DE102013207066A1 - Method for storing altitude profile data, as well as profile data and method for generating a height profile from these profile data - Google Patents

Abstract

The invention relates to a method for storing profile data for a predetermined height profile, comprising the following steps: (a) providing a regular grid (1) with polygonal grid fields (2) of a first level in one plane, wherein corner points (3) of the grid fields ( 2) represent raster points and a height value is stored for each raster point, (b) determining an error value for each raster field (2) which describes a deviation of an altitude profile approximated by the height values of the raster field (2) from the predetermined height profile, (c) Subdivide the grid fields (2), the error value of which exceeds a predetermined threshold value, into polygonal partial grid fields (4, 5), with at least the corner points (3) of the partial grid fields (4, 5) that do not have corner points (3) of the grid fields (2 ) coincide, elevation values are stored, (d) assigning one indicator to each grid (2), the indicator describing whether the ras subfield (2) is subdivided or not, and storing these indicators together with the elevation values, and (e) repeating steps (b) to (d) for all partial grid fields (4, 5), the partial grid fields (4, 5) being one i-th levels form grid fields (2) of an i + 1-th level, the height values being stored in a predetermined sequence, so that the height values are each assigned to a specific corner point (3) of the grid field (4, 5) based on the sequence and the indicators of a j-th level determine for which corner points (3) of the j + 1-th level elevation values are stored.

Description

The present invention relates to a method for storing profile data for a given height profile, as well as corresponding profile data and a method for generating a height profile from such profile data.

Profile data is used, for example, in geographic information systems to describe a terrain. A terrain is an area in three-dimensional space with special properties. A terrain can be steadily and locally reversibly uniquely reproduced in a plane, as long as there are no overhangs and as long as buildings are not part of the terrain. Thus, a terrain is a 2.5-dimensional structure. A terrain can thus be represented by a two-dimensional surface, wherein each point of the two-dimensional surface is assigned a height value.

In practice, elevation values are stored to a predetermined grid and the terrain is approximated by a plurality of contiguous triangles (triangulation). The vertices of a triangle are each represented by a height value.

It is known to arrange and store the elevation values in a regular or regular grid, in an irregular triangle mesh and with a semi-regular triangle mesh.

In a regular triangular grid, the elevation values are stored at each vertex of a triangle arranged in the plane, with all triangles in the plane being congruent to each other. Since a height value is needed for each vertex, the memory requirements are very high. For example, for a flat layer, the same amount of height values per surface must be stored as for a strongly contoured area. However, to ensure that the strongly contoured area can be displayed correctly, the individual triangles must not be too large. Data compression can somewhat reduce storage requirements.

Irregular triangular networks (TIN: Triangle Irregular Grid) store irregular triangular networks. With a method for producing such irregular triangular nets height profiles can be very well approximated, since in larger contoured areas smaller triangular areas and in less contoured areas larger triangular areas are provided. In this way, on the one hand, strongly contoured areas can be reproduced exactly and, on the other hand, flat areas with a small amount of data can be stored correctly. Such a triangle mesh solves on the one hand the problem of the memory requirement and on the other hand the problem of the accuracy of the representation of a height profile. The processing and production of such irregular triangular networks is very complex. The search for altitude values in irregular triangular meshes is complicated. Such an irregular triangle mesh is not hierarchical and it is always necessary to store a complete description of the triangular meshes.

Furthermore, ROAM networks (Real-time Optimal Adapting Meshes) are known. These nets are semi-regular triangular meshes representing a height field. They require significantly fewer triangles than a regular grid, but a little more than an irregular triangle mesh.

For the irregular triangular meshes and the semi-regular triangular meshes, the elevation values are stored along with the XY coordinates. This means that for each altitude point all three space coordinates are stored.

The present invention has for its object to provide a method for storing profile data for a given height profile, as well as corresponding profile data and a method for generating a height profile of such profile data, so that a predetermined height profile is accurately reproduced by the profile data, the amount of data low and the process is easy and quick to execute.

The object is achieved by the method for storing profile data, the corresponding profile data and the method for generating a height profile, as defined in the independent claims. Advantageous embodiments of the invention are defined in the respective subclaims.

• (a) Vorsehen eines regelmäßigen Rasters mit polygonalen Rasterfeldern eines nullten Levels in einer Ebene, wobei Eckpunkte der Rasterfelder Rasterpunkte darstellen, und zu jedem Rasterpunkt ein Höhenwert gespeichert wird,
• (b) Bestimmen eines Fehlerwertes für ein jedes Rasterfeld, der eine Abweichung eines durch die Höhenwerte des Rasterfelds approximierten Höhenprofils vom vorgegebenen Höhenprofil beschreibt,
• (c) Unterteilen der Rasterfelder, deren Fehlerwert einen vorbestimmten Schwellwert überschreitet, in polygonale Teilrasterfelder, wobei zumindest zu den Eckpunkten der Teilrasterfelder, die nicht mit Eckpunkten der Rasterfelder zusammen fallen, Höhenwerte gespeichert werden,
• (d) Zuordnen jeweils eines Indikators zu einem jeden Rasterfeld, wobei der Indikator beschreibt, ob das Rasterfeld unterteilt ist oder nicht, und Speichern dieser Indikatoren zusammen mit den Höhenwerten, und
• (e) Wiederholen der Schritte (b) bis (d) für alle Teilrasterfelder, wobei die Teilrasterfelder eines i-ten Levels Rasterfelder eines i + 1-ten Level bilden, wobei die Höhenwerte in einer vorbestimmten Reihenfolge gespeichert werden, so dass die Höhenwerte aufgrund der Reihenfolge jeweils einem bestimmten Eckpunkt des Rasterfeldes zugeordnet sind und durch die Indikatoren eines j-ten Levels bestimmt ist, für welche Eckpunkte des j + 1-ten Levels Höhenwerte gespeichert sind.

The method according to the invention for storing profile data for a given height profile comprises the following steps:

• (a) providing a regular grid with polygonal grids of a zeroth level in a plane, where vertices of the grids represent halftone dots, and an elevation value is stored for each grid point,
• (b) determining an error value for each grid which describes a deviation of a height profile approximated by the height values of the grid from the given height profile,
• (c) dividing the grids whose error value exceeds a predetermined threshold into polygonal sub-grids, wherein at least to the vertices of the sub-grids that do not coincide with vertices of the grids, height values are stored,
• (d) assigning one indicator to each grid, the indicator describing whether the grid is subdivided or not, and storing these indicators together with the elevation values, and
• (e) repeating steps (b) through (d) for all sub-grids, wherein the sub-grids of an i-th level form rasters of i + 1-th level, wherein the altitude values are stored in a predetermined order such that the altitude values are the order are each assigned to a specific vertex of the grid and is determined by the indicators of a j-th level, for which vertices of the j + 1-th level height values are stored.

These procedures generate semi-regular profile data. Grids that reflect a given height profile with the desired accuracy are not subdivided, whereas grids that do not reflect the given height profile with the desired accuracy are subdivided into sub-grids. These partial grid fields can in turn be subdivided (recursion). This subdivision can be repeated as often as desired. As a result, it is fundamentally possible to approximate a given height profile as exactly as desired. On the other hand, flat areas in a given height profile can be approximated very accurately with a large-area grid, without the need for further subdivisions and thus additional height values.

Thus, a very precise approximation of the given height profile is generated, as well as a division into small grids only executed where it is necessary. This allows the number of altitude values to be kept low.

The indicators describe which grid is subdivided or not subdivided. Since the indicators are stored together with the altitude values and assigned to the respective grid, the stored profile data contains the information about the division of the individual grid fields. The height values are stored for a respective grid in a predetermined order, so that the height values are each assigned to a specific vertex of the respective grid due to the order. As a result, it is not necessary to store the X and Y coordinates of the vertices of the grids. This information is contained in the profile data implied by the order of the altitude values in combination with the indicators with which the individual sections of the stored altitude values are uniquely assignable to a grid. Since the X and Y coordinates of the corner points of the grids are not explicitly stored, the data volume of the profile data is reduced considerably. A small volume of data requires little storage space and can be processed much faster than a large volume of data. This means that the profile data according to the invention in a mobile geographic information system can be easily stored and processed without an exceptionally high computing power is necessary. Compared to conventional profile data, it is possible to significantly increase the quality of the resolution with the same computing power.

The grids have a polygonal shape, in particular a triangular or rectangular shape. Preferably, the shape of the grid is uniformly triangular or rectangular. The size of the individual grids of the different levels is of course different.

The altitude values can be stored as absolute altitude values or as relative altitude values to a reference altitude. Preferably, a portion of the altitude values are stored as absolute altitude values and the remaining altitude values are stored as relative altitude values, wherein the relative altitude values relate to a particular, stored absolute altitude value. For example, an absolute height value is stored in each zeroth level grid, and all other elevation values of this zeroth level grid and the corresponding sub-grids are stored as relative altitude values with reference to this one absolute altitude value. It may also be useful to store an absolute height value for each grid of a level and to store the remaining altitude values as relative altitude values.

The absolute height values are preferably arranged at vertices of the grids, which in turn represent a regular grid, ie that an absolute height value is provided at a predetermined point of the respective grid.

In determining the error value of a grid with respect to the following elevation profile, an approximate elevation profile is represented by a plurality of triangular faces, wherein the elevation values form vertices of the triangular faces and an error in the predetermined elevation profile is calculated according to a predetermined metric.

For this purpose, an arbitrary metric for calculating the deviation of the approximated height profile from the given height profile can be used per se. As a metric, for example, a distance of the approximated height profile can be used for the given height profile, wherein the distance may be a maximum, a mean distance or a weighted distance. The mean distance is understood to mean a distance which is calculated at several points on the respective triangular surface and then averaged. A weighted distance is a distance that is calculated at several points on a triangular surface and then weighted and averaged. An alternative metric is the integral over the absolute distance between the approximated height profile and the given height profile.

The regular grid of the first level may have a predetermined number, in particular four, polygonal grids. This is particularly useful in a rectangular or square area. The corresponding sub-grids are preferably quadrants of the original grid. The further subdivision of the grids is again by partial grids in the form of quadrants. The resulting data structure corresponds to a quad-tree.

The repeated subdivision of the grids represents a recursion.

• – Lesen von erfindungsgemäßen Profildaten, wobei anhand der Indikatoren die jeweiligen Höhenwerte bestimmten Rasterfeldern zugeordnet werden und anhand der Reihenfolge, mit der die Höhenwerte in den Profildaten gespeichert sind, die Höhenwerte bestimmten Eckpunkten der Rasterfelder bzw. deren zweidimensionalen Koordinaten zugeordnet werden, und
• – Erzeugen von aneinander hängenden Dreiecksflächen, deren Eckpunkte jeweils benachbarte Höhenwerte sind, wobei die Dreiecksflächen das Höhenprofil darstellen.

A method according to the invention for producing a height profile comprises the following steps:

• Reading of profile data according to the invention, wherein the respective elevation values are assigned to specific grids based on the indicators, and the elevation values are assigned to certain vertices of the grids or their two-dimensional coordinates based on the order in which the elevation values are stored in the profile data;
• - Creating triangular faces hanging from one another, whose vertices are respectively adjacent height values, the triangular areas representing the height profile.

With the method according to the invention profile data can be stored very efficiently. NASA is aware of SRTM data with 90 m horizontal resolution, which requires uncompressed 37.7 GB of storage and lossless compressed 13.7 GB of storage space. The profile data according to the invention require only 1 GB storage space for the entire world with a horizontal resolution of 90 m.

Furthermore, the invention allows a very fast approximation of the height profile by creating the triangles (triangulation).

The inventive method is constructed incrementally. Once you have read a full level, you can already create a rough elevation profile. The more levels have been read, the more detailed the elevation profile becomes. This is advantageous, for example, for streaming from a server to a client, in particular in a client located in a motor vehicle, because after the first levels have been transmitted, it is already possible to generate a height profile which is increasingly specified with the further transmission.

The procedure is hierarchical. As a result, individual height values can be accessed very quickly.

The invention will be explained in more detail by way of example with reference to the drawing. The drawing shows in:

1 the method according to the invention is simplified schematically in a flow chart,

2 a rectangular grid with zeroth level grids,

3a the rectangular grid 2 with two partial grids of the first level,

3b . 3c the arrangement of height values in the sub-grid fields 3a , and

4 the grid 2 with partial grids of the second level.

In the embodiment explained below for storing profile data for a predetermined height profile is of a regular grid 1 for a zeroth level ( 2 ) in four grids 2 is divided. The grids 2 are rectangular fields and each represent a quadrant 1, 2, 3, 4 of the grid 1 or a maximum area to which a terrain can be imaged. The grids 2 have vertices 1-9 on which height values are to be stored.

The method of storing profile data will be described below with reference to FIG 1 explains:
The process starts with step S1 (start).

In step S2, the grids become 2 established. In the present embodiment, there are four grids 2 ( 2 ).

In the following step (S3), the height values 1-9 become the corner points 3 calculated. A given height profile is in electronically readable form of any conventional format. The grid 1 is brought into line with the range of the given height profile, which is to be implemented in profile data. At the corners 3 then the altitude values are calculated and buffered.

These height values are used to approximate the height profile (S4) by connecting the height values of three adjacent vertices with a planar triangle. For example, in the present embodiment, the triangles include vertices 1, 2, 4 and 2, 4, 5 and 2, 3, 5 and 3, 5, 6 and 4, 5, 7 and 5, 7, 8 and 5, 6, 8 and 6, 8, 9. Thus, there are eight contiguous triangles whose vertices correspond to the respective height values above the plane of the grid 1 are arranged.

For each grid 2 an error value is determined (S5), which indicates how much the approximated height profile of the corresponding grid differs from the given height profile. The metric for calculating the error value may be, for example, the maximum distance between the area of the approximated height field and the predetermined height field. Other types of metric are possible, for example the integral over the absolute value of the distances between the approximated height field and the given height field. In the present embodiment, each grid comprises 2 two triangular surfaces whose deviation from the given height profile is determined.

For each grid 2 An indicator is determined whether the deviation of the approximated height profile is greater than a predetermined maximum allowable deviation. When determining the indicator, the error value is compared with a threshold value. If the error value is greater than the threshold, then the indicator is set to 1 or to "yes" and if the error value is less than the predetermined threshold, then the indicator is set to 0 or "No".

If the deviation of the approximated height profile from the given height profile is greater than a maximum predetermined deviation, then the grid becomes 2 in partial grids 4 divided.

In the present embodiment, the grids are 2 of the first and second quadrants in partial grids 4 ( 3a ). In quadrants 3 and 4 there is no subdivision into partial grids. This means that the indicators have the following values: 1, 1, 0, 0.

In the 2 shown grids 2 represent a zeroth level. The indicators and altitude values for the zeroth level are stored in the following data field: 1, 1, 0, 0 4 bits: yes, yes, no, no dH1 Height difference to H5 dH2 Height difference to H5 dh3 Height difference to H5 dh4 Height difference to H5 H5 Height value for corner point 5 dh6 Height difference to H5 dH7 Height difference to H5 DH8 Height difference to H5 DH9 Height difference to H5

For vertex 5, the altitude value is stored as absolute value H5. For the other corner points 1-4 and 6-9, the height values are stored as height differences to the height value H5 only. By storing the height differences, the necessary number space can be kept smaller than when storing absolute height values. In addition, the provision of height differences allows a very fast triangulation of the approximated height profile.

In the following step S8 it is checked whether there is another raster of the respective level. As in the present embodiment, the zeroth level is only a single raster 1 The decision is no. This means that all rasters of the respective level have been processed.

The process goes to step S9, in which it is checked whether at least one indicator is set (S9). If this is the case then it means that at least one grid 2 in partial grids 4 is divided. The partial grid fields 4 ( 3a ) then represent grids in the next level, the first level. Therefore, in step S10, a counter i indicating the number of the level is incremented by one. The process then proceeds to step S2, in which case the grid 2 from the previous level forms the grid in the next level. Accordingly, the partial grid form 4 from the previous level, the grids of the current level. Steps S2 to S8 are executed the same way as the previous level with the new grids. Since height values already exist at certain vertices of the respective grids, only a portion of the height values need to be determined at the level above the zeroth level. 3b shows, for example, the vertices of the first quadrant according to 3a the partial grid fields 4 , where in 3b only the corner points 1, 2, 3, 4, 5 are to be assigned height values, since the other four corner points, which are marked with an "R", are already calculated and stored in the preceding level. The height values already calculated are in the level zero the height values to the vertices 1, 2, 4 and 5 ( 2 respectively. 3a ). In the second quadrant, the in 2 and 3a is arranged below the first quadrant, then only four vertices 1, 2, 3, 4 are to be assigned height values ( 3c ), since the elevation values of the other vertices have already been calculated at the previous level or at the previous grid. 4 shows the grid 2 , where black points represent all the vertices on which elevation values have been determined in the zeroth and first levels.

The indicators and elevation values of the first grid of the first level are stored in the following data field: 1, 0, 0, 1 Height difference to H3 dH1 Height difference to H3 dH2 Height difference to H3 H3 Height value for corner point 3 dh4 Height difference to H3 DH5 Height difference to H3

The indicators 1, 0, 0, 1 mean that the grids 4 of the first level in the first and fourth quadrant ( 3a ) again into further partial grids 5 to divide. These further partial grids 5 then form the grids of a second level.

The height value to the corner point H3 is stored as an absolute height value. The remaining altitude values are stored as relative altitude values, referring to the altitude value H3.

After saving the indicators and height values in step S7, it is again checked in step S8 whether there is another raster of the first level. In this case, there is another raster, the second quadrant of the raster of the 0th level, for which the altitude values are to be determined. Therefore, the process flow branches back to the step S3.

The steps S3 to S8 or S2 to S8 are repeated until the error values in all grids of the highest level are below the threshold value. Basically any number of levels can be created. In each level, elevation values are calculated for at least one or more rasters. The data fields are stored consecutively in a file, wherein for the respective grid first the indicators are stored with a numerical value and then the height values with one numerical value each. The altitude values are stored in a predetermined order. In the present embodiment, the height values in the order 1, 2, 3, 4, 5, 6, 7, 8, 9 of the in 2 stored vertices, which at higher levels already existing height values are omitted (see, eg. 3b and 3c ). In the present embodiment, the order is starting with the vertex top left, then horizontal right, then the next line in the grid and then the height values of the bottom line each from left to right.

As data fields, only the values in the left column of the data fields shown in the two tables above are stored in a file. By providing the indicators and the particular order in which the elevation values are arranged in the data fields, the individual elevation values can each be assigned to a specific vertex.

In the context of the invention, of course, other orders can be selected and of course, another form of the grid can be selected. It can also be useful triangular grids. A grid can also contain a different number of grids. In principle, it is also possible that the number of grids per grid varies in each level. However, it is advantageous to use the same shape and / or number of grids in all levels and also to maintain the same order of assignment of the height values to the vertices. As a result, both when saving and when reading the profile data, the height values can be assigned to the respective corner points very simply, quickly and unambiguously with the same rule set.

When reading the profile data, the indicators and the respective height values of a certain level are read first, then the height values are assigned to the vertices of the grid. Then the indicators and the altitude values of the next level are read. These elevation values are assigned vertices from the grid of the present level based on the indicators of the previous level and the predetermined order. This is repeated until the elevation values of all levels have been read and assigned to corresponding vertices. The elevation profile is then generated from contiguous triangular faces, where the vertices of the triangular faces are each adjacent elevation values. The contiguous triangular surfaces then form the height profile.

The profile data according to the invention are used, for example, for displaying a terrain on a geographic information system. Here it is possible to first calculate a preliminary terrain profile after reading the elevation values of a few levels. After reading altitude values of further levels, the terrain profile can be further refined.

LIST OF REFERENCE NUMBERS

1

grid

2

grid

3

vertex

4

Sub-grid of level zero or grid of level one

5

Partial Grid of Level One or Grid of Level Two

Claims (13)

A method for storing profile data for a given height profile, comprising the following steps: (a) providing a regular grid ( 1 ) with polygonal grids ( 2 ) of a first level in a plane, where vertices ( 3 ) of the grids ( 2 ) Represent grid points, and store a height value for each grid point, (b) determine an error value for each grid ( 2 ), which is a deviation of one from the height values of the grid ( 2 ) describes the approximate height profile from the given height profile, (c) dividing the grid fields ( 2 ) whose error value exceeds a predetermined threshold, into polygonal sub-grids ( 4 . 5 ), whereby at least to the vertices ( 3 ) of the partial grid fields ( 4 . 5 ) that are not connected to corner points of the grids ( 2 ) fall together, height values are stored, (d) assigning one indicator to each grid ( 2 ), where the indicator describes whether the grid ( 2 ) and storing these indicators together with the height values, and (e) repeating steps (b) to (d) for all sub-grids ( 4 . 5 ), whereby the partial grids ( 4 . 5 ) of an i-th level grids ( 2 ) of an i + 1-th level, wherein the altitude values are stored in a predetermined order, so that the altitude values due to the order of each one corner point ( 3 ) of the grid ( 2 ) and is determined by the indicators of a j-th level for which vertices ( 3 ) of the j + 1-th level altitude values are stored. Method according to claim 1, characterized in that the grids ( 2 ) and the partial grids ( 4 . 5 ) are triangular or rectangular and in particular square. A method according to claim 1 or 2, characterized in that a part of the altitude values are stored as absolute altitude values and the remaining altitude values as relative altitude values, wherein a relative altitude value defines a height difference to a respective absolute altitude value. Method according to claim 3, characterized in that only height values of the first level are stored as absolute height values and the height values of all other levels are stored as relative height values. Method according to Claim 3 or 4, characterized in that the absolute height values at vertices ( 3 ) of the grids ( 2 ), which in turn form a regular grid ( 1 ). Method according to one of claims 3 to 5, characterized in that when determining the error value, the approximated height profile is represented by a plurality of triangular surfaces , wherein the height values vertices ( 3 ) of the triangular faces, and an error is calculated according to a predetermined metric. A method according to claim 6, characterized in that the metric comprises the calculation of a distance of the approximated height profile to the predetermined height profile, wherein the distance is a maximum, a mean distance or a weighted distance. Method according to one of claims 1 to 7, characterized in that the regular grid ( 1 ) of the first level a predetermined number, in particular four, polygonal grids ( 2 ) having. Method according to one of claims 1 to 8, characterized in that the grid ( 1 ) in each level the same number of grids ( 2 ) exhibit. Method according to one of claims 1 to 9, characterized in that the grids ( 2 ) all have the same shape. Profile data produced by a method according to one of claims 1 to 10. A method for generating a height profile, comprising the steps of: - reading the profile data according to claim 11, wherein on the basis of the indicators the respective height values of particular grid fields ( 2 ) and the height values of certain vertices (based on the order in which the height values are stored in the profile data) ( 3 ) of the grids ( 2 ) or whose 2-dimensional coordinates are assigned, and - generating triangular faces hanging from one another, whose vertices ( 3 ) are respectively adjacent height values, the triangular areas representing the height profile. A method according to claim 12, characterized in that first the indicators and altitude values of some levels are read and from this a height profile is generated, and thereafter further indicators and altitude values of higher levels are read and the altitude profile is refined.

Images