JP2006284704A - Three-dimensional map simplification device and three-dimensional map simplification method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a three-dimensional map simplification technology capable of simplifying building data so as to enhance user's visibility, and also capable of reducing storage capacity for holding the three-dimensional map. <P>SOLUTION: A three-dimensional map simplification device includes: an importance calculating means 12 for calculating the importance of each element constituting a three-dimensional map image; a grouping means 17 for dividing elements arranged on the ground and whose importance is smaller than a prescribed threshold into groups so that the elements adjacent to each other belong to the same group; and an element combining means 18 for combining elements belonging to each group to make one combined element. As to less important elements, by combining a plurality of elements to make one element, the number of meshes of the element in the three-dimensional map image can be reduced. Thus, the data amount of the three-dimensional map to be displayed can be reduced. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a three-dimensional map simplification technique that simplifies the data of a three-dimensional map represented by three-dimensional data, and more particularly to a three-dimensional map simplification technique with less discomfort due to simplification.

  In recent years, research and development of a technique for displaying a three-dimensional map and performing navigation, position confirmation, and the like on an information terminal such as a portable camera or a mobile phone is being performed (see, for example, Patent Documents 1 and 2). Such portable information terminals are required to be reduced in size and power consumption. Moreover, when transmitting map information by radio | wireless communication, in order to display map information on a portable terminal in real time, reduction of the information amount of map information is required. Accordingly, there is a need for a 3D map simplification technique for displaying a 3D map with a simple configuration.

  By the way, a 3D model is used for editing, acquiring, and displaying a three-dimensional map. The 3D model has various expression methods such as a volume model, a solid model, and a spline patch. Among them, a 3D mesh model (hereinafter referred to as “3D mesh”) that approximates the surface of an object with a polygon is widely used because a high-definition model and animation can be expressed relatively easily (see Non-Patent Document 1). ).

  Usually, the 3D model includes information representing color, material, and the like called attribute data in addition to the mesh information. In the 3D map, a 3D mesh having a texture as an attribute is used. The texture is usually given as an image. Pasting an image given as a texture onto a 3D mesh is called “mapping”. By mapping the texture to the 3D mesh, a colored 3D mesh is obtained.

  Originally, a 3D mesh often has a huge amount of information. Therefore, in order to distribute this 3D mesh information over the network, manage the database, or display it at high speed, it is necessary to reduce the amount of data. Here, the reduction of the data amount is called “simplification”.

  By the way, as described above, the 3D mesh used for the three-dimensional map has both mesh information and texture information. Therefore, it is necessary to simplify each piece of information. Usually, the simplification of the texture (image) can be easily achieved by reducing the resolution (vertical and horizontal size of the image). However, a device is required for the 3D mesh.

  In the three-dimensional map, each component such as a building, a house, and a signboard is configured by a 3D mesh having a simple shape such as a rectangular parallelepiped or a cylinder. Even in a three-dimensional map of a relatively small area, the number of components is enormous. Therefore, a large amount of calculation is required for creating a database, transmission, and display. In particular, it is difficult to transmit and display data as it is in a portable terminal that has strong restrictions on the calculation capability, storage capacity, and display resolution of a CPU (central processing unit).

  Therefore, several methods for displaying a 3D map at high speed by proposing an effective data structure of the 3D map have been proposed.

  Patent Document 1 adopts a method of forming a block in which a plurality of structures such as buildings are built as one block that represents a schematic shape of the plurality of structures existing in the block. Then, based on the position information and azimuth information of the viewpoint, the texture (image) data is subjected to deformation processing and brightness / darkness processing according to the viewpoint, and the processed image data is pasted and mapped to the surface of the corresponding block.

  A display example of a three-dimensional map by this method is shown in FIG. As shown in FIG. 8, each block of the three-dimensional map is formed as one rectangular parallelepiped block 100. Image data 101 and 102 representing textures subjected to deformation processing and light / dark processing are attached to the surface of each rectangular parallelepiped block 100. The correspondence between the cuboid block 100 and the image data 101 and 102 is always maintained.

  Non-Patent Document 1 prepares simplified building shape data at a plurality of levels, and sets the level according to the distance from the viewpoint, thereby simplifying the building shape according to the distance from the viewpoint. The method of displaying is described.

FIG. 9 is a diagram illustrating an example of building data in Non-Patent Document 1. In this method, building data S _j = {S _j ¹ ,..., S _j ⁷ } of each level as shown in FIG. 9 is generated in advance and stored for all buildings j included in the three-dimensional map. The building data is configured such that the higher the level, the greater the degree of simplification.

When displaying a three-dimensional map, the distance d _i from the viewpoint to each building is calculated. Then, the level is determined according to the distance d _i and the building data of that level is displayed. At this time, the higher the distance from the viewpoint, the higher the level is set. As a result, low-level building data is selected for the building near the viewpoint, and higher-level building data is selected as the distance from the viewpoint increases. Therefore, simplified building data is displayed as the distance from the viewpoint increases. This reduces the amount of calculation for arranging polygons on the display screen.
JP 2001-5994 A JP 11-219106 A Yamane Yamane, “Creation and Display of Building Shapes for Automatic Virtual City Generation”, January 2003, Graduate School of Systems and Information Engineering, University of Tsukuba, 2003 Master thesis, [Search March 20, 2005], Internet <URL: http://www.sie.tsukuba.ac.jp/H15syuron/200205245.pdf> H. Hoppe, "Progressive meshes", ACM SIGGRAPH 1996, pp.99-108.

  If the method described in Patent Document 1 is used, mesh data can be greatly reduced. However, since each block is approximated by a rectangular parallelepiped block, the degree of approximation becomes very rough. As a result, there is a problem that visibility of the landscape for the user is lowered.

  In the method of Non-Patent Document 1, it is necessary to store building data of each level in a memory. Therefore, a large capacity storage device is required. In particular, it is difficult to hold a wide-area three-dimensional map on a portable terminal or the like having a large storage capacity restriction.

  Accordingly, an object of the present invention is to provide a 3D map simplification technique that can simplify building data so as to improve the visibility of the user and reduce the storage capacity for holding the 3D map.

  The first configuration of the three-dimensional map simplification device according to the present invention is a three-dimensional map simplification device that simplifies a three-dimensional map image. For each element constituting a three-dimensional map image, the importance of calculating the importance of the element A calculation means; a grouping means for grouping elements close to each other into the same group for elements whose importance is smaller than a predetermined threshold among elements arranged on the ground; and belonging to each group Element coupling means for coupling elements into a single coupling element.

  According to this configuration, the number of meshes of elements in the three-dimensional map image can be reduced by combining a plurality of elements into a single element for elements with low importance. Therefore, it is possible to reduce the amount of data of the 3D map to be displayed. In addition, if simplified 3D map data is transmitted to a mobile terminal or the like, or stored in an internal storage device such as a mobile terminal, wide-area 3D map data can be stored even in a mobile terminal with a small storage capacity. Can be held. Furthermore, since it is not necessary to create simplified building data at a plurality of levels, it is possible to reduce the storage capacity of the three-dimensional map and facilitate the creation of data.

  In addition, since elements with low importance are simplified and elements with high importance are displayed in detail, the visibility of the user is not deteriorated. Rather, it becomes easier to capture the features of the landscape, so that the visibility is improved.

  Here, “element” refers to an element (object, object) constituting a three-dimensional map such as a building, a road, a sign, or a signboard. Each element is represented by a 3D mesh whose surface is approximated by a polygon.

  According to a second configuration of the three-dimensional map simplification device of the present invention, in the first configuration, the element combining unit projects, for each group, each element belonging to the group onto the ground, and projects the two-dimensional Projection vertex determining means for obtaining the vertex of the projected figure; ground projection figure generating means for obtaining a convex hull of the two-dimensional projection figure and using this as a new ground projection figure S of the element; height of each element belonging to the group A height determining means for determining a representative value of the height of the element of the group from; and a connecting element that generates the connecting element by extending the ground projection figure S in the height direction by the representative value of the height Generating means.

  With this configuration, it is possible to combine the elements belonging to each group without a sense of incongruity and generate a combined element.

  Here, as the “representative value”, a maximum value, an average value, or the like is used.

  The second configuration of the 3D map simplification device according to the present invention is the first or second configuration, wherein the importance calculation unit is configured to measure a unit of the mesh with respect to one of the meshes constituting the element. The degree of importance including the surface complexity, which is a value obtained by integrating the inner product of the normal vector and the unit normal vector of another mesh with respect to all meshes constituting the element, is calculated.

  With this configuration, an element having a complicated shape such as a signboard has a large surface complexity, and an element having a relatively simple shape like an ordinary building has a small surface complexity. Therefore, by using the surface complexity as an index, it is possible to distinguish between a building and other ancillary structures to some extent, and the importance determination accuracy is improved.

A fourth configuration of the 3D map simplification device according to the present invention is a 3D map simplification device that simplifies texture images belonging to each element of the 3D map, and divides the texture original image into a plurality of closed regions. Area dividing means; representative color determining means for determining a representative color for each closed area; color replacement means for replacing the color of each closed area with the representative color; and the representative color determining means Histogram generating means for generating a color histogram H = {H (C)} of a region (C represents a color); a set of frequency peaks from the color histogram (hereinafter referred to as “peak set”) P = {P ( Peak extraction means for extracting C _j )}; n ₀ colors from the peak set P so that the distance in the color space with the color having the highest frequency in the order of the frequency is equal to or greater than a predetermined threshold. Representatives to be selected as representative colors Characterized by comprising a; selecting means.

  With this configuration, when subtracting the texture, the color that occupies the most area is determined as the representative color, and the color space distance between the representative colors is set to a certain threshold or more, so that the contrast between the representative colors is increased. You can turn on. For this reason, the amount of texture data can be reduced by color reduction by optimal color selection, and the visibility of the reduced texture can be ensured.

  A first configuration of the 3D map simplification method according to the present invention is a 3D map simplification method that simplifies a 3D map image, and calculates the importance of each element constituting the 3D map image. An importance calculation step; among elements arranged on the ground, a grouping step of grouping elements close to each other into the same group for elements whose importance is smaller than a predetermined threshold; and each of the groups An element combining step that combines elements belonging to the above to form one combined element.

  According to a second configuration of the method for simplifying a three-dimensional map according to the present invention, in the first configuration, the element combining step projects, for each group, each element belonging to the group onto the ground and projects the two-dimensional A projection vertex determining step for obtaining a vertex of the projected figure; a convex hull of the two-dimensional projected figure is obtained, and a ground projected figure generating step using this as a new ground projected figure S of the element; the height of each element belonging to the group A height determining step of determining a representative value of the height of the element of the group from the above; and a connecting element that generates the connecting element by extending the ground projection figure S in the height direction by the representative value of the height A generating step.

  According to a third configuration of the 3D map simplification method of the present invention, in the first or second configuration, in the importance calculation step, a unit of the mesh with respect to one mesh among the meshes constituting the element. The degree of importance including the surface complexity, which is a value obtained by integrating the inner product of the normal vector and the unit normal vector of another mesh with respect to all meshes constituting the element, is calculated.

A fourth configuration of the 3D map simplification method according to the present invention is a 3D map simplification method that simplifies texture images belonging to each element of the 3D map, and divides the texture original image into a plurality of closed regions. An area dividing step; a representative color determining step for determining a representative color for each closed area; and a color replacement step for replacing the color of each closed area with the representative color. In the representative color determining step, Histogram generation step for generating a closed region color histogram H = {H (C)} (C represents color); a set of frequency peaks from the color histogram (hereinafter referred to as “peak set”) P = {P (C _j)} peak extraction step for extracting, from the peak set P, n ₀ or so that the distance on a color space between and more power is greater color in degrees descending order is equal to or greater than a predetermined threshold value Characterized in that it has a; representative color selecting step of determining the elected color representative color.

  The program which concerns on this invention is a program which performs the three-dimensional map simplification method of any one structure of the said 1st thru | or 4.

  As described above, according to the present invention, by combining less important elements, it is possible to reduce the number of meshes of elements in the stereoscopic map image, simplify building data, and reduce the amount of data. Become. In addition, wide-area three-dimensional map data can be held even in a portable terminal having a small storage capacity. Furthermore, since it is not necessary to create simplified building data at a plurality of levels, it is possible to reduce the storage capacity of the three-dimensional map and facilitate the creation of data.

  Further, by simplifying the elements with low importance and displaying the elements with high importance in detail, it becomes easier to capture the features of the landscape, and the visibility of the user can be improved.

  Further, with respect to the texture data of the elements, the texture data amount is reduced by the color reduction by the optimum color selection, and the visibility of the reduced texture can be ensured.

  The best mode for carrying out the present invention will be described below with reference to the drawings.

  FIG. 1 is a block diagram showing the configuration of the 3D map simplifying apparatus according to the first embodiment of the present invention. The three-dimensional map simplification device according to the first embodiment includes, as its configuration, mesh storage means 1, texture storage means 2, mesh simplification module 3, simplified mesh storage means 4, texture simplification module 5, and simplified texture storage means. 6, texture mapping means 7, landscape data storage means 8, and output means 9 are provided.

  The mesh storage means 1 stores each element (polygon) of a detailed three-dimensional map as mesh information. The texture storage unit 2 stores image (hereinafter referred to as “texture”) information to be pasted on the surface of each element of the three-dimensional map stored in the mesh storage unit 1. It is assumed that the mesh information and texture information of these detailed three-dimensional maps are given in advance.

  The mesh simplification module 3 simplifies the mesh information (hereinafter referred to as “detailed mesh”) stored in the mesh storage means 1 and generates a simplified mesh. The simplified mesh storage means 4 temporarily stores the simplified mesh generated by the mesh simplification module 3.

  The texture simplification module 5 simplifies the texture information (hereinafter referred to as “detailed texture”) stored in the texture storage unit 2 and generates a simplified texture. The simplified texture storage means 6 temporarily stores the simplified texture generated by the texture simplification module 5.

  The texture mapping means 7 texture-maps the simplified texture to the simplified mesh, and synthesizes simplified stereoscopic map data (hereinafter referred to as “landscape data”) from a predetermined viewpoint position and viewing direction. The landscape data storage unit 8 stores the landscape data synthesized by the texture mapping unit 7. The output unit 9 outputs the landscape data stored in the landscape data storage unit 8. As the output means 9, a transmission device that transmits landscape data to the mobile terminal, a display device that displays the landscape data on the display of the mobile terminal, or the like is used.

  Here, the mesh simplification module 3 includes an element extraction means 11, an importance calculation means 12, an element classification means 13, an important element storage means 14, a fine element storage means 15, a geometric information simplification means 16, and a grouping means. 17, element coupling means 18, and simplified mesh output means 19.

  The element extraction unit 11 sequentially reads each element from the mesh storage unit 1. The importance level calculation means 12 calculates the importance level of each element read by the element extraction means 11. The importance calculation method will be described later.

  The element classification unit 13 classifies each element read by the element extraction unit 11 into an important element and a fine element based on the importance of the element. This classification method will be described later. The elements classified as important elements by the element classification means 13 are temporarily stored in the important element storage means 14. In addition, the elements classified as the fine element are stored in the fine element storage means 15. The “important element” refers to an element that can be easily noticed or an object (landmark) that can be easily seen when a three-dimensional map is viewed. “Small elements” refers to elements that are not important elements.

  The geometric information simplification means 16 deletes unnecessary fine elements or reduces unnecessary meshes for the fine elements stored in the fine element storage means 15. The grouping means 17 performs grouping so that the adjacent structures included in the fine elements are in the same group. The element combination means 18 combines the fine elements belonging to each grouped group into one combined element. The simplified mesh output means 19 outputs each important element and each fine element (or a combined element combined in a group) to the simplified mesh storage means 4.

  The element combining unit 18 includes a projection vertex determining unit 21, a projection figure generating unit 22, a height determining unit 23, and a connecting element generating unit 24.

  For each group, the projection vertex determining means 21 projects each object belonging to the group onto the ground, and calculates the vertex of the projected two-dimensional projection figure. The projection figure generation means 22 obtains the convex hull of the calculated two-dimensional projection figure and sets it as the new ground projection figure S of the element. The height determining means 23 determines the representative value h of the height of the element of the group from the height of each element belonging to the group. The joining element generation unit 24 creates a joining element formed by joining the respective small elements in the group by extending the ground projection figure S in the height direction by the representative value h of the height.

  On the other hand, the texture simplification module 5 includes important texture generation means 25. Each of these generates an important texture and a fine texture based on the detailed texture stored in the texture storage means 2.

  Here, the “important texture” refers to a texture that can be easily recognized as a landmark or an information display among textures of important elements and textures of fine elements. The “fine texture” refers to the texture of each element other than the important texture.

  The important texture generation unit 25 includes an area division unit 27, a representative color determination unit 28, and a color replacement unit 29.

  The area dividing means 27 divides the original image of the detailed texture into a plurality of closed areas. The representative color determining means 28 determines a representative color in the closed region for each closed region. The color replacement unit 29 replaces the color of each closed region with the representative color.

  Further, the representative color determining means 28 includes a histogram generating means 30, a peak extracting means 31, and a representative color selecting means 32.

The histogram generation unit 30 generates a closed region color histogram H = {H (C)} generated by the region dividing unit 27 (C represents a color). The peak extraction means 31 extracts a set of frequency peaks (hereinafter referred to as “peak set”) P = {P (C _j )} from the color histogram. The representative color selection means 32 selects n ₀ colors from the peak set P so that the distance in the color space from the peak in the descending order of the frequency to the color having the higher frequency is equal to or greater than a predetermined threshold value. decide.

  Although the 3D map simplification device of FIG. 1 can be configured by hardware, each functional block is created by a program, and each functional block is realized by executing the program by a computer. Also good.

  The operation of the 3D map simplifying apparatus according to the present embodiment configured as described above will be described below.

(1) Overall Flow of 3D Map Simplification Method FIG. 2 is a diagram illustrating the overall flow of the 3D map simplification method of the first embodiment.

  First, in step S <b> 1, the mesh simplification module 3 generates a simplified mesh based on the detailed mesh stored in the mesh storage unit 1. Thereby, the number of meshes of the detailed mesh is reduced without impairing the visibility of each element, and the data amount of the mesh is reduced. The generated simplified mesh is stored in the simplified mesh storage unit 4.

  Next, the texture simplification module 5 generates a simplified texture based on the texture (image) stored in the texture storage unit 2. Thereby, the data amount of each texture is reduced, maintaining the visibility of the texture of each element. The generated simplified texture is stored in the simplified texture storage unit 6.

  Next, in step S3, the texture mapping means 7 pastes the simplified texture on the simplified mesh of each element by texture mapping, and arranges each element at each position on the map plane, and a predetermined viewpoint. Create landscape data from The created landscape data is stored in the landscape data storage means 8.

  Finally, in step S4, the output unit 9 outputs the landscape data stored in the landscape data storage unit 8, and ends the series of processes.

  Hereinafter, the processing of the above step S1 and step S2 will be described in detail.

(2) Simplified mesh generation algorithm (step S1)
FIG. 3 is a flowchart showing the overall flow of the simplified mesh generation algorithm (step S1).

  In the simplified mesh generation algorithm, first, in steps S11 to S18, importance levels for all elements stored in the mesh storage means 1 are calculated and classified into important elements or fine elements. At this time, for the fine information, the number of meshes is reduced by simplifying the geometric information. In steps S19 and S20, the number of elements is reduced by integrating or deleting the fine elements. Hereinafter, the details will be described in order.

In step S <b> 11, first, the element extraction unit 11 reads an unselected element b _i from the mesh storage unit 1.

In step S12, the importance calculation means 12 calculates the importance I (b _i ) of the element b _i . Here, importance I _{(b i)} the volume of the elements _{b i} (or surface _area) I 1 _{(b i),} the complexity _I 2 _(b i), as a weighted linear sum of the attributes _I 3 _{(b i)} It is represented by Formula (1). Here, w ₁ , w ₂ , and w ₃ are weighting factors, and are constants that are determined as appropriate during implementation.

Volume elements _{b i} (or surface _area) I 1 _{(b i)} can be easily calculated by the vertex coordinates of the mesh constituting the component _{b i.} (The reason why the volume or the surface area is used here is that there are elements having a volume of 0 and a finite surface area as in traffic signs.)

The complexity I ₂ (b _i ) of the element b _i is calculated as follows.

(A) First, unit normal vectors of k mesh faces constituting the element b _i are obtained. Here, a set of k mesh faces constituting the element b _i is denoted as Ξ (b _i ) = {ξ ₁ (b _i ),..., Ξ _k (b _i )}. A set of unit normal vectors of the surface ξ _i (∈Ξ (b _i )) is denoted as N = {n ₁ ,..., N _k }. The unit normal vector can be easily calculated based on the vertex coordinates of each mesh.

(A) Next, a set of unit normal vectors is arbitrarily selected from N as a reference unit normal vector (hereinafter referred to as “reference normal vector”). Here, the reference normal vector is n ₁ .

(C) Next, the variance σ ² of the inner product of the reference normal vector n ₁ and the other unit normal vectors is obtained by Equation (2), and this is defined as the complexity I ₂ of the element b _i .

The attribute I ₃ (b _i ) is an importance parameter determined by the visibility of the element b _i , and is represented by Expression (3).

Here, h (b) represents the height of the element b. h (b _ar ) represents the average height of building elements whose distance from the element b _i is within a predetermined range. b _ne represents the nearest building element of element b _i . d (b _i , b _ne ) represents the distance between the element b _i and the element nearest to the element b _i . Further, a ₁ , a ₂ , and a ₃ are weighting factors, which are constant values determined as appropriate.

The attribute I ₃ (b _i ) is used when the element b _i faces the corner of the intersection, when the difference between the average height of the surrounding building elements and the height of the element b _i is large, and the element b _i Takes a large value when there are no adjacent building elements.

In this way, the complexity and attributes of the element b _i are quantified, and the importance level I (b _i ) is expressed by the linear sum of Expression (1).

In step S13, the element classification unit 13 classifies the element b _i into an important element or a fine element by determining whether or not the importance level I (b _i ) is greater than a predetermined threshold value T _{I. Here,} if _I (b i)> _{T I,} element classification means 13 determines the elements _{b i} and key elements, and temporarily stores the key elements _{b i} important element storage unit 14 (step S14), and Step S18 Migrate to If I (b _i ) ≦ T _I , the element classification unit 13 determines that the element b _i is a fine element, and proceeds to step S15.

In step S15, the element classification means 13 further classifies the fine element b _i into any one of “road”, “traffic guide sign”, “small object”, and “building”. Hereinafter referred to small classification of the Hoso瑣elements _{b i} with _{U i.}

Here, the determination of the small classification U _i of the fine element b _i by the element classification means 13 is executed as follows.

Referring to texture (A) Hoso瑣elements b _i, arrows, diagonal lines, crosswalks, when a predetermined texture specific to the road, such as a zebra zone is detected, the small classification U _i of the Hoso瑣elements b _i is Assume that it is a “road”.

Even when the texture is not detected, the altitude of the fine element b _i is equal to or less than the predetermined threshold T _lev and the variance value σ ² (n) calculated by the above equation (2) is the predetermined threshold T If it is less than or equal to _σ , the small category U _i of the fine element b _i is assumed to be “road”. That is, it is determined that a fine element having a low altitude and a small variance value σ ² (n) is close to the ground and flat, and therefore, such a fine element is a “road”.

(A) When the variance value σ ² (n) of the fine element b _i calculated by the above equation (2) is 0, the small value of the fine element b _i is small if the following three conditions are satisfied: The classification U _i is assumed to be a “traffic guide sign”.

(A) consisting Hoso瑣texture tensioned element b _i is two colors (such as blue and white).
(B) normal n of Hoso瑣elements b _i is substantially parallel to the ground surface (reference surface).
(C) The lower end of the small rod element b _i is not in contact with the ground (that is, suspended in the air).

(C) When the volume (or surface area) of the fine rod element b _i is equal to or less than the predetermined threshold value T _S , the small classification U _i of the fine stripe element b _i is assumed to be “small object”. Examples of such a small object include a signboard and a post.

(D) above (a) if none ~ in (c), small classification U _i of the Hoso瑣elements b _i is assumed to be "building".

As described above, the small classification U _i of all Hoso瑣elements b _i are determined.

In step S <b> 16, the element classification unit 13 temporarily saves the fine detail element b _i and its small classification U _i in the fine detail element storage means 15.

In step S17, the geometric information simplification means 16 simplifies the fine element b _i as follows.

(1) When the small classification U _i simplicity Hoso瑣element b _i of the road and the traffic sign of "road" or "traffic signs", the Hoso瑣elements b _i is a flat or curved volumetric 0. In this case, only the mesh vertices on the sides constituting the face are left, and all the mesh vertices in the face are deleted. In addition, when there is a straight line portion on the side constituting the surface and there are vertices of the mesh other than both ends of the straight line, all such vertices are also deleted. Then, the mesh is reconfigured with only the remaining vertices, and a simplified mesh of the fine element b _i is generated.

(2) Simplification of small objects All the fine elements whose small classification U _i is “small objects” in step S15 are deleted.

(3) Simplification of Buildings For the fine elements (hereinafter referred to as “building elements”) whose subclass U _i is “buildings” in step S15, the adjacent building elements are integrated. However, this process is performed in a later step S19.

In step S18, if the above-described series of processing (S11 to S17) for all the elements b _i in the mesh storage unit 1 has not been completed, the process returns to step S11. If completed, the process proceeds to step S19.

  In step S <b> 19, the grouping unit 17 and the element combining unit 18 perform the integration process of the building elements stored in the small element storage unit 15. This building element integration processing will be described in more detail later. Due to the building element integration process, the narrow elements are further simplified.

  In step S20, the simplified mesh output means 19 reads out the important elements stored in the important element storage means 14 and the simplified fine elements stored in the fine element storage means 15, and stores the simplified mesh storage. The data is stored in the means 4, and the simplified texture generation algorithm is terminated.

[Building element integration processing]
Next, the building element integration process in step S19 will be described in more detail.

  FIG. 4 is a flowchart showing the overall flow of the integration process of the building element integration process.

First, in step S31, the grouping unit 17 reads out building elements stored in the fine element storage unit 15 and groups building elements that are close to each other into the same group. In the following, I referred to as a group of building elements G _i, the building element belonging to the group G _i b _i, and _j. G _i = {b _{i, j} }.

In step S32, the element coupling means 18 extracts a group _{G i} of a single building element. Then, in step S33, it performs a process of combining the building elements in the group _{G i} (see FIG. 5).

  Next, in step S34, if the building element combining process has not been performed for all groups, the process returns to step S32, and the process for the next group is performed. When building element combination processing has been performed for all groups, the building element integration processing ends.

FIG. 5 is a flowchart showing the flow of building element combination processing in step S32. In this case, it referred to a group of building elements _{G i,} the group _{G i} belonging to the building element b _{i, j,} building elements b _i, the vertices of the mesh belonging to the _j p _{i, j,} and _k.

First, in step S41, the projection vertex determining means 21 extracts the vertices p _{i, j, k} on the ground for all elements b _{i, j} (∀b _{i, j} ∈G _i ) belonging to the group Gi ( (Refer FIG.5 (b)). Here, whether or not it is on the ground can be determined by whether or not the height coordinates of each vertex coincide with the height of a predetermined reference surface (the ground surface). The polygon formed by the extracted vertex set {p _{i, j, k} } is a ground projection figure obtained by projecting each building element of group G _i onto the ground.

In step S42, the projection figure generating means 22 removes vertices other than the bending points from all the extracted vertices p _{i, j, k} . In other words, if there are mesh vertices other than the end points of the polygonal straight line portion composed of all extracted vertices p _{i, j, k} , the vertex is removed because it is redundant. (See FIG. 5C).

In step S43, the projected figure generation means 22 obtains a convex hull for the polygon formed by the remaining vertices {p _{i, j, k} }. This is referred to as terrestrial projected figure S of the new coupling elements b _i (see FIG. 5 (d)).

In step S44, the height determining means 23 determines the height h (b _i ) of the new coupling element b _i from the height of each building element of G _i . In this case, the height h (b _i ) is determined as a maximum value, an average value, or the like of the height of each building element of G _i .

In step S45, the connecting element generating unit 24 generates a polygonal column obtained by extending the ground projection figure S by the height h (b _i ) toward the sky, and this is used as the connecting element b _i , and the narrow element storing unit 15 is used. Save to.

  By the processing as described above, the detailed mesh of the three-dimensional map is simplified, and the data amount can be greatly reduced. Further, since the meshes are reduced while ensuring the visibility of the less important objects according to the importance of the elements, the visibility of the three-dimensional map is not lowered. Rather, since the extra elements are reduced, the map structure becomes simple and the effect of improving visibility can be expected.

(3) Simplified texture generation algorithm (step S2)
Finally, the simplified texture generation process performed in step S2 will be described.

  First, the texture simplification module 5 refers to the classification of each element generated by the element classification means 13, and selects the important elements and “road”, “traffic guide sign”, and “building” in the detailed elements. Textures are sequentially read from the texture storage means 2. Then, each texture is simplified by executing the following texture simplification algorithm for each read texture.

  FIG. 6 is a flowchart showing the overall flow of the simplified texture generation algorithm (step S2).

First, in step S51, the region dividing unit 27 generates a feature image by applying a Sobel filter to the texture image. Next, in step S52, Hough transformation is performed on the feature image to detect a straight line portion. Then, using the obtained straight line, the original image increase (texture) is divided into a plurality of closed regions {s _i }. Hereinafter, the representative color determining unit 28 determines a representative color in each of the divided areas.

In step S54, the representative color determining unit 28 selects one closed region s _i . In step S55, the representative color {C _i ^(j) } of the closed region s _i is determined. Here, a plurality of colors are determined as the representative color. Details of the representative color determination process will be described later.

Next, in step S56, the color replacement unit 29 selects, for each pixel in the closed region s _i , a representative color having the closest color space distance to the color of the pixel, and the representative that has selected the color of the pixel. Replace with color. As a result, the texture in the closed region s _i is reduced to the number of elements of the representative color {C _i ^(j) }.

In step S57, it is determined whether or not the processing in steps S54 to S56 has been performed for all the closed regions {s _i }. If there is any processing that has not been performed, the process returns to step S54 to perform processing for the closed region. When the processes of steps S54 to S56 are finished for all closed regions {s _i }, the simplified texture generation algorithm is finished.

  Next, details of the representative color determination process in step S55 will be described.

[Representative color determination processing]
FIG. 7 is a flowchart showing the representative color determination process.

First, in step S61, the histogram generation means 30 generates a color histogram H _i = {H _i (C)} for the closed region s _i . Here, C represents a color, and H _i (C) represents the frequency of the color C in the closed region s _i .

In step S62, peak extraction unit 31 extracts the peak of the frequency from the color histogram _{H i.} Hereinafter, this set of peaks is denoted as Q _i = {q _i (C)}. q _i represents each peak point.

In step S63, the peak extraction means 31 extracts the peak q _i (C _j ) having the maximum frequency H _i (C _j ) from each peak in the peak set Q _i . Let the color C _j corresponding to this peak q _i (C _j ) be the 0th representative color C _i ⁽⁰⁾ . The already selected peak q _i (C _j ) is deleted from the peak set Q _i . That is, the set Q _i is updated as Q _i ← Q _i − {q _i (C _j )}.

In step S64, the representative color selection unit 32 normalizes the frequency H _i (C _j ) of each peak q _j (∈Q _i ) of the peak set Q _i with the maximum frequency H _i (C _i ⁽⁰⁾ ). (Formula (6)). The normalized frequency is denoted as R _i (C _j ). In step S65, the selected color number n held as an internal variable is set to zero.

In step S66, the representative color selection unit 32 extracts the color C _j having the maximum frequency R _i (C _j ) from the peaks q _j (∈Q _i ) of the peak set Q _i .

In step S67, the representative color selecting unit 32, the extracted color C _j, the shortest distance d _j in the color space from the representative color that has already been determined is calculated by Equation (7).

In step S68, the representative color selection unit 32 calculates the evaluation value x _j for the extracted color C _j by using the equation (8). From the equation (8), the evaluation value x _j is a weighted average of the shortest distance d _j and the frequency R _i (C _j ). w ₁ and w ₂ are weighting factors, and are set to appropriate constant values. That is, as the evaluation value x _j is larger shortest distance d _j of the representative color already determined is increased, also, frequency R _{i (C j)} the larger the evaluation value x _j increases.

The reason why the shortest distance d _j of the representative color of the previously determined evaluation values x _j are considered, intended that the representative value each other are selected to prevent as much as possible to become only approximate color is there. When only approximate color is selected, the closed region s _i a state such as mashed monochromatic, which may lower the visibility is lost information of the original texture almost.

In step S68, the representative color selecting unit 32 compares with the threshold _{x th} evaluation value _{x j.} If x _j > x _th , the representative color selection means 32 determines the extracted color C _j as the representative color C _i ⁽ⁿ⁾ (step S70). If x _j ≦ x _th , the extracted color C _j is discarded and the process proceeds to step S71.

In step S71, the representative color selection unit 32 deletes the evaluated peak q _i (C _j ) from the peak set Q _i . That is, the set Q _i is updated as Q _i ← Q _i − {q _i (C _j )}.

In step S72, the representative color selecting unit 32 determines whether the selected color number n has reached the specified value n _0. If not, the process returns to step S66. If it has reached, the representative color determination process is terminated.

Through the above processing, n ₀ representative colors are determined. The determined representative color is used for color reduction of the color of the closed region s _i , is used in as many pixels as possible, and colors that are separated from each other in the color space as much as possible are selected.

It is a block diagram showing the structure of the three-dimensional map simplification apparatus which concerns on Example 1 of this invention. It is a figure showing the whole flow of the three-dimensional map simplification method of Example 1. FIG. It is a flowchart showing the whole flow of the production | generation algorithm (step S1) of a simplified mesh. It is a flowchart showing the whole flow of the integration process of the integration process of a building element. It is a flowchart showing the flow of the combination process of the building element in step S32. It is a flowchart showing the whole flow of the production | generation algorithm (step S2) of the simplification texture. It is a flowchart showing a representative color determination process. It is a display example of the three-dimensional map by the method of patent document 1. FIG. It is a figure which shows the example of the building data in a nonpatent literature 1.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Mesh memory | storage means 2 Texture memory | storage means 3 Mesh simplification module 4 Simplified mesh memory | storage means 5 Texture simplification module 6 Simplified texture memory | storage means 7 Texture mapping means 8 Landscape data storage means 9 Output means 11 Element extraction means 12 Importance Calculation means 13 Element classification means 14 Important element storage means 15 Detailed element storage means 16 Geometric information simplification means 17 Grouping means 18 Element combination means 19 Simplified mesh output means 21 Projection vertex determination means 22 Projected figure generation means 23 Height Determination means 24 Joining element generation means 25 Important texture generation means 27 Region division means 28 Representative color determination means 29 Color replacement means 30 Histogram generation means 31 Peak extraction means 32 Representative color selection means

Claims (9)

A 3D map simplification device for simplifying a 3D map image,
Importance calculation means for calculating the importance of each element constituting the three-dimensional map image;
Grouping means for grouping elements close to each other into the same group with respect to elements having a degree of importance smaller than a predetermined threshold among elements arranged on the ground;
And element combination means for combining elements belonging to each group into one combined element;
3D map simplification device with The element coupling means includes
Projection vertex determining means for projecting each element belonging to the group to the ground and obtaining the vertex of the projected two-dimensional projection figure for each group;
A ground projection figure generating means for obtaining a convex hull of the two-dimensional projection figure and using this as a new ground projection figure S of the element;
Height determining means for determining a representative value of the height of the element of the group from the height of each element belonging to the group;
And a connecting element generating means for generating the connecting element by extending the ground projection figure S by a representative value of the height in the height direction;
The three-dimensional map simplification device according to claim 1, comprising: The importance calculation means includes:
A surface complexity that is the value obtained by integrating the inner product of the unit normal vector of one mesh and the unit normal vector of another mesh for all the meshes that make up the element for one of the meshes that make up the element The three-dimensional map simplification device according to claim 1, wherein an importance level including a property is calculated. A 3D map simplification device that simplifies texture images belonging to each element of a 3D map,
Area dividing means for dividing the texture original image into a plurality of closed areas;
Representative color determining means for determining a representative color for each closed region;
Color replacement means for replacing the color of each closed region with the representative color;
Have
The representative color determining means includes
Histogram generating means for generating a color histogram H = {H (C)} of the closed region (C represents a color);
Peak extraction means for extracting a set of frequency peaks (hereinafter referred to as “peak set”) P = {P (C _j )} from the color histogram;
Representative color selection means for selecting n ₀ colors from the peak set P in order of decreasing frequency and selecting a representative color so that the distance in the color space from the higher frequency is equal to or greater than a predetermined threshold value. ;
A three-dimensional map simplification device characterized by comprising: A 3D map simplification method for simplifying a 3D map image,
An importance calculation step for calculating the importance of each element constituting the three-dimensional map image;
A grouping step of grouping elements adjacent to each other into elements that are close to each other with respect to elements having an importance less than a predetermined threshold among elements arranged on the ground;
And an element combining step of combining elements belonging to each group into one combined element;
A method for simplifying a three-dimensional map. In the element combining step,
A projection vertex determining step for projecting each element belonging to the group onto the ground for each group and obtaining a vertex of the projected two-dimensional projection figure;
Obtaining a convex hull of the two-dimensional projected figure and using it as a new ground projected figure S of the element;
A height determining step of determining a representative value of the height of the element of the group from the height of each element belonging to the group;
And a connecting element generating step of generating the connecting element by extending the ground projection figure S by a representative value of the height in the height direction;
The three-dimensional map simplification method according to claim 5, wherein: In the importance calculation step,
A surface complexity that is the value obtained by integrating the inner product of the unit normal vector of one mesh and the unit normal vector of another mesh for all the meshes that make up the element for one of the meshes that make up the element 7. The method of simplifying a three-dimensional map according to claim 5 or 6, wherein an importance level including sex is calculated. A 3D map simplification method for simplifying texture images belonging to each element of a 3D map,
An area dividing step for dividing the texture original image into a plurality of closed areas;
A representative color determining step for determining a representative color for each of the closed regions;
A color replacement step of replacing the color of each closed region with the representative color;
Have
In the representative color determination step,
A histogram generation step for generating a color histogram H = {H (C)} of the closed region (C represents a color);
A peak extracting step of extracting a set of frequency peaks (hereinafter referred to as “peak set”) P = {P (C _j )} from the color histogram;
A representative color selection step of selecting n ₀ colors from the peak set P in order of increasing frequency and determining a representative color so that the distance in the color space with the color having the higher frequency is equal to or greater than a predetermined threshold. ;
A 3D map simplification method characterized by comprising: A program for executing the 3D map simplification method according to any one of claims 5 to 8.

Images