table storage unit 130

And the weight parameter estimated by the parameter estimation unit 160

And parameters controlling spatial correlation

And high-resolution data that is the spatial data to be converted with the target granularity

Prediction distribution

High resolution data by seeking

Is calculated.

Predictive distribution

Is the following formula (15).

here,

Is an average and can be represented by the following formula (16).

Also,

Is a correlation matrix and can be expressed by the following equation (17).

The high resolution data calculation unit 170 then converts the high resolution data that is the conversion target spatial data of the calculated target granularity.

Average of

And correlation matrix

Are stored in the high-resolution data storage unit 180.

The high resolution data storage unit 180 is high resolution data that is spatial data to be converted with the target granularity calculated by the high resolution data calculation unit 170.

Average of

And correlation matrix

Is stored.

  Then, the resolution increasing apparatus 10 performs the processes of the spatial interpolation unit 150, the parameter estimation unit 160, and the high resolution data calculation unit 170 for all target granularity candidates.

  The input unit 190 receives an input of a target granularity ID that is an ID of a target granularity candidate.

  Then, the input unit 190 passes the received target granularity ID to the search unit 200.

  The means for inputting the target granularity ID to the input unit 190 may be anything such as a keyboard, mouse, menu screen, or touch panel. The input unit 190 can be realized by a device driver of an input unit such as a mouse or a menu screen control software.

The search unit 200 is a high-resolution data obtained by increasing the resolution of the spatial data to be converted for the area set of the granularity specified by the target granularity ID received by the input unit 190.

Search for.

Specifically, as shown in the upper diagram of FIG. 2, high-resolution data which is spatial data to be converted of a target granularity candidate having the same ID as the target granularity ID received by the input unit 190.

Average of

Is retrieved from the high-resolution data storage unit 180.

And the search part 200 is a search result.

To the output unit 210.

The output unit 210 is acquired from the search unit 200.

Is output as high-resolution data.

  For example, as shown in the lower diagram of FIG. 2, the high-resolution data is output to the granularity of the target granularity candidate A from the spatial data to be converted with a coarse granularity.

  In the above description, display on the display is taken as an example, but the output unit 210 may be configured to output such as printing to a printer, sound output, and transmission to an external device. The output unit 210 may include an output device such as a display or a speaker. The output unit 210 may be realized by driver software for an output device, driver software for an output device, an output device, or the like.

<Operation of the high resolution device according to the embodiment of the present invention>
FIG. 3 is a flowchart showing a resolution enhancement processing routine according to the embodiment of the present invention.

  When the high resolution device 10 executes an instruction for performing the high resolution processing, the high resolution processing routine shown in FIG. 3 is executed.

First, in step S100, the space interpolation unit 150 estimates, for each of a plurality of domains, hyperparameters for interpolating the auxiliary spatial data of the domain into the auxiliary spatial data of various granularities.
In step S110, the high resolution device 10 selects the first target granularity ID.

  In step S120, for each of the plurality of domains, the space interpolation unit 150 arbitrarily outputs auxiliary spatial data of the domain using the hyperparameters estimated in step S100 for the auxiliary spatial data of the domain. Are interpolated into the auxiliary spatial data of the target granularity.

  In step S130, the parameter estimation unit 160 determines the target granularity based on the spatial data to be converted and the average and variance of the domain values for each area in the target granularity for each of the plurality of domains. Estimate the parameters of the regression model to convert to the spatial data to be converted.

  In step S140, the high-resolution data calculation unit 170 sets a target based on the spatial data to be converted, each auxiliary spatial data having a granularity targeted for each of a plurality of domains, and the regression model. Calculate the spatial data to be converted for granularity.

  In step S150, the high resolution data storage unit 180 stores the average and correlation matrix of the high resolution data that is the spatial data to be converted with the target granularity calculated in step S140.

  In step S160, the resolution increasing apparatus 10 determines whether or not processing has been performed for all target granularity IDs.

  When the processing has not been performed for all target granularity IDs (NO in step S160), in step S170, the resolution increasing apparatus 10 selects the next target granularity ID and repeats the processing in steps S120 to S160.

  On the other hand, when the process is performed for all target granularity IDs (YES in step S160), the process ends.

  FIG. 4 is a flowchart showing a search processing routine according to the embodiment of the present invention.

  In step S200, an input of a target granularity ID that is an ID of a target granularity is received.

  In step S210, the search unit 200 searches the area set having the granularity specified by the target granularity ID received in step S200 for high-resolution data obtained by increasing the resolution of the spatial data to be converted.

  In step S220, the output unit 210 outputs the high resolution data acquired from the search unit 200.

  As described above, according to the high resolution device according to the embodiment of the present invention, for each of a plurality of domains indicating the type of value, auxiliary spatial data of the domain is used by an arbitrary spatial interpolation method. And interpolating to the auxiliary spatial data of the target granularity, based on the spatial data to be converted and each of the auxiliary spatial data of the granularity targeted for each of a plurality of domains, In order to estimate the parameters of the regression model for converting the target spatial data to the target spatial data of the target granularity, the spatial data is accurately obtained using auxiliary spatial data with different granularities. High resolution can be achieved.

  The present invention is not limited to the above-described embodiment, and various modifications and applications can be made without departing from the gist of the present invention.

For example, in the above-described embodiment, high resolution data is previously stored for all granularities.

Average of

And correlation matrix

Are stored in the high-resolution data storage unit 180, but each time the target granularity ID is input, the resolution may be increased with the learned model and output.

In addition, the input unit 190 receives spatial data to be converted.

The target granularity ID is received, stored in the conversion target spatial data storage unit 120, and the spatial data to be converted is increased in resolution as the target granularity of the input target granularity ID, and this is output. It is good also as a structure.

  In the present specification, the embodiment has been described in which the program is installed in advance. However, the program can be provided by being stored in a computer-readable recording medium.

10 High Resolution Device 100 Auxiliary Spatial Data Storage Unit 110 Target Granularity Area Center Storage Unit 120 Conversion Target Spatial Data Storage Unit 130 Granularity Area Corresponding Table Storage Unit 140 Operation Unit 150 Spatial Interpolation Unit 160 Parameter Estimation Unit 170 High Resolution Data Calculation Unit 180 High-resolution data storage unit 190 Input unit 200 Search unit 210 Output unit

Claims (7)

The conversion target spatial data consisting of values associated with each area in the granularity indicating the unit of the area that divides the space is targeted using auxiliary spatial data of different types of values. A high resolution device for learning a regression model for conversion to spatial data,
Spatial interpolation for interpolating the auxiliary spatial data of the domain into the auxiliary spatial data of the target granularity using an arbitrary spatial interpolation method for each of a plurality of domains indicating the type of value And
A parameter estimation unit that estimates parameters of the regression model based on the spatial data to be converted and each of the auxiliary spatial data of the granularity to be the target for each of the plurality of domains;
High resolution device. For each of the plurality of domains, the spatial interpolation unit uses Gaussian process regression as the spatial interpolation method, and the auxiliary spatial data of the granularity to be the target is the area at the granularity to be the target. Find the mean and variance of the domain values for each
The parameter estimation unit includes the spatial data to be converted, and the average and variance of the domain values for each area in the granularity to be the target, obtained for each of the plurality of domains by the spatial interpolation unit. The high resolution device according to claim 1, wherein the parameter of the regression model is estimated based on the parameter. The conversion of the target granularity based on the spatial data to be converted, each of the auxiliary spatial data of the target granularity for each of the plurality of domains, and the regression model. The high-resolution device according to claim 1, further comprising: a high-resolution data calculation unit that calculates the target spatial data. The conversion target spatial data consisting of values associated with each area in the granularity indicating the unit of the area that divides the space is targeted using auxiliary spatial data of different types of values. A high resolution method for learning a regression model for conversion to spatial data,
For each of a plurality of domains indicating the type of value, a spatial interpolation unit uses the auxiliary spatial data of the domain as the target using the arbitrary spatial interpolation method as the auxiliary spatial data of the granularity. Interpolate to
A parameter estimation unit estimates the parameters of the regression model based on the spatial data to be converted and each of the auxiliary spatial data of the granularity to be the target for each of the plurality of domains. Resolution method. For each of the plurality of domains, the spatial interpolation unit uses Gaussian process regression as the spatial interpolation method, and the auxiliary spatial data of the granularity to be the target is the area at the granularity to be the target. Find the mean and variance of the domain values for each
The parameter estimation unit includes the spatial data to be converted, and the average and variance of the domain values for each area in the granularity to be the target, obtained for each of the plurality of domains by the spatial interpolation unit. The high resolution method according to claim 4, wherein the parameter of the regression model is estimated based on the parameter. The high-resolution data calculation unit is configured to determine the target based on the spatial data to be converted, each of the auxiliary spatial data of the granularity to be the target for each of the plurality of domains, and the regression model. The high resolution method according to claim 4 or 5, wherein the spatial data to be converted having the granularity is calculated.   The program for functioning a computer as each part of the resolution increasing apparatus of any one of Claims 1-3.