CN104199937A - Multi-website POI position mapping method and device - Google Patents

Abstract

A position mapping method and device for multi-site POIs. The method includes the following specific steps: S110: Establish a correction unit set; S120: Perform region splitting in each correction unit of the correction unit set; S130: Semantic-based control point matching; S140: Multi-site POI control point set with the same name , Automatic and uniform acquisition of check point sets; S150: Calculation of position mapping model coefficients; S160: Evaluation of position mapping accuracy. The invention evenly divides POI positions into several units, uniformly selects control points, comprehensively utilizes semantic matching and least squares methods to fully automatically select control points, and uses a "divide and conquer" strategy to establish an optimal position mapping model.

Description

A location mapping method and device for multi-site POIs

technical field

The present invention relates to the field of spatial information, in particular to a spatial position mapping method and device, which are used to establish position mapping relationships among POIs of multiple websites.

Background technique

POI (Point Of Interest, point of interest) is an important content of geographic information services. important role. With the development of Internet technology, Internet POI information resources are increasingly abundant, and the rich PO.I data contained in the Internet has become an important data source for expanding and updating POI databases. However, there is a problem of inconsistent location information in multi-site POI data sources. It is time-consuming and labor-intensive to establish a location mapping relationship with traditional manual/human-computer interaction methods, and the accuracy is low.

1. Control point selection method

To establish the location mapping relationship of multi-site POIs, the reasonable selection of control points is a key step. The number and distribution of control points will seriously affect the accuracy of position mapping. Unreasonable point distribution will directly lead to local deformation. In order to reduce the adverse effects caused by inaccurate selection of control points and uneven distribution, a large number of redundant control points are needed, and the selected POI control points should be evenly distributed within the geographical range of the POI point set. The number, distribution and precision of control points will directly affect the effect of location mapping.

In the prior art, there are mainly three methods for selecting control points: purely manual selection of control points, human-computer interactive selection of control points, and matching and automatic generation of control points.

(1) Manual selection of control points: The control point selection methods of POIs on different websites can be manually interpreted according to the operator's experience and attribute information of POIs. The POI coordinate information to be mapped is (r, c), and the reference POI coordinate information is (x, y), and corresponding coordinate point pairs are established to generate control points. This method has a relatively large dependence on the workload and experience of the operator, and the time consumption is large, and the selection accuracy of the control points is difficult to guarantee.

(2) Human-computer interactive selection of ground control points: In this method, the operator first selects several control points through human-computer interaction based on experience, and uses these control points to determine approximate transformation parameters, and then uses recognition technology to identify In order to obtain enough control points within the geographical range, the conversion parameters can be determined more accurately. This method realizes the combination of man and machine, reduces the consumption of time to a certain extent, and the result has better objectivity. In the selection process, the decisive factors are the accuracy of the control points selected by human-computer interaction and the reliability of the recognition technology.

(3) Automatic selection of control points: This process of selecting control points does not require human-computer interaction and manual participation, and relies entirely on the automatic identification of the computer to find out the control points and then establish a position mapping model. This method can minimize Time overhead and human cost. At present, the automatic selection method of control points has been studied more in the field of remote sensing images, but less involved in the field of multi-site POI. For the multi-site POI data, the control points are mainly selected automatically through the recognition technology, but the positions of the control points matched by this method are often unevenly distributed, which leads to local deformation due to the ill-conditioned condition of the normal equation coefficient matrix during the position mapping solution; and In a large geographical range, only relying on recognition technology without combining with methods such as regional division, location mapping is difficult to achieve the global optimal effect, and cannot meet the requirements of practical applications for POI accuracy.

2. Position mapping method

Common position mapping methods in the prior art include average displacement method, similar transformation method, polynomial method, etc.:

(1) Average displacement method: the location mapping relationship is established by adding displacements to the geographic locations of different data sources, and the displacements are usually the average displacements of common points. The average displacement model is expressed in matrix form as follows:

$\left\{\begin{array}{c}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}\\ \mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}\end{array}\right.$

Among them, r, c and x, y are the location information of different source POIs.

The average displacement method requires at least one control point for position mapping. It is suitable for data with a small geographical range of the data source, and the systematic error is more obvious than the accidental error, but for a large geographical range, the accidental error and the systematic error are not obvious. The data mapping for is not good, and increasing the number of control points does little to improve the location mapping.

(2) Similarity transformation method: Transformation of graphics into similar graphics in Euclidean space, such as translation, rotation, and scaling. The similarity transformation model is expressed in matrix form as follows:

$\left\{\begin{array}{c}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}\\ \mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}\end{array}\right.$

Among them, r, c and x, y are the location information of different source POIs.

The similarity transformation method requires at least two control points to perform position mapping. It has the advantages of clear geometric relations, regular transformation formulas, and appropriate extrapolation. Increasing the amount of control point data can partially improve the position mapping effect, but for large geographical ranges, Data in which neither random nor systematic errors are significant map well.

(3) Polynomial transformation method: use polynomials to represent the location mapping relationship between different geographical locations.

The second-order polynomial transformation model is as follows:

$\left\{\begin{array}{c}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; r</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 4</span>}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; c</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 5</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; m</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}\\ \mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; r</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; 4</span>}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; c</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; 5</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}\end{array}\right.$

Among them, r, c and x, y are the location information of different source POIs.

The second-order polynomial transformation method requires at least 6 control points for position mapping. The position mapping effect can be improved by increasing the number of control points, and the mapping effect is better for data with insignificant accidental and systematic errors, but for a large geographical range, only the polynomial transformation method cannot achieve the global optimal mapping effect.

To sum up, the current establishment method of multi-site POI location mapping relationship is still in the purely manual or semi-automatic stage. This method is time-consuming, inefficient, and costly. For POI point sets with a large geographical range, only common locations are used. If the mapping method is not combined with the matching technique, the position mapping cannot achieve the global optimal effect. Therefore, how to automatically select and evenly select control points, and establish the position mapping relationship of POIs between websites has become a technical problem that needs to be solved urgently in the prior art.

Contents of the invention

The purpose of the present invention is to solve the problem that the current multi-site POI location mapping relationship cannot be fully automated, and for POI point sets with a large geographical range, the common location mapping method cannot achieve the global optimal effect. The location mapping relationship of multi-site POIs is established by combining the semantic matching method and the least squares principle.

For reaching this purpose, the present invention adopts following technical scheme:

A multi-site POI position mapping method, comprising the steps of:

S110. Establish a correction unit set:

The minimum and maximum geographic coordinates are calculated respectively according to the geographic coordinate set of the POI point set, and the minimum and maximum geographic coordinates obtained are the geographic range of the POI point set, and a certain geographic interval is used to divide the geographic range of the POI point set into N*N small A grid, the full set of N*N grids constitutes a set of correction units;

S120. Perform region splitting in each correction unit of the correction unit set:

After the correction unit set is established, in order to ensure the uniform distribution of control points in each correction unit of the correction unit set, each correction unit is evenly split into multiple coordinate sub-grids to form the coordinate sub-grid of control points, In order to ensure the uniform distribution of check points in each correction unit of the correction unit set, each correction unit is evenly split into multiple coordinate sub-grids to form the coordinate sub-grid of check points;

S130. Semantic-based control point matching:

For POI names from different websites, use the Chinese word segmentation tool to perform word segmentation, and then determine whether they are the same control point according to the phrase string after word segmentation;

S140. Automatic and uniform acquisition of control point sets and check point sets of the same name of POIs of multiple websites:

In the range of each control point sub-grid, use the method of semantic matching to automatically obtain control points and add them to the set of control points. In the range of each check point sub-grid, use the method of semantic matching to automatically obtain at least one check point , and join the checkpoint set;

S150. Solving position mapping model coefficients

Using the set of control points and checkpoints obtained in the above steps, a second-order polynomial transformation model is selected to solve the position mapping model coefficients for each correction unit.

Among them, the second-order polynomial transformation model is expressed in matrix form by formula (1):

$\left(\begin{array}{c}x\\ \mathrm{the\; y}\end{array}\right)=\left(\begin{array}{ccccccccccc}1& r& c& r^2{c}^2& r\&times;c& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 1& r& c& r^2{c}^2& r\&times;c\end{array}\right)\left(\begin{array}{c}{m}_0\\ m_1\\ m_2\\ m_3\\ m_4\\ m_5\\ {\mathrm{no}}_0\\ {\mathrm{no}}_1\\ {\mathrm{no}}_2\\ {\mathrm{no}}_3\\ {\mathrm{no}}_4\\ {\mathrm{no}}_5\end{array}\right)$ Formula 1)

Among them, r, c are the geographical coordinates of POI of a certain website, x, y are the geographical coordinates of POI of another website, m ₀ , m ₁ , m ₂ , m 3 , m ₄ , m ₅ , n ₀ , n _{1 ,} n ₂ , n ₃ , n ₄ , and n ₅ are position mapping coefficients in the second-order polynomial model, and the geographic coordinates are longitude and latitude.

Preferably, in the step of establishing a correction unit set, the geographical interval is 4°, 2°, 1°, 30' or 15'.

Preferably, in the region splitting step in each correction unit of the correction unit set, each correction unit is evenly split into 6*6 coordinate sub-grids to form a coordinate sub-grid of control points, and each The correction unit is evenly split into 12*12 coordinate sub-grids to form the coordinate sub-grids of the checkpoints.

Preferably, in the step of calculating the coefficients of the position mapping model, the error equation is first listed according to the second-order polynomial transformation model, and then the position mapping coefficient in the error equation is calculated according to the method of indirect adjustment.

Preferably, after S150, there is also S160: Position Mapping Accuracy Evaluation

For the correction unit set, traverse each correction unit, use the position mapping coefficient Δ obtained by each correction unit to calculate the middle error, maximum error and minimum error of the check point set, and complete the POI position mapping accuracy of multiple sites assessment.

The present invention also discloses a multi-site POI position mapping device, which includes the following units:

Correction unit set establishment unit (210):

The minimum and maximum geographic coordinates are calculated respectively according to the geographic coordinate set of the POI point set, and the minimum and maximum geographic coordinates obtained are the geographic range of the POI point set, and a certain geographic interval is used to divide the geographic range of the POI point set into N*N small A grid, the full set of N*N grids constitutes a set of correction units;

Region split unit (220):

After the correction unit set is established, in order to ensure the uniform distribution of control points in each correction unit of the correction unit set, each correction unit is evenly split into multiple coordinate sub-grids to form the coordinate sub-grid of control points, In order to ensure the uniform distribution of check points in each correction unit of the correction unit set, each correction unit is evenly split into multiple coordinate sub-grids to form the coordinate sub-grid of check points;

Semantic-based control point matching unit (230):

For POI names from different websites, use the Chinese word segmentation tool to perform word segmentation, and then determine whether they are the same control point according to the phrase string after word segmentation;

Control point set, checkpoint set acquisition unit (240):

In the range of each control point sub-grid, use the method of semantic matching to automatically obtain control points and add them to the set of control points. In the range of each check point sub-grid, use the method of semantic matching to automatically obtain at least one check point , and join the checkpoint set;

Position mapping model coefficient calculation unit (250)

Using the set of control points and checkpoints obtained by the above units, a second-order polynomial transformation model is selected, and the coefficients of the position mapping model are calculated for each correction unit.

Among them, the second-order polynomial transformation model is expressed in matrix form by formula (1):

$\left(\begin{array}{c}x\\ \mathrm{the\; y}\end{array}\right)=\left(\begin{array}{ccccccccccc}1& r& c& r^2{c}^2& r\&times;c& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 1& r& c& r^2{c}^2& r\&times;c\end{array}\right)\left(\begin{array}{c}{m}_0\\ m_1\\ m_2\\ m_3\\ m_4\\ m_5\\ {\mathrm{no}}_0\\ {\mathrm{no}}_1\\ {\mathrm{no}}_2\\ {\mathrm{no}}_3\\ {\mathrm{no}}_4\\ {\mathrm{no}}_5\end{array}\right)$ Formula 1)

Among them, r, c are the geographical coordinates of POI of a certain website, x, y are the geographical coordinates of POI of another website, m ₀ , m ₁ , m ₂ , m 3 , m ₄ , m ₅ , n ₀ , n _{1 ,} n ₂ , n ₃ , n ₄ , and n ₅ are position mapping coefficients in the second-order polynomial model, and the geographic coordinates are longitude and latitude.

Preferably, in establishing the set of correction units, the geographical interval is 4°, 2°, 1°, 30' or 15'.

Preferably, in the area splitting unit, each correction unit is evenly split into a 6*6 coordinate subgrid to form a control point coordinate subgrid, and each correction unit is evenly split into a 12*12 Coordinate subgrid, which constitutes the coordinate subgrid of checkpoints.

Preferably, in the location mapping model coefficient solving unit, firstly, the error equation is listed according to the second-order polynomial transformation model, and then the location mapping coefficient in the error equation is calculated according to the method of indirect adjustment.

Preferably, it also includes: a position mapping accuracy evaluation unit (260)

For the correction unit set, traverse each correction unit, use the position mapping coefficient Δ obtained by each correction unit to calculate the middle error, maximum error and minimum error of the check point set, and complete the POI position mapping accuracy of multiple sites assessment.

The invention evenly divides POI positions into several units, uniformly selects control points, comprehensively utilizes semantic matching and least squares methods to fully automatically select control points, and uses a "divide and conquer" strategy to establish an optimal position mapping model.

Description of drawings

Fig. 1 is the flow chart of the multi-site POI position mapping method according to the specific embodiment of the present invention;

Fig. 2 is a schematic diagram of splitting the correction unit set and the control points and checkpoint areas in the correction unit according to a specific embodiment of the present invention;

Figure 3 is a roadmap for automatic acquisition of multi-site POI control point sets with the same name;

Fig. 4 is a roadmap for automatic acquisition of multi-site POI checkpoint sets with the same name;

5 is a block diagram of a multi-site POI location mapping device according to a specific embodiment of the present invention;

Figure 6 is a map of the location mapping results of multi-site POIs.

Detailed ways

The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, but not to limit the present invention. In addition, it should be noted that, for the convenience of description, only some structures related to the present invention are shown in the drawings but not all structures.

The principle of the present invention is: firstly, according to the geographic scope of the multi-site POI, a correction unit set is established. In each correction unit, the control point set and the check point set are obtained by means of semantic matching method through region splitting. The location mapping coefficients are calculated using the control point set by the least square principle, and finally the location mapping accuracy is evaluated.

Referring to FIG. 1 , a flow chart of a multi-site POI position mapping method according to the present invention is disclosed.

Specifically, it includes the following steps:

S110: Establish a correction unit set

The minimum and maximum geographic coordinates are calculated respectively according to the geographic coordinate set of the POI point set, and the minimum and maximum geographic coordinates obtained are the geographic range of the POI point set, and a certain geographic interval is used to divide the geographic range of the POI point set into N*N small A grid, the N*N grid ensemble constitutes a set of correction units.

Each POI contains geographic coordinates, and the complete set of geographic coordinates of POIs in the POI point set constitutes the geographic coordinate set of the POI point set.

The minimum and maximum geographic coordinates are respectively calculated according to the geographic coordinate set of the POI point set, and the obtained minimum and maximum geographic coordinates are the geographic range of the POI point set.

In this embodiment, the Beijing area (39.26°N, 115.25°E, 41.03°N, 117.30°E) is used as the test area, and the test data are two sets of POI data sets from different websites.

North latitude 39.26° and 41.03° are the minimum and maximum geographic coordinates of the Beijing POI point set in the latitude direction, constituting the geographical range of the POI point set in the latitude direction; east longitude 115.25° and 117.30° are the minimum and maximum coordinates of the Beijing POI point set in the longitude direction. The maximum geographic coordinates constitute the geographic range of the POI point set in the longitude direction.

The geographic range of the POI point set is divided into N*N small grids by using a certain geographic interval. The smaller the geographical interval, the larger the value of N. Each small grid corresponds to a correction unit, and the full set of N*N grids constitutes a set of correction units.

In this embodiment, correction unit sets can be respectively established according to five geographical intervals of 4°, 2°, 1°, 30', and 15'.

The commonly used geographic grid intervals are shown in Table 1.

Table 1 Geographic grid interval

S120: Perform region splitting in each correction unit of the correction unit set

After the correction unit set is established, in order to ensure the uniform distribution of control points in each correction unit of the correction unit set, each correction unit is evenly split into multiple coordinate sub-grids to form the coordinate sub-grid of control points, In order to ensure the uniform distribution of check points in each correction unit of the correction unit set, each correction unit is evenly split into multiple coordinate sub-grids to form the coordinate sub-grids of check points.

In a preferred embodiment, each correction unit is evenly split into 6*6 coordinate sub-grids to form a coordinate sub-grid of control points. Each correction unit is evenly split into 12*12 coordinate sub-grids to form the coordinate sub-grids of checkpoints, as shown in Figure 2.

S130: Semantic-based control point matching

For POI names from different websites, use the Chinese word segmentation tool to perform word segmentation, and then determine whether they are the same control point according to the phrase string after word segmentation.

For example, in this embodiment, the POI name a of a certain website is the National Bureau of Surveying and Mapping Information, and the POI name b of another website is the National Bureau of Surveying and Mapping. The phrase string after participle a of the POI name of a certain website is {{country}, {surveying and mapping}, {information}, {bureau}}, and the phrase string after participle b of the POI name of another website is {{country}, { Surveying}, {bureau}} both refer to the National Bureau of Surveying, Mapping and Geographic Information, but the names are inconsistent. It can be determined that the two are the same place as the control point object.

S140: Automatic and uniform acquisition of control point sets and checkpoint sets of POIs with the same name for multiple sites

In the range of each control point subgrid, the control points are automatically obtained by using the method of semantic matching, and added to the set of control points, as shown in Figure 4.

Similarly, in each checkpoint subgrid range, at least one checkpoint is automatically obtained by using the semantic matching method, and added to the checkpoint set, as shown in Figure 5.

This step is to use the method of steps S110-S130 to traverse all control point sets and check point sets to ensure that each grid gets corresponding control points and check points.

Those skilled in the art can know that step S130 can also be directly integrated into step S140, directly in each control point sub-grid and each check point sub-grid, using a matching method to make each grid obtain a corresponding control points and checkpoints.

S150: Calculating position mapping model coefficients

Solve for positional model coefficients using control points.

Establishing a location mapping model between different website POIs is equivalent to establishing a location mapping relationship, and the location mapping model is expressed by the location mapping model coefficients. Therefore, by determining the location mapping model coefficients through calculation, the location mapping model between different website POIs can be better established.

The control points are used to calculate the position mapping coefficients, and the check points are used to evaluate the model reliability of the position mapping relationship.

The calculation of position mapping model coefficients depends on sufficient control points and checkpoints, the control points are used to calculate the position mapping coefficients, and the checkpoints are used to evaluate the accuracy of position mapping.

In a specific embodiment, the control points and check point sets obtained in the above steps are used, and a second-order polynomial transformation model is selected to calculate the position mapping model coefficients for each correction unit.

Among them, the second-order polynomial transformation model is expressed in matrix form by formula (1):

$\left(\begin{array}{c}x\\ \mathrm{the\; y}\end{array}\right)=\left(\begin{array}{ccccccccccc}1& r& c& r^2{c}^2& r\&times;c& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 1& r& c& r^2{c}^2& r\&times;c\end{array}\right)\left(\begin{array}{c}{m}_0\\ m_1\\ m_2\\ m_3\\ m_4\\ m_5\\ {\mathrm{no}}_0\\ {\mathrm{no}}_1\\ {\mathrm{no}}_2\\ {\mathrm{no}}_3\\ {\mathrm{no}}_4\\ {\mathrm{no}}_5\end{array}\right)$ Formula 1)

Among them, r, c are the geographical coordinates of POI of a certain website, namely longitude and latitude, x, y are the geographical coordinates of POI of another website, namely longitude and latitude, m ₀ , m ₁ , m ₂ , m ₃ , m ₄ , m ₅ , n ₀ , n ₁ , n ₂ , n ₃ , n ₄ , and n ₅ are the position mapping coefficients in the second-order polynomial model. The location mapping relationship between two different websites can be established through this transformation model.

Specifically, the basic principle of solving the position mapping model is equivalent to the indirect adjustment method of measurement. First, the error equation is listed according to the second-order polynomial transformation model, and then the undetermined parameters in the error equation are calculated according to the indirect adjustment method (position mapping coefficient).

Compared with the homogeneous equation, the error equation means that the number of equations is more than the number of unknowns. According to the principle of least squares, the most probable value of unknowns is obtained.

In the present embodiment, adopt second-order polynomial model to obtain the most probable value of position mapping coefficient, list error equation as follows by formula (1):

$\left(\begin{array}{c}{v}_x\\ v_{\mathrm{the\; y}}\end{array}\right)=\left(\begin{array}{ccccccccccc}1& r& c& r^2{c}^2& r\&times;c& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 1& r& c& r^2{c}^2& r\&times;c\end{array}\right)\left(\begin{array}{c}{\mathrm{\&Delta;m}}_0\\ {\mathrm{\&Delta;m}}_1\\ {\mathrm{\&Delta;m}}_2\\ {\mathrm{\&Delta;m}}_3\\ {\mathrm{\&Delta;m}}_4\\ {\mathrm{\&Delta;m}}_5\\ {\mathrm{\&Delta;\; n}}_0\\ {\mathrm{\&Delta;\; n}}_1\\ {\mathrm{\&Delta;n}}_2\\ {\mathrm{\&Delta;\; n}}_3\\ {\mathrm{\&Delta;n}}_4\\ {\mathrm{\&Delta;\; n}}_5\end{array}\right)-\left(\begin{array}{c}x\\ \mathrm{the\; y}\end{array}\right)$ Formula (2)

Namely: v=BΔ-1

in, $B=\left(\begin{array}{ccccccccccc}1& r& c& r^2{c}^2& r\&times;c& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 1& r& c& r^2{c}^2& r\&times;c\end{array}\right),$

$\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\left(\begin{array}{c}\mathrm{<span\; class="notranslate">\; x</span>}\\ \mathrm{<span\; class="notranslate">\; the\; y</span>}\end{array}\right)$

m _i = m _{i initial} + Δm _i

Among them, V includes V _x , V _y , indicating the correction of the geographical coordinates of another website POI, that is, longitude x, latitude y, B is the coefficient matrix of the error equation, B passes through the geographical coordinates of a certain website POI, longitude r, latitude y, Latitude c, calculated; l includes x and y, which is the geographic coordinates of another website POI, longitude x, latitude y. The position mapping coefficient m _i is composed of the initial value m _i of the position mapping coefficient and the correction number Δm _i . Generally speaking, the initial calculation value m _{i of the position mapping coefficient is initially} set to 0, and at this time m _i =Δm _i , so Δm _i is used to represent the position mapping coefficient m _i below.

According to the calculation method of indirect adjustment, by matrix operation:

Δ＝(B ^T B) ^-1 B ^T 1

Solve to calculate the position mapping coefficient Δ.

The basic idea of this step is to establish the error equation based on the least squares adjustment method. The coordinates of the known control points are substituted into the equation to solve the position mapping coefficients.

After calculating the position mapping coefficients, directly input the POI coordinates to obtain the coordinates after position mapping. For example, after the location mapping coefficients between two different websites have been calculated, after inputting the geographical coordinates of a certain website POI (namely longitude r, latitude c), use formula 1 to obtain the longitude coordinates of another website POI through matrix operation x and latitude coordinates y.

S160: Position Mapping Accuracy Evaluation

For the correction unit set, traverse each correction unit, use the position mapping coefficient Δ obtained by each correction unit to calculate the middle error, maximum error and minimum error of the check point set, and complete the POI position mapping accuracy of multiple sites assessment.

That is, each correction unit calculates a corresponding position coefficient, and the position coefficient of the correction unit is used in each correction unit to calculate the calculation error of the correction unit through checkpoints. Those skilled in the art should know that the method of using check points to calculate the error is any other conventional method in the field, for example, the conventional method of measurement adjustment such as: conditional adjustment with parameters, Indirect adjustment; the conventional method of accuracy assessment, such as: calculating the error by introducing the control point set into the model.

In this embodiment, the accuracy is shown in Table 2, the error unit is km, and the evaluation result of the position mapping accuracy is shown in FIG. 6 .

Table 2 Accuracy statistics table

In this embodiment, compared with the current manual and semi-automatic methods of selecting control points required for establishing a multi-site POI location mapping model, the automatic and uniform selection of control points is fully realized, and the work efficiency is greatly improved.

In the present embodiment, as shown in table 2, adopt traditional method to set up POI position mapping relation of different websites, the middle error of longitude direction is 1.25km, and the middle error of latitude direction is 0.074km; Five-divided geographical interval as an example), the medium error in the longitude direction is 0.88m, and the medium error in the latitude direction is 0.01m. The accuracy of the location mapping has been improved from the kilometer level to the decimeter level, which meets the actual application requirements for POI location mapping accuracy. Require.

Referring to Fig. 5, the present invention also discloses a multi-site POI position mapping device, including the following units

Correction unit set establishment unit 210:

The minimum and maximum geographic coordinates are calculated respectively according to the geographic coordinate set of the POI point set, and the minimum and maximum geographic coordinates obtained are the geographic range of the POI point set, and a certain geographic interval is used to divide the geographic range of the POI point set into N*N small A grid, the full set of N*N grids constitutes a set of correction units;

Region splitting unit 220:

After the correction unit set is established, in order to ensure the uniform distribution of control points in each correction unit of the correction unit set, each correction unit is evenly split into multiple coordinate sub-grids to form the coordinate sub-grid of control points, In order to ensure the uniform distribution of check points in each correction unit of the correction unit set, each correction unit is evenly split into multiple coordinate sub-grids to form the coordinate sub-grid of check points;

Semantic-based control point matching unit 230:

For POI names from different websites, use the Chinese word segmentation tool to perform word segmentation, and then determine whether they are the same control point according to the phrase string after word segmentation;

The control point set and checkpoint set acquisition unit 240:

In the range of each control point sub-grid, use the method of semantic matching to automatically obtain control points and add them to the set of control points. In the range of each check point sub-grid, use the method of semantic matching to automatically obtain at least one check point , and join the checkpoint set;

Position mapping model coefficient calculation unit 250:

Using the set of control points and checkpoints obtained by the above units, a second-order polynomial transformation model is selected, and the coefficients of the position mapping model are calculated for each correction unit.

Among them, the second-order polynomial transformation model is expressed in matrix form by formula (1):

$\left(\begin{array}{c}x\\ \mathrm{the\; y}\end{array}\right)=\left(\begin{array}{ccccccccccc}1& r& c& r^2{c}^2& r\&times;c& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 1& r& c& r^2{c}^2& r\&times;c\end{array}\right)\left(\begin{array}{c}{m}_0\\ m_1\\ m_2\\ m_3\\ m_4\\ m_5\\ {\mathrm{no}}_0\\ {\mathrm{no}}_1\\ {\mathrm{no}}_2\\ {\mathrm{no}}_3\\ {\mathrm{no}}_4\\ {\mathrm{no}}_5\end{array}\right)$ Formula 1)

Among them, r, c are the geographical coordinates of POI of a certain website, x, y are the geographical coordinates of POI of another website, m ₀ , m ₁ , m ₂ , m 3 , m ₄ , m ₅ , n ₀ , n _{1 ,} n ₂ , n ₃ , n ₄ , and n ₅ are position mapping coefficients in the second-order polynomial model, and the geographic coordinates are longitude and latitude.

Wherein, in establishing the correction unit set unit, the geographical interval is 4°, 2°, 1°, 30' or 15'.

Among them, in the area splitting unit, each correction unit is evenly split into 6*6 coordinate sub-grids to form the coordinate sub-grid of control points, and each correction unit is evenly split into 12*12 coordinates Subgrid, which constitutes the coordinate subgrid of the checkpoints.

Wherein, in the location mapping model coefficient solving unit, the error equation is first listed according to the second-order polynomial transformation model, and then the location mapping coefficient in the error equation is calculated according to the method of indirect adjustment;

Among them, the most probable value of the position mapping coefficient is obtained by using the second-order polynomial model, and the error equation is listed by formula (1) as follows

$\left(\begin{array}{c}{v}_x\\ v_{\mathrm{the\; y}}\end{array}\right)=\left(\begin{array}{ccccccccccc}1& r& c& r^2{c}^2& r\&times;c& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 1& r& c& r^2{c}^2& r\&times;c\end{array}\right)\left(\begin{array}{c}{\mathrm{\&Delta;m}}_0\\ {\mathrm{\&Delta;m}}_1\\ {\mathrm{\&Delta;m}}_2\\ {\mathrm{\&Delta;m}}_3\\ {\mathrm{\&Delta;m}}_4\\ {\mathrm{\&Delta;m}}_5\\ {\mathrm{\&Delta;\; n}}_0\\ {\mathrm{\&Delta;n}}_1\\ {\mathrm{\&Delta;n}}_2\\ {\mathrm{\&Delta;\; n}}_3\\ {\mathrm{\&Delta;n}}_4\\ {\mathrm{\&Delta;n}}_5\end{array}\right)-\left(\begin{array}{c}x\\ \mathrm{the\; y}\end{array}\right)$ Formula (2)

Namely: v=BΔ-1

in, $B=\left(\begin{array}{ccccccccccc}1& r& c& r^2{c}^2& r\&times;c& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 1& r& c& r^2{c}^2& r\&times;c\end{array}\right),$

m _i = m _{i initial} + Δm _i

Among them, V includes V _x , V _y , indicating the correction of the geographical coordinates of another website POI, that is, longitude x, latitude y, B is the coefficient matrix of the error equation, B passes through the geographical coordinates of a certain website POI, longitude r, latitude y, Latitude c, calculated; I includes x and y, which is the geographical coordinates of another website POI, longitude x, latitude y. The position mapping coefficient m _i is composed of the initial value m _i of the position mapping coefficient and the correction number Δm _i . Generally speaking, the initial calculation value m _{i of the position mapping coefficient is initially} set to 0, and at this time m _i =Δm _i , so Δm _i is used to represent the position mapping coefficient m _i below.

According to the calculation method of indirect adjustment, by matrix operation:

Δ＝(B ^T B) ^-1 B ^T 1

Solve to calculate the position mapping coefficient Δ.

Wherein, it also includes a position mapping accuracy evaluation unit (260)

For the correction unit set, traverse each correction unit, use the position mapping coefficient Δ obtained by each correction unit to calculate the middle error, maximum error and minimum error of the check point set, and complete the POI position mapping accuracy of multiple sites assessment.

That is, each correction unit has a set of position coefficients, and the position coefficient of the correction unit is used in each correction unit, and the calculation error of the correction unit is calculated by using checkpoints.

The invention evenly divides POI positions into several units, uniformly selects control points, comprehensively utilizes semantic matching and least squares methods to fully automatically select control points, and uses a "divide and conquer" strategy to establish an optimal position mapping model.

Obviously, those skilled in the art should understand that each unit or each step of the present invention described above can be realized by a general-purpose computing device, and they can be concentrated on a single computing device. Alternatively, they can be implemented by a computer device executable program codes, so that they can be stored in a storage device and executed by a computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps can be made into a single integrated circuit module for realization. As such, the present invention is not limited to any specific combination of hardware and software.

The above content is a further detailed description of the present invention in conjunction with specific preferred embodiments. It cannot be determined that the specific embodiments of the present invention are limited thereto. Under the present invention, several simple deduction or substitutions can also be made, all of which should be considered as belonging to the protection scope of the present invention determined by the submitted claims.

Claims (10)

1. A multi-site POI position mapping method, comprising the steps: S110. Establish a correction unit set: The minimum and maximum geographic coordinates are calculated respectively according to the geographic coordinate set of the POI point set, and the minimum and maximum geographic coordinates obtained are the geographic range of the POI point set, and a certain geographic interval is used to divide the geographic range of the POI point set into N*N small A grid, the full set of N*N grids constitutes a set of correction units; S120. Perform region splitting in each correction unit of the correction unit set: After the correction unit set is established, in order to ensure the uniform distribution of control points in each correction unit of the correction unit set, each correction unit is evenly split into multiple coordinate sub-grids to form the coordinate sub-grid of control points, In order to ensure the uniform distribution of check points in each correction unit of the correction unit set, each correction unit is evenly split into multiple coordinate sub-grids to form the coordinate sub-grid of check points; S130. Semantic-based control point matching: For POI names from different websites, use the Chinese word segmentation tool to perform word segmentation, and then determine whether they are the same control point according to the phrase string after word segmentation; S140. Automatic and uniform acquisition of control point sets and check point sets of the same name of POIs of multiple websites: In the range of each control point sub-grid, use the method of semantic matching to automatically obtain control points and add them to the set of control points. In the range of each check point sub-grid, use the method of semantic matching to automatically obtain at least one check point , and join the checkpoint set; S150. Solving position mapping model coefficients Using the control points and check point sets obtained in the above steps, select the second-order polynomial transformation model, and solve the position mapping model coefficients for each correction unit respectively; Among them, the second-order polynomial transformation model is expressed in matrix form by formula (1): $\left(\begin{array}{c}x\\ \mathrm{the\; y}\end{array}\right)=\left(\begin{array}{ccccccccccc}1& r& c& r^2{c}^2& r\&times;c& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 1& r& c& r^2{c}^2& r\&times;c\end{array}\right)\left(\begin{array}{c}{m}_0\\ m_1\\ m_2\\ m_3\\ m_4\\ m_5\\ {\mathrm{no}}_0\\ {\mathrm{no}}_1\\ {\mathrm{no}}_2\\ {\mathrm{no}}_3\\ {\mathrm{no}}_4\\ {\mathrm{no}}_5\end{array}\right)$ Formula 1) Among them, r, c are the geographical coordinates of POI of a certain website, x, y are the geographical coordinates of POI of another website, m ₀ , m ₁ , m ₂ , m 3 , m ₄ , m ₅ , n ₀ , n _{1 ,} n ₂ , n ₃ , n ₄ , and n ₅ are position mapping coefficients in the second-order polynomial model, and the geographic coordinates are longitude and latitude. 2. multi-site POI position mapping method according to claim 1, is characterized in that: In the step of establishing a correction unit set, the geographical interval is 4°, 2°, 1°, 30' or 15'. 3. multi-site POI position mapping method according to claim 1, is characterized in that: In the region splitting step in each correction unit of the correction unit set, each correction unit is evenly split into a 6*6 coordinate sub-grid to form a coordinate sub-grid of control points, and each correction unit is evenly divided into Split into 12*12 coordinate sub-grids to form the coordinate sub-grids of checkpoints. 4. multi-site POI position mapping method according to claim 1, is characterized in that: In the step of calculating the coefficients of the position mapping model, the error equation is first listed according to the second-order polynomial transformation model, and then the position mapping coefficient in the error equation is calculated according to the method of indirect adjustment; Among them, the most probable value of the position mapping coefficient is obtained by using the second-order polynomial model, and the error equation is listed by formula (1) as follows $\left(\begin{array}{c}{v}_x\\ v_{\mathrm{the\; y}}\end{array}\right)=\left(\begin{array}{ccccccccccc}1& r& c& r^2{c}^2& r\&times;c& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 1& r& c& r^2{c}^2& r\&times;c\end{array}\right)\left(\begin{array}{c}{\mathrm{\&Delta;m}}_0\\ {\mathrm{\&Delta;m}}_1\\ {\mathrm{\&Delta;m}}_2\\ {\mathrm{\&Delta;m}}_3\\ {\mathrm{\&Delta;m}}_4\\ {\mathrm{\&Delta;m}}_5\\ {\mathrm{\&Delta;\; n}}_0\\ {\mathrm{\&Delta;n}}_1\\ {\mathrm{\&Delta;n}}_2\\ {\mathrm{\&Delta;\; n}}_3\\ {\mathrm{\&Delta;\; n}}_4\\ {\mathrm{\&Delta;n}}_5\end{array}\right)-\left(\begin{array}{c}x\\ \mathrm{the\; y}\end{array}\right)$ Formula (2) Namely: v=BΔ-1 in, $B=\left(\begin{array}{ccccccccccc}1& r& c& r^2{c}^2& r\&times;c& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 1& r& c& r^2{c}^2& r\&times;c\end{array}\right),$ $\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\left(\begin{array}{c}\mathrm{<span\; class="notranslate">\; x</span>}\\ \mathrm{<span\; class="notranslate">\; the\; y</span>}\end{array}\right)$ m _i = m _{i initial} + Δm _i V includes V _x , V _y , indicating the correction of the geographical coordinates of another website POI, B is the coefficient matrix of the error equation, and B is calculated by the geographical coordinates of a certain website POI; 1 includes x and y, which is another website POI Geographical coordinates, m _{i initially} represents the initial value of the calculation of the position mapping coefficient, and Δm _i represents the correction number of the calculation of the position mapping coefficient; According to the calculation method of indirect adjustment, by matrix operation: Δ＝(B ^T B) ^-1 B ^T 1 Solve to calculate the position mapping coefficient Δ. 5. multi-site POI position mapping method according to claim 1, is characterized in that: After S150, there is also S160: Position Mapping Accuracy Evaluation, For the correction unit set, traverse each correction unit, use the position mapping coefficient Δ obtained by each correction unit to calculate the middle error, maximum error and minimum error of the check point set, and complete the POI position mapping accuracy of multiple sites assessment. 6. A multi-site POI position mapping device, comprising the following units: Correction unit set establishment unit (210): The minimum and maximum geographic coordinates are calculated respectively according to the geographic coordinate set of the POI point set, and the minimum and maximum geographic coordinates obtained are the geographic range of the POI point set, and a certain geographic interval is used to divide the geographic range of the POI point set into N*N small A grid, the full set of N*N grids constitutes a set of correction units; Region split unit (220): After the correction unit set is established, in order to ensure the uniform distribution of control points in each correction unit of the correction unit set, each correction unit is evenly split into multiple coordinate sub-grids to form the coordinate sub-grid of control points, In order to ensure the uniform distribution of check points in each correction unit of the correction unit set, each correction unit is evenly split into multiple coordinate sub-grids to form the coordinate sub-grid of check points; Semantic-based control point matching unit (230): For POI names from different websites, use the Chinese word segmentation tool to perform word segmentation, and then determine whether they are the same control point according to the phrase string after word segmentation; Control point set, checkpoint set acquisition unit (240): In the range of each control point sub-grid, use the method of semantic matching to automatically obtain control points and add them to the set of control points. In the range of each check point sub-grid, use the method of semantic matching to automatically obtain at least one check point , and join the checkpoint set; Position mapping model coefficient calculation unit (250) Using the control points and check point sets obtained by the above units, select the second-order polynomial transformation model, and solve the position mapping model coefficients for each correction unit; Among them, the second-order polynomial transformation model is expressed in matrix form by formula (1): $\left(\begin{array}{c}x\\ \mathrm{the\; y}\end{array}\right)=\left(\begin{array}{ccccccccccc}1& r& c& r^2{c}^2& r\&times;c& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 1& r& c& r^2{c}^2& r\&times;c\end{array}\right)\left(\begin{array}{c}{m}_0\\ m_1\\ m_2\\ m_3\\ m_4\\ m_5\\ {\mathrm{no}}_0\\ {\mathrm{no}}_1\\ {\mathrm{no}}_2\\ {\mathrm{no}}_3\\ {\mathrm{no}}_4\\ {\mathrm{no}}_5\end{array}\right)$ Formula 1) Among them, r, c are the geographical coordinates of POI of a certain website, x, y are the geographical coordinates of POI of another website, m ₀ , m ₁ , m ₂ , m 3 , m ₄ , m ₅ , n ₀ , n _{1 ,} n ₂ , n ₃ , n ₄ , and n ₅ are position mapping coefficients in the second-order polynomial model, and the geographic coordinates are longitude and latitude. 7. The multi-site POI position mapping device according to claim 6, characterized in that: In establishing the set of correction units, the geographical interval is 4°, 2°, 1°, 30' or 15'. 8. The multi-site POI position mapping device according to claim 6, characterized in that: In the area splitting unit, each correction unit is evenly split into a 6*6 coordinate subgrid to form a control point coordinate subgrid, and each correction unit is evenly split into a 12*12 coordinate subgrid The grid constitutes the coordinate subgrid of the checkpoints. 9. The multi-site POI position mapping device according to claim 6, characterized in that: In the location mapping model coefficient solving unit, the error equation is first listed according to the second-order polynomial transformation model, and then the location mapping coefficient in the error equation is calculated according to the method of indirect adjustment; Among them, the most probable value of the position mapping coefficient is obtained by using the second-order polynomial model, and the error equation is listed by formula (1) as follows $\left(\begin{array}{c}{v}_x\\ v_{\mathrm{the\; y}}\end{array}\right)=\left(\begin{array}{ccccccccccc}1& r& c& r^2{c}^2& r\&times;c& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 1& r& c& r^2{c}^2& r\&times;c\end{array}\right)\left(\begin{array}{c}{\mathrm{\&Delta;m}}_0\\ {\mathrm{\&Delta;m}}_1\\ {\mathrm{\&Delta;m}}_2\\ {\mathrm{\&Delta;m}}_3\\ {\mathrm{\&Delta;m}}_4\\ {\mathrm{\&Delta;m}}_5\\ {\mathrm{\&Delta;\; n}}_0\\ {\mathrm{\&Delta;\; n}}_1\\ {\mathrm{\&Delta;\; n}}_2\\ {\mathrm{\&Delta;n}}_3\\ {\mathrm{\&Delta;n}}_4\\ {\mathrm{\&Delta;n}}_5\end{array}\right)-\left(\begin{array}{c}x\\ \mathrm{the\; y}\end{array}\right)$ Formula (2) Namely: v=BΔ-1 in, $B=\left(\begin{array}{ccccccccccc}1& r& c& r^2{c}^2& r\&times;c& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 1& r& c& r^2{c}^2& r\&times;c\end{array}\right),$ m _i = m _{i initial} + Δm _i V includes V _x , V _y , indicating the correction of the geographical coordinates of another website POI, B is the coefficient matrix of the error equation, and B is calculated by the geographical coordinates of a certain website POI; 1 includes x and y, which is another website POI Geographical coordinates, m _{i initially} represents the initial value of the calculation of the position mapping coefficient, and Δm _i represents the correction number of the calculation of the position mapping coefficient; According to the calculation method of indirect adjustment, by matrix operation: Δ＝(B ^T B) ^-1 B ^T 1 Solve to calculate the position mapping coefficient Δ. 10. The multi-site POI position mapping device according to claim 1, characterized in that: Also included: Position Mapping Accuracy Assessment Unit (260) For the correction unit set, traverse each correction unit, use the position mapping coefficient Δ obtained by each correction unit to calculate the middle error, maximum error and minimum error of the check point set, and complete the POI position mapping accuracy of multiple sites assessment.