GB2416222A

GB2416222A - Cleaning spatial data

Info

Publication number: GB2416222A
Application number: GB0415494A
Authority: GB
Inventors: Dawn Moffat
Original assignee: LASER SCAN
Current assignee: LASER SCAN
Priority date: 2004-07-12
Filing date: 2004-07-12
Publication date: 2006-01-18
Also published as: GB0415494D0

Abstract

A method and apparatus for improving the accuracy of a user's spatial data. The data is split into two kinds of features, those found in the background, for example maps produced by Ordnance Survey (RTM) in Great Britain, and those found in the foreground which relate to areas with which the user is concerned. The current invention, by removing errors both 12 and 14, both within features and between features, in the foreground data allows a user to more accurately define the areas with which they are concerned. This may be further enhanced by associating foreground and background features. The method is particularly helpful when looking at changes within the background data, such as those caused by the use of GPS within map production where background data has shifted significantly as well as becoming more accurate. This is because the foreground data requires shifting to match up with the background data.

Description

. : :..

241 6222 Apparatus for Improving the Accuracv of Spatial Data This invention relates to an apparatus and a method for processing spatial data. The invention is particularly applicable to use within systems for improving the accuracy of mapped data.

Mapping agencies, such as Ordnance Survey in the UK, have produced 2 dimensional maps of land using historical 1:2500 scale mapping for many years. These topological maps, often referred to as base maps, have been used by other groups of people, such as a land registry, to record features such as buildings and areas of land significant to them. In marking foreground features significant to them users build up their own individual Geographical Information System (GIS). The features may be added individually or as a group, resulting in layers which may be selectively viewed by the user.

As may be expected if foreground features are entered manually into GlSs, small errors such as the repeated input of a single point inevitably occur. Whilst the small errors in the source data used to record foreground features may have been insignificant when the accuracy of base maps was much less, the use of GPS, which can measure accurately to +1metre, in the recording of maps for public use has made foreground feature accuracy much more important. Hence, it has become more important to identify and remove small manual errors within a user's data.

All that is easily identifiable and removed using tools that are currently available are basic geometric errors such as duplicate points and redundant points. Duplicate points are points, so close together, that the distance between them is regarded as insignificant by the ' c. e: : ë: ate system's user. Redundant points, on the other hand are points that are unnecessary for defining a feature. For example, if there are 4 points defining a straight line then by definition 2 of those points are unnecessary for defining the line and are, therefore, redundant. The presence of both of these points results in slower system performance, hence, it is desirable to remove them. Other common errors, also result in a slower system performance, but require a person to identify and remove them. This is time consuming and results in extra expense for a user.

Recently, mapping agencies have begun to use the global positioning system (GPS) to produce the standard base maps against which foreground features are marked. The use of GPS has led to the realisation that the original base maps created before GPS was used are incorrect and many features have shifted significantly on the new base maps compared to the original base maps. This has meant that any foreground features marked against the features of the original base map are no longer in the correct position. Hence many companies have been working to try to shift foreground data so that it is aligned with the new base map data. This is considered a preferable alternative to deleting the foreground data and re-entering it against the positionally corrected base map data.

Additionally, mapping agencies tend to improve their maps on a year-byyear basis. This allows them to take into account "real world" changes. For example, a person may have built an extension on a house or a field boundary may have been altered. The current methods for correcting foreground data to match base map data do not take any of these "real-world" alterations into account. Hence, updated foreground data may not be accurate.

This means that "real-world" alterations have to be added manually after the maps were updated. This is also time consuming and results in extra expense. l

8 66. 8 .

However, the previous common perception that the translation of data from the "original" position to that calculated using GPS would be sufficient to achieve this is proving to be false. At the moment one method being used is where points are identified and assigned a link file. The link file provides a series of mappings between the old and new positions of a subset of points and allows a shifted feature to be identified according to a particular co- ordinate. This is not sufficient to produce a good match, however, as only 1 in 10 or 1 in 20 points are identified leading to poor accuracy for points not included in the link file.

Another method is known as "rubber sheeting" where the foreground features are manipulated and stretched to fit the new base map. Neither the sole use of link files nor "rubber-sheeting" have proved very effective with many foreground and background features not matching up. Additionally small errors within the source data which references foreground features may increase due to the data manipulation and then become significant.

According to a first aspect of the invention there is provided a method of cleaning spatial data comprising receiving original foreground and background data the original foreground and background data each defining a feature, identifying an error within a feature, removing the error within the feature, identifying an error in a relationship between two or more features, removing the error in the relationship between the features, associating a foreground feature with one or more background features and presenting the processed data to a user. This has the advantage that any spatial data processed in this way has any errors and inconsistencies within it removed allowing a user to place greater reliance upon this data.

tee a:. .. ce: . : : Preferably, the error within a feature is identified by recognising two points closer together than a value pre-defined by the user, one or more points defining a line which does not deviate to a greater amount than a value pre-defined by the user from a straight line between two points, the angle between two lines is approaching zero (less than a value predefined by the user) and the length of the two lines is less than a value predefined by the user or the angle between one of the lines and a major line is less than the value predefined by the user.

Preferably, identifying an error in a relationship between two or more features comprises identifying two points closer together than a value predefined by the user, a point and a line closer together than a value predefined by the user or two lines closer together than a value predefined by the user. This allows a better match between foreground and background features.

Preferably, any errors are automatically removed resulting in a time, and therefore expense saving for the user.

Preferably, the method also comprises the step of translating or transforming foreground data to match with shifted background data. This allows the user to transform data collected before GPS was used into data which corresponds to background data collected using GPS.

Preferably the method further comprises the step of altering a foreground shape to match the shape of an associated background shape when the background shape is changed. This has the advantage that if background features are altered then the positions of foreground i Be: e: ee' : features can more easily be matched to the altered background features to which they correspond.

According to another aspect of the invention there is provided apparatus for cleaning spatial S data comprising an input adapted to receive original foreground and background data, the original foreground and background data comprising a feature, first identification means arranged to identify an error within a feature, first correction means arranged to remove the error within the feature, second identification means arranged to identify an error in a relationship between two or more features, second correction means arranged to remove the error in the relationship between the features, association means arranged to associate a foreground feature with one or more background features and an output adapted to present the processed data to a user.

According to a further aspect of the invention there is provided a computer readable medium carrying a computer program which when executed on a processor carries out the steps of receiving original foreground and background data the original foreground and background data comprising a feature, identifying an error within a feature, removing the error within the feature, identifying an error in a relationship between two or more features, removing the error in the relationship between the features, associating a foreground feature with one or more background features and presenting the processed data to a user.

According to a further aspect of the invention there is provided a server arranged to receive original foreground and background data the original foreground and background data comprising a feature, identify an error within a feature remove the error within the feature, identify an error in a relationship between two or more features, remove the error in the It t t C try It C t' It try It relationship between the features and associate a foreground feature with one or more background features output the cleaned data to a client.

According to a further aspect of the invention there is provided a client comprising a user interface, a server connection, output for sending foreground and background data to a server, server input for receiving cleaned spatial data and means for viewing cleaned spatial data.

Embodiments of the invention will now be described, by way of example, and with reference to the drawings in which: Figure 1 illustrates the preferred workflow employed when improving accuracy; Figure 2 illustrates a possible relationship between background and foreground features; Figure 3 illustrates types of geometric errors; Figure 4 illustrates a spike; l 5 Figure 5 illustrates a kick back; Figure 6 illustrates types of topological errors; and Figures 7a, 7b and 7c illustrate correction of different types of topological error.

With reference to Figure 1, the source data which records foreground features and the data which records the background features making up the base map are retrieved and transferred to a system where the methods described below may be carried out in step 10.

The data recording foreground and background features consists of numbers representing the co-ordinates that define a feature's shape. The system may either be a proprietary or non-proprietary system. '

.:e ce: ee. a: ee.: Preferably, the data is transferred into an open system such as an Oracle (TM) database in order to allow the data to be manipulated further. This has the advantage that the system is more accessible to users allowing them to adapt the system according to their needs.

It is advantageous to correct minor errors, such as the manual errors described above within the foreground features either with respect to themselves, other foreground features or background features. Preferably the errors in foreground features are corrected, firstly to remove any errors within a feature, step 12, and secondly, to remove any errors between features, step 14.

Once a foreground feature's minor errors are removed it is simpler to relate foreground features to background features in step 16. This is done because one foreground feature does not always correspond to one background feature. For example, in the case of Figure 2, the features labelled 24 may represent a front garden, house and back garden respectively. If the user was a land registry then it may wish to record different people's freehold. Therefore, for the land registry a foreground feature 26 may consist of the three features 24, as all this constitutes one person's freehold.

The association of foreground and background features of step 16 is desirable as it allows changes, known as real world changes, such as the addition of an extension or change in field boundaries to be logged automatically within the system. In addition to this any foreground feature changes required because of changes to base map data may be easily checked to ensure foreground features match up to the base map features with which they are associated. l e

. Any differences between the foreground and background data may be presented to a user and corrected accordingly by the user in steps 18 and 20 respectively. The processed data is then presented to the user for their use in step 22.

Foreground data errors are identified and corrected during the processes of geometric and topological cleaning (steps 12 and 14 respectively). With reference to Figure 3, there are four main types of geometric errors which can occur. These are duplicate points 28 where two points occur within a short distance. Redundant points where extra points not required to outline a feature occur, for example, multiple points along a straight line. Spikes 32 where there are two short lines having an angle approaching zero between them. Finally, there are kick backs 34 which are spikes where the angle between one of the lines next to the spike is approaching zero.

A duplicate point 28 occurs when two points have been entered on a single line so close together that they appear to mark the same point. This may occur for example, if the user entered a point twice. Duplicate points 28 may be scanned for on lines. This is done by taking the source data and comparing consecutive points. If the distance between the two points is less than the tolerance value defined by the user then the duplicate point 28 may be presented to the user, for example as a table of errors. Alternatively the user may choose for the duplicate point 28 to be automatically removed. The duplicate point 28 may be replaced by either one of the duplicate points as the distance between the points is defined as insignificant by the user.

If any section of a feature, in this case a straight line between points 38 and 40, is defined by more points than are necessary, then the extra "redundant" points, for example 30, may be l ' l c: c: r identified and removed leading to greater simplification and accuracy of the user's data. In order to do this the first 38 and last 40 points of the section of the feature are identified, i.e. the first and last pair of numbers in the data defining the feature. The data for the section of the feature, between the first and last points 38 and 40 respectively is scanned, as described for duplicate points. The point furthest in a perpendicular direction from the straight line 42 defined by the first 38 and last 40 points is identified. The perpendicular distance 44 between that point and the straight line 42 is calculated and if it is less than the tolerance value defined by the user then the presence of redundant points 30 is either presented to the user or automatically removed with the straight line 42 being used to join the first 38 and last 40 identified points.

This method may be repeated for different sections of the same feature or for different features. In order to ensure that the entire feature is covered when examining a feature for errors the last pair of numbers to be examined, rather than being the end numbers are overshot by one, i.e. are the second pair of numbers in the list.

Figure 4 illustrates a spike 32. The spike function acts to identify and remove any spikes occurring within a user's features. The function identifies a spike 32 if the angle between two lines, 46 and 48, is less than angle 50, which is specified by the user according to the accuracy of their original data and the length of each of the two lines 46 and 48 do not exceed the user defined tolerance value 36. If these conditions are met then the presence of a spike 32 may be presented to the user, for example as an entry in a table of errors.

Alternatively, the error may automatically be removed. l 1,

c: :: :: ace: : Figure 5 illustrates a kick back 34. The kick back function identifies kick backs 34 using the following rules: 1. the length of line 54 is less than that of line 52 or line 56; 2. the angle between line 52 and line 54, and line 54 and line 56 is less than that specified by the user; and 3. the distances between line 52 and the meeting point of line 54 and line 56 (point 55) and line 56 and the meeting point of line 52 and line 54 are less than the tolerance value 36.

The user may then specify a parameter in order to determine the way in which the identified kick back 34 is removed. For example, the user may choose to: (i) remove the leading pair of ordinates 53; (ii) the trailing pair of ordinates 57; or (iii) remove the pair of ordinates 55 furthest from the vector defined by the first 53 and last 57 pair of ordinates of the kick back being analysed.

Alternatively, at the beginning of the process of geometric cleaning the user may choose for all kick backs 34 to be rectified using one of the above methods. The user may choose to do this for all identified kick backs 34 or only for a group of features. For example, all identified kick backs 34 in buildings should have the trailing pair of ordinates 57 removed, or all kick backs 34 identified in a specified area of the map should have the pair of ordinates 55 furthest from the vector defined by the first 53 and last 55 pair of ordinates removed.

Finally, the tolerance values 36 defined by the user for the different types of geometric errors may have different values.

t ce,, c.e Topological cleaning is also carried out in order to improve the accuracy of the spatial relationships between features and areas of the data. With reference to Figure 6, types of topological errors include gaps 72 and slivers 70 where shapes either don't meet or overlap respectively and undershoots 68 and overshoots 66 where lines don't meet up or overlap respectively.

Preferably topological cleaning is carried out using a dynamic topological database such as Radius Topology where any changes are registered and persistently stored without the data having to constantly rebuild the topology map. The database preferably consists of a set of rules, an engine for modifying data according to the set of rules and a schema according to which data is stored within the database.

In order to carry out topological cleaning features are defined using three primitives: nodes 60 representing the end of a line or intersection of two or more lines, edges 62 representing lines and area boundaries and faces 64 representing areas. The primitives are stored and maintained in a database which may be edited directly by users. Where edges 62 overlap or do not quite meet are known as overshoots 66 and undershoots 68. Where faces 64 overlap or do not quite meet are known as slivers 70 and gaps 72.

Any edges 62 or nodes 60 that are shared between 2 or more features may be stored as a common edges or nodes within the database. In this way if one node's 60 position is altered then the geometries of all edges 62 ending in the node 60 are automatically altered to take this change into account. This is also true if a common edge's position is altered. This may d, I t e ace be carried out for edges and nodes which are common to base map and foreground features as well as shapes within the same plane.

As with geometric errors the user may set a tolerance value 36. If the error, for example an undershoot, occurs at less than the tolerance value 36 then the undershoot is automatically corrected, however, if it is greater than the tolerance value 36 then extra edges are created.

Alternatively, there may be a baseline tolerance value 36 defined within the system carrying out the topological cleaning. Preferably gaps and overlaps are automatically corrected.

Additionally, features may be grouped into classes, such as A and B roads and motorways may be grouped into three different classes. The classes may then be allocated different priorities, for example motorways may be given a higher priority than A roads which may have a higher priority than B roads. This has the advantage that features may be corrected according to their priority rating, as discussed below.

With reference to Figure 7, a user may set rules according to which the topological errors in the foreground features are automatically corrected. For example, undershoots 68, overshoots 66, gaps 72 or slivers 70 may be automatically corrected using one of the three following rules. Preferably, any feature being added to the database will move to match up with features already contained within the database. Each feature will also be allocated a new and old priority value and it is according to these values that the features are matched up using class priority rules, as discussed below.

Class priority rules define the priority ranking of features and therefore which feature will be moved and which will remain static. The rules consist of a class identifier and old and new l 1 : : I: I: : : priority values. Preferably, unless otherwise set by the user, the old priority value is of a higher priority than that of the new priority value and therefore any new features move to match features already contained within the database.

This is illustrated in Figure 7. Here, with reference to Figure 7a, if two nodes 60 from two different features are closer together than the tolerance value 36 set by the user then the features will be altered to share the nodes. In this instance a lower priority feature's node 60 is moved to create the shared node 74 with a higher priority feature's node 60.

Secondly, with reference to Figure 7b, if a lower priority feature's node 60 and a higher priority feature's edge 62 are closer together than the tolerance value 36 set by a user then the node 60 of the lower priority feature will be altered to split the edge 62 of the higher priority feature resulting in a shared node 74. Finally, with reference to Figure 7c if edges 62 of a lower and higher priority feature are closer together than the tolerance value 36 set by a user then the edge 62 of the lower priority feature will be corrected to split the edge 62 of the lower priority feature resulting in a shared node 74. The tolerance values 36 for each of these situations may be set as the same value or as different values. A default baseline tolerance may also be used if no tolerance values have been entered by the user.

Once the foreground data has been cleaned there should be a good match between the underlying base map data and the foreground data which includes the features the user wants to reference. The user may then associate base map features with foreground features. For example, a user may own a street of houses and has only outlined the property which is owned by them. In this case the foreground features will include a number . I t, ' 1 1 # of base map features, including the separate houses on the streets, the front and back gardens of those houses.

The relationship is stored within the database and may be used to ensure that any modified foreground data still represents the areas on the base map that it previously did. Preferably, the data is associated and stored automatically by the system, however, it is also possible to carry out these steps manually.

The association of foreground and background features allows the user to more readily alter the foreground features to take into account changes in the background due to real map changes. Additionally, it also allows the user to ensure that any foreground features entered into the system maintain the same relationships with features in the base map irrespective of any alterations in the base map. This is especially important if the base map data is being shifted, such as the shift that has occurred do to the use of GPS to measure the base map features.

After data association has occurred any gaps and overlaps existing between the base map and foreground features may be identified and corrected using the same methods as described with respect to topological cleaning. Namely, any gaps and overlaps are identified and if they are is less than the tolerance value specified by the user according to the accuracy of their original data then the presence of a gap or overlap may be presented to the user, for example as an entry in a table of errors. Alternatively, the error may automatically be removed. l

l:t.' : l :' :: Ill Any errors identified during the steps of geometric and topological cleaning and data association which cannot be automatically corrected by the system may be presented to the user in a separate table. This allows the user to manually fix the identified error.

If the base map data has altered significantly, as is the case with the change over from data used before GPS was used for surveying and after it was used for surveying then the user's data will need to be shifted in some way. This may be done as described above by using "link files" to identify points within the pre- and post-shift data and ensure that they are shifted by the same amount. Alternatively the foreground data may be stretched and manipulated in order to match up with the shifted background data. The more accurate foreground data will result in more accurate shifted data which requires less modification after shifting.

The shifted cleaned data, however, whilst being more accurate with respect to base map data being collected using GPS than data that has not been cleaned or associated, may still not be aligned accurately enough with the new base map features to be of use to the user.

In order to allow for this the steps involved with geometric and topological cleaning, data association and the detection of gaps and overlaps between the base map and foreground features are repeated, after the foreground data has been manipulated.

The repetition of these steps results in the foreground data corresponding more accurately with GPS measured background data. In order to ensure a good match between the foreground and base map features and also to provide a measure of quality of the shifted data the resulting data is compared with the data and computed data associations as it was before it was shifted to match up with GPS measured background data.

t t t e t t t t t t Whilst it is preferable for geometric cleaning, topological cleaning, data association and correction of any differences between background and foreground features to occur both before and after shifting of the data it is possible for these steps to be run either before or after the shifting of the data. Furthermore, in certain cases only some of the steps need to be run either before or after the data is shifted. For example if there are no geometric errors in the foreground data, geometric cleaning does not need to be carried out. The same is true for topological cleaning.

Although the methods described above are currently being utilised with respect to 2 dimensional spatial features, it is equally possible to improve the accuracy of 3-dimensional features using the techniques described above.

Summary

The data will match Land Line (TM) or OSGB MasterMap (TM) data collected using GPS exactly, allowing reliance upon the data for all business operations Topological snapping automatically adjusts data for Real World Change Tools available to manage the process effectively Setting of user appropriate rules allows a user control how their data is adjusted Ability to automatically process data in bulk Solution works with most mainstream GIS products Enables full exploitation of OSGB MasterMap (TM) Enables version control of changes within a database

Claims

. r e ce. .. . . . . . . CLAIMS

1. A method of cleaning spatial data comprising: - receiving original foreground and background data the original foreground

and background data each defining a feature;

- identifying an error within a feature; - removing the error within the feature; - identifying an error in a relationship between two or more features; - removing the error in the relationship between the features; and - associating a foreground feature with one or more background features thereby to produce cleaned spatial data

2. A method of cleaning spatial data as claimed in Claim 1 wherein the spatial data relates to maps.

3. A method of cleaning spatial data as claimed in Claims 1 or 2 wherein the error within a feature is identified by recognising two points closer together than a predetermined tolerance value.

4. A method of cleaning spatial data as claimed in Claims 1 or 2 wherein the error within a feature is identified by recognising one or more points defining a line which does not deviate to a greater amount than a predetermined tolerance value from a straight line between two points.

.,' e e ee e e e ee e I e e e e e ce e e e e es e e e e e as e e e e e

5. A method of cleaning spatial data as claimed in Claims 1 or 2 wherein the error within a feature is identified by recognising that the angle between two lines is less than a value predefined by the user and the length of the two lines is less than a predetermined tolerance value.

6. A method of cleaning spatial data as claimed in Claim 5 wherein the angle between one of the lines and a major line is less than a predetermined tolerance value.

7. A method of cleaning spatial data as claimed in Claim 5 wherein the error is removed using a function.

8. A method of cleaning spatial data as claimed in Claim 7 wherein a parameter of the function determines which pair of ordinates are removed.

9. A method of cleaning spatial data as claimed in Claim 8 wherein the parameter of the function is set for a group of features.

10. A method of cleaning spatial data as claimed in any preceding claim wherein the error within a feature is automatically removed.

11. A method of cleaning spatial data as claimed in any preceding claim wherein identifying an error in a relationship between two or more features comprises identifying two points closer together than a value predefined by the user.

. a c, ... C

12. A method of cleaning spatial data as claimed in any preceding claim wherein identifying an error in a relationship between two or more features comprises identifying a point and a line closer together than a value predefined by the user.

S

13. A method of cleaning spatial data as claimed in any preceding claim wherein identifying an error in a relationship between two or more features comprises identifying two lines closer together than a value predefined by the user.

14. A method of cleaning spatial data as claimed in Claims 11 to 13 wherein the error in a relationship between two or more features is automatically removed.

15. A method of cleaning spatial data as claimed in any preceding claim further comprising the step of assigning a common node to two or more lines sharing a node.

16. A method of cleaning spatial data as claimed in any preceding claim further comprising the step of assigning a common node to two or more features having intersecting lines at the point of intersection.

17. A method of cleaning spatial data as claimed in either of Claims 15 or 16 further comprising the step of altering the geometries of any feature having a common node when the common node's position is altered.

. se, e

18. A method of cleaning spatial data as claimed in any preceding claim further comprising the step of assigning a common line to two or more features sharing a line.

19. A method of cleaning spatial data as claimed in Claim 18 further comprising the step of altering the geometries of any feature having a common line when the common line's position is altered.

20. A method of cleaning spatial data as claimed in any preceding claim further comprising the step of: - aligning foreground data with shifted background data.

21. A method of cleaning spatial data comprising the step of aligning foreground data with shifted background data following the steps claimed in Claim 1.

22. A method of cleaning spatial data comprising the step of aligning foreground data with shifted background data before the steps claimed in Claim 1.

23. A method of cleaning spatial data as claimed in any of Claims 20 to 21 wherein the step of aligning foreground data with shifted background data comprises: - transforming the foreground data by stretching it to match up to shifted

background data.

24. A method of cleaning spatial data as claimed in Claim 14 wherein the data further undergoes the steps of Claim 1 after the method of Claim 14 or 15 is carried out. e e c

ë e

25. A method of cleaning spatial data as claimed in any preceding claim further comprising the step of altering a foreground feature to match the feature of an associated background feature when the background feature is changed.

26. Apparatus for cleaning spatial data comprising: - an input adapted to receive original foreground and background data the original foreground and background data comprising a feature; - first identification means arranged to identify an error within a feature; - first correction means arranged to remove the error within the feature; - second identification means arranged to identify an error in a relationship between two or more features; - second correction means arranged to remove the error in the relationship between the features; IS - association means arranged to associate a foreground feature with one or

more background features; and

- an output adapted to present the processed data to a user.

27. A computer readable medium carrying a computer program which when executed on a processor carries out the steps of: - receiving original foreground and background data the original foreground and baseman data comprising a feature; - identifying an error within a feature; - removing the error within the feature; - identifying an error in a relationship between two or more features; c.

: : : :: A:: :: at: ace - removing the error in the relationship between the features; - associating a foreground feature with one or more background features; and - presenting the processed data to a user.

28. A server arranged to: - receive original foreground and background data the original foreground and

background data comprising a feature;

- identify an error within a feature; - remove the error within the feature; - identify an error in a relationship between two or more features; - remove the error in the relationship between the features; and - associate a foreground feature with one or more background features output the cleaned data to a client.

29. A client comprising a: - user interface; - a server connection; output for sending foreground and background data to a server; - server input for receiving cleaned spatial data; and - means for viewing cleaned spatial data.

30. A method for processing data substantially as herein described with reference to and as shown in any combination of the accompanying drawings.

31. Apparatus for processing data substantially as herein described with reference to and as shown in any combination of the accompanying drawings.