JP5237837B2 - Spatial data management device, spatial data management method, and spatial data management program - Google Patents

Description

  The present invention relates to a spatial data management device, a spatial data management method, and a spatial data management program.

  In recent years, database management systems (spatial DB management systems) capable of managing objects such as buildings existing in a two-dimensional or three-dimensional space have been developed. Such an object search is required for a map search system specialized in a navigation function, an administrative system such as city planning and disaster prevention, and a geographical information system applying facility management such as gas and water. Therefore, the spatial DB management system is used as data management middleware for a geographic information system.

  An object is composed of a graphic element expressing its position and shape, and attribute elements such as characters and numerical values expressing its contents. In the space DB management system, it is required to search at high speed an object that matches a search condition including a graphic element from a large number of objects stored in a large-capacity secondary storage device.

  The space DB management system includes a range search and a k nearest neighbor search as object searches. In the range search, a figure having an arbitrary shape is specified, and an object included in or intersecting with the figure is searched. In the geographic information system, range search is used when a map is drawn on a screen. In this case, an object included in the screen range is searched. When the user moves the drawing range or enlarges or reduces the drawing range, the range search process frequently occurs.

  On the other hand, in the k nearest neighbor search, an arbitrary point is designated, and the top k objects closest to the point are searched. The k nearest neighbor search is used, for example, in the map search system when searching for the top 10 restaurants near the destination of the tourist destination. In the map information system, nearest neighbor search is used when the closest object is designated with the mouse pointer when editing graphic data.

  The space DB management system is also being used in the development of software for embedded devices (embedded software) that uses maps. Typical examples are navigation software for in-vehicle information terminals and portable information terminals such as car navigation devices and PNDs (Personal Navigation Devices). The main functions of the navigation software are a map display on the screen, a map search for shops and facilities, and a route search from the current location to the destination. The internal processing of these functions is based on object search. When a developer uses a spatial DB management system, it is not necessary to implement object search processing for each function, and therefore, software development man-hours can be reduced.

  The spatial DB management system creates a spatial index in order to process an object search at high speed. The spatial DB management system can spatially narrow down objects to be searched by using a spatial index. As a result, the number of disk access processes between the secondary storage device and the main storage device that occupy most of the search processing time can be reduced. This disk access processing occurs when an object is read from the secondary storage device to the main storage device.

  Non-Patent Document 1 discloses a technique in which a PMR quadtree has a data structure as a representative spatial index. In the PMR quadtree, an area obtained by recursively dividing a spatial area into four is encoded in the order of Z order, and managed by a one-dimensional tree structure index using the code value of the area as a key value.

`` Speeding up construction of PMR quadtree-based spatial indexes '' VLDB Journal, Vol.11, Issue.2, pp.109-137 (2002)

  In the development on a platform where use restrictions of hardware resources such as embedded software are severe, it is necessary to consider the following two restrictions.

  The first restriction is a restriction on the size of the embedded software. It is necessary to match the embedded software program, data, and index with a predetermined capacity of the secondary storage device or the main storage device. In the development of an embedded system, there is a high need to make hardware components short, small, and thin, and therefore developers use a small-capacity secondary storage device or main storage device for the hardware components of the embedded system. As a result, it is necessary to reduce the size of the space DB in consideration of restrictions on the size of the embedded software. The space DB is composed of a management program, objects, and a space index. Among these, it is required to reduce the capacity necessary for storing a spatial index that occupies a large proportion.

The second restriction is a restriction on the processing time of the embedded software. In an embedded system, it is required to execute a certain process and finish the process within a predetermined time.
Range search is most often used in object search. Therefore, even when a spatial index after capacity reduction is used, it is necessary to be able to realize a range search as fast as the spatial index before capacity reduction.

In the method of managing spatial data using the PMR quadtree used in Non-Patent Document 1 described above, since there is a redundant part in the data structure of the PMR quadtree, the two constraints (small capacity and high speed) are sufficiently satisfied. I cannot say that.
For example, the data structure of the PMR quadtree becomes a one-dimensional tree structure index such as a B-tree, and the elements of the spatial region information and the object information are stored in leaf nodes. This spatial region is a region that includes or intersects with an object. Here, when a plurality of objects are included or intersect in one space area, elements having the same space area information are stored redundantly.

  In view of the above, the main object of the present invention is to solve the above-described problems and to carry out search processing of objects arranged in a space area at a small capacity and at a high speed.

In order to solve the above problems, the present invention is a spatial data management device for managing spatial data in which an object is arranged in a multidimensional space,
The spatial data management device includes a storage unit that stores spatial data, and an object search unit that searches the spatial data for objects that exist within a designated search range,
The storage unit stores a region index tree that stores the spatial data as a balance tree that limits the number of child nodes from each node;
Each node of the stored region index tree stores a predetermined number or less of spatial region information obtained by dividing the spatial data at least once, and the spatial region information is stored on the multidimensional space of the spatial region. Contains a key value that is the result of mapping to a scalar value that uniquely identifies the position,
A link to each child node is associated with a non-leaf node having a child node of the region index tree according to a range of key values that each child node can take, and the child node of the region index tree A leaf node that does not have a link from the spatial region information to an object where a circumscribed rectangle intersects the spatial region,
The object search unit
When the search range is specified through the input device,
In order from the root node of the region index tree, a search code value obtained by mapping a representative point of the search range to a scalar value and a key value of a spatial region of each node of the region index tree are collated, whereby a non-leaf node A space traced from the search range for each round with respect to a procedure for extracting a spatial region including a representative point of the search range and an object linked from the spatial region by tracing the link to the leaf node and the object Repeat by updating the representative points of the area excluding the area as a new search code value,
When there is no region excluding the spatial region traced from the search range, for each object extracted by the process of tracing each node of the region index tree, the circumscribed rectangle of the object is collated with the search range, and at least a part Is output as an object that intersects the search range.
Other means will be described later.

  According to the present invention, search processing for objects arranged on a space area can be carried out with a small capacity and at a high speed.

It is a block diagram which shows the spatial data management apparatus regarding one Embodiment of this invention. It is explanatory drawing which shows the area | region index tree regarding one Embodiment of this invention. FIG. 2A shows a development view of the space area, and FIG. 2B shows a tree structure for creating the development view. It is explanatory drawing which shows the area | region index tree in the local area regarding one Embodiment of this invention. FIG. 3A shows a development view of the space area, and FIG. 3B shows a tree structure for creating the development view. It is explanatory drawing which shows the outline | summary of the search process of the area | region index tree which the object search part regarding one Embodiment of this invention performs. FIG. 4A shows a development view of the space area, and FIG. 4B shows a tree structure for creating the development view. It is explanatory drawing which shows the outline | summary of the registration process of the area | region index tree which the object registration part regarding one Embodiment of this invention performs. FIG. 5A shows a development view of the space area, and FIG. 5B shows a tree structure for creating the development view. It is explanatory drawing which shows the outline | summary of the deletion process of the area | region index tree which the object deletion part regarding one Embodiment of this invention performs. FIG. 6A shows a development view of the space area, and FIG. 6B shows a tree structure for creating the development view. It is a flowchart which shows the search process of the area | region index tree which the object search part regarding one Embodiment of this invention performs. It is a flowchart which shows the search process of the space area | region in the search process of the area | region index tree regarding one Embodiment of this invention. It is a flowchart which shows the intersection determination process of the object in the search process of the area | region index tree regarding one Embodiment of this invention. It is a flowchart which shows the update process of the search code value in the search process of the area | region index tree regarding one Embodiment of this invention. It is a flowchart which shows the registration process of the area | region index tree which the object registration part regarding one Embodiment of this invention performs. It is a flowchart which shows the detail of the registration process of the area | region index tree which the object registration part regarding one Embodiment of this invention performs. FIG. 12A shows a main process of registration, and FIG. 12B shows a subroutine process called from the main process. It is a flowchart which shows the deletion process of the area | region index tree which the object deletion part regarding one Embodiment of this invention performs. FIG. 13A shows a main process for deletion, and FIG. 13B shows a subroutine process called from the main process.

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

  FIG. 1 is a configuration diagram showing a spatial data management device 1. The “space” in the “spatial data” of the present embodiment is not limited to a three-dimensional solid space, but refers to an entire n-dimensional (n is a natural number of 2 or more) vector space. Hereinafter, the present embodiment will be described by exemplifying a plane vector space of n = 2, but the present embodiment is not limited to n = 2 and can be expanded to n dimensions. In the n-dimensional space, one space is equally divided into (2 n) spaces by one space division process.

The spatial data management device 1 is configured as a computer including a central processing unit 91, a main storage device 92, a secondary storage device 93, an input device 94, and an output device 95.
The central processing unit 91 performs processing of programs stored in the main storage device 92 and executes various types of processing.
The main storage device 92 is configured as a memory that stores programs and data executed by the central processing unit 91. The main storage device 92 stores programs for configuring the object search unit 30, the object registration unit 40, the object deletion unit 50, and the data switching unit 60 as processing units.
The secondary storage device 93 is configured as a large-capacity storage device such as a hard disk drive or a flash memory disk. In the secondary storage device 93, as the spatial data handled by each processing unit of the main storage device 92, the region index tree 21, the in-region object table 22, and the object detail table 23 are stored in association with each other as a spatial database. .
The input device 94 is a device for inputting search conditions and the like necessary for searching for objects. For example, the in-vehicle information terminal includes a software keyboard using a touch panel in addition to a hardware keyboard.
The output device 95 is a device that displays an object search result, and includes a liquid crystal display or the like.

Data stored in the secondary storage device 93 will be described.
The area index tree 21 stores spatial index information as a tree structure with a low data amount and high-speed search (see FIG. 2 for details).
The in-area object table 22 stores information on whether or not an object is included for each area indicated by the area index tree 21.
The object detail table 23 stores detailed information of the object indicated by the in-region object table 22. The detailed information includes a list of coordinate values representing the actual shape of the object and attribute information of the object.

The object search unit 30 uses the area index tree 21 to perform a range search for objects included in an arbitrary range (search range).
There are the following two types of range search. One is a case where an object included in a search range is searched. The other is a case of searching for an object included in or intersecting with the search range. The object search unit 30 of the present embodiment targets the latter range search including the cross relationship. The latter range search is used for the process of displaying the map of the vehicle position and the vicinity of the destination on the in-vehicle information terminal in the example of the in-vehicle information terminal. Note that the former range search is a stricter range search condition, and can basically be realized by the same processing.
In order to execute a range search, the object search unit 30 includes a search initial setting unit 31, an intersection determination unit 32, a search start point setting unit 33, and an absolute coordinate value conversion unit.

The object registration unit 40 newly registers (adds) an object to the spatial data (region index tree 21, in-region object table 22, and object detail table 23) in the secondary storage device 93.
The object registration unit 40 includes a relative coordinate value conversion unit 41 and an area division unit 42.
The relative coordinate value conversion unit 41 converts the circumscribed rectangle of the object from an absolute coordinate value to a relative coordinate value.
The area dividing unit 42 divides the space area into four when the object nodes of the area index tree 21 overflow due to the insertion of objects.

The object deletion unit 50 deletes an already registered object from the spatial data (region index tree 21, in-region object table 22, and object detail table 23) in the secondary storage device 93. The object deletion unit 50 includes an area merging unit 51.
The area merging unit 51 combines the four spatial areas into one when the object nodes of the area index tree 21 can be merged by deleting the object.

The data switching unit 60 switches the set of spatial data (region index tree 21, in-region object table 22, and object detail table 23) that is a target of each object processing (search processing, registration processing, deletion processing). A plurality of sets of spatial data are stored in the secondary storage device 93.
For example, the data switching unit 60 stores in advance a spatial range (Japan national version, Kanto version, Tokyo version, etc.) handled by each set of spatial data, and a position indicated by a search condition specified via the input device 94 ( For example, a set of spatial data (for example, the Tokyo version) corresponding to the most detailed map scale is selected from a set of spatial data including Tokyo Station.

Thus, by preparing a plurality of sets of spatial data, the range search can be speeded up, although the capacity of the secondary storage device is consumed. Note that, for example, a region where the driver frequently drives is set in the spatial data set (detailed map).
On the other hand, only one set of spatial data (for example, the Japanese national version) may be prepared. At that time, the data switching unit 60 can be omitted from the spatial data management device 1.

FIG. 2A is an expanded view in which the information indicated by the area index tree 21 is expanded for easy understanding. In this development view, a horizontal (x) 16 section × vertical (y) 16 section space area is shown as viewed from above (z-axis), and seven objects (A to G) are arranged. .
The numbers (for example, the origin “0”, the center point “192”, and the like) in the development view indicate Morton code values encoded in the order of the Z order corresponding to the described position of the number. The Morton code value is an example of a mapping value that maps a position on an n-dimensional space region to a one-dimensional scalar value. In the present embodiment, Morton code values are exemplified, but other mapping calculation values such as Hilbert code values encoded with a Hilbert curve may be used instead.
A gray rectangle in the development view indicates a rectangle that circumscribes each object, and object IDs (A to F) are described therein.
A dotted line in the development view indicates a boundary of each spatial region generated by dividing the spatial region into four equal areas by a perpendicular bisector of each axis. For example, the upper right area starting from the Morton code value “192” (lower left point) is a larger (8 × 8 section) area as a result of dividing the 16 spatial areas into one. On the other hand, the area near the center starting from the Morton code value “60” (lower left point) is a smaller (2 × 2 section) area as a result of dividing the 16-space area into three.

  The spatial data management device 1 represents the position of each spatial region as a set of Morton code values obtained by converting the representative point (lower left point) and the number of divisions (number of divisions). Since the spatial region is recursively divided into four areas with equal areas, the position and shape of the spatial region can be uniquely specified by a set of Morton code values and the number of divisions. The representative point of the spatial area is a coordinate value that is a set of the minimum value of the x coordinate and the y coordinate in the area.

Hereinafter, a calculation method for converting the coordinate value (7, 10) to the Morton code value “157” will be described.
(Procedure 1) (x, y) = (7, 10) is input as a coordinate value.
(Procedure 2) The input decimal coordinate values (7, 10) are converted into binary coordinate values (0111, 1010).
(Procedure 3) The binary coordinate values (0111, 1010) are alternately arranged in the order of the y-coordinate and the x-coordinate in order from the first bit of the binary number to be “10011101”.
The leading “1” of “10011101” is the leading “1” of the y-coordinate “1010”.
The second digit “0” of “10011101” is the leading “0” of the x-coordinate “0111”.
The third digit “0” of “10011101” is the second digit “0” of the y coordinate “1010”.
(Procedure 4) When “10011101” is converted into a decimal number, the Morton code value “157” is obtained.

  Due to the nature of the Morton code value, the Morton code value corresponding to the lower left coordinate of an arbitrary rectangle in the space area is always smaller than the Morton code value corresponding to the upper right coordinate of the rectangle. The Morton code value corresponding to the coordinates inside the rectangle is included between the Morton code values of the lower left coordinates and the upper right coordinates of the rectangle. If these properties of Morton code values are used for multidimensional range search, the multidimensional range as a search condition can be simplified to a one-dimensional scalar value range. Furthermore, since Morton code values can be assigned close code values between spatially close coordinate values, the range of one-dimensional scalar values can be narrowed to speed up the range search.

In the above example, the Morton code value calculation method on the two-dimensional plane (x, y) has been shown. However, the Morton code value is not limited to two dimensions, and can be applied to three or more n dimensions. Can be extended. The calculation of the n-dimensional Morton code value can be realized by replacing (procedure 3) with the following (procedure 3b).
(Procedure 3b) n-dimensional coordinate values expressed in binary numbers are alternately arranged in the following order.
(First bit of n-dimensional coordinate value) → (First bit of n−1-dimensional coordinate value) → (Omitted) → (First bit of one-dimensional coordinate value) → (Second bit of n-dimensional coordinate value) → ( (2nd digit bit of n-1 dimensional coordinate value) → (abbreviated) → (2nd digit bit of 1D coordinate value) → (abbreviated) → (last bit of 1D coordinate value)

FIG. 2B is a configuration diagram showing the information of the area index tree 21 shown in FIG. 2A as a tree structure. In the area index tree 21, each node and a link connecting the nodes have a tree structure. The area index tree 21 has a tree structure of a balance tree (B-tree, B + tree, etc.), and includes non-leaf nodes, leaf nodes, and object nodes. These nodes are stored in the secondary storage device in a one-to-one relationship in units of pages. A page is a minimum input / output unit between the secondary storage device and the main storage device in the database management system.
In the following description, an example in which a B + tree is used as an example of a balance tree will be described. The B + tree is an index structure that is most often used in database management systems in that search and update processing can be performed at high speed. In the B + tree, a non-leaf node includes a key value and a pointer to a child node, and a leaf node includes a key value and a record pointer. Then, adjacent leaf nodes are connected by a one-way link.

The leaf nodes and non-leaf nodes of the area index tree 21 are Morton code values (indicated by numerical values in the figure) that are key values and the number of divisions for creating a spatial area from the lower left coordinates indicated by the Morton code values. A maximum of three sets (indicated by the numerical value in parentheses in the figure) and (indicated by the gray part of each node in the figure) can be stored. For example, the upper left space area in FIG. 2A is expressed as a set of “Morton code value = 128, number of divisions = 1”.
Since the set of Morton code values and the number of divisions represents one space area in FIG. 2A, in the following description, the set of Morton code values and the number of divisions is referred to as “space area information in a node”. Is written.
In this way, a spatial region generated by dividing the spatial region into four equal areas is uniquely represented by a set of Morton code values and the number of divisions. The Morton code value is a one-dimensional scalar value, so the magnitude relationship can be defined without depending on the order of the spatial domain (two-dimensional plane, three-dimensional space), etc., and the key value of the B + tree node (one-dimensional key value) Function as.

The non-leaf (stem) node of the region index tree 21 stores a maximum of three spatial region information. Non-leaf nodes link non-leaf nodes or leaf nodes as child nodes. Spatial region information in non-leaf nodes is arranged in descending or ascending order, and pointers to child nodes are indicated by arrows in white portions between the spatial region information. The pointer to the child node indicates a physical address on the disk storing the page corresponding to the child node.
Note that the maximum number of pieces of spatial region information is three only as an example. The maximum number of space area information is determined by the page size, the size of the space area information, the size of the pointer to the child node, and the like. If the maximum number of space area information is “i” in the data structure of the B + tree, the number of pointers to child nodes is “i + 1”.
A pointer set between the first space area information and the second space area information of the parent node indicates a child node indicated by the pointer. At this time, the Morton code value of each spatial region information of the child node falls within the range from the Morton code value of the first spatial region information to the Morton code value of the second spatial region information.

A leaf node of the region index tree 21 stores a maximum of three spatial region information. Leaf nodes link to object nodes. The space area information of leaf nodes is arranged in descending order or ascending order. From each space area information (for example, Morton code value “32”) of the leaf node, there is a link (for example, object “A”) to an object node indicating an object overlapping the space area indicated by the space area information.
Note that the maximum number of pieces of space area information is three as an example. If the maximum number is “i”, pointers to “i” object nodes are stored.

Furthermore, the plurality of spatial region information stored in the same leaf node constitute the leaf node so that the Morton code values of the spatial region information do not overlap each other.
Thereby, the disk capacity required for the area index tree 21 can be reduced as compared with the following comparative example. The comparative example is an example in which a PMR quadtree is configured using a set of key value and object information as an element of a leaf node of the region index. In this comparative example, since one space region contains a plurality of objects, a plurality of elements having the same key value are stored in the leaf nodes, resulting in redundancy.

  The object node of the region index tree 21 includes one piece of information on the spatial region indicated by the leaf node to be linked, and information on a plurality of objects overlapping (partially touching or containing all) (in this case, up to three pieces). Is stored. The object information includes a spatial area (Morton code value and the number of divisions) and an object (ID and position in the area) in an in-area object table 22 described later.

Table 1 shows the in-region object table 22. The in-region object table 22 manages spatial region information (a set of Morton code values and the number of divisions) and object information (a set of an object ID and an in-region position of the object) in association with each other. Information on one spatial region and information on a plurality of objects corresponding to the spatial region are included in the object node of the region index tree. Since one page is allocated to one object node, information on objects overlapping the same space area is stored in one page. As a result, information on objects overlapping the same area can be acquired by one disk access, which contributes to speeding up the range search.
The spatial area information in the intra-area object table 22 is the same as the spatial area information in the non-leaf nodes and leaf nodes of the area index tree 21.

  The object ID of the area object table 22 is the same as the object ID of the object node of the area index tree 21. The object ID is information for uniquely identifying an object, for example, a record pointer of the object. The record pointer of the object indicates a physical address on the disk where the record in the table is stored.

The in-region position of the object in the in-region object table 22 indicates the relative coordinate value of the circumscribed rectangle (lower left coordinate to upper right coordinate) of the object indicated by the object ID.
For example, the circumscribed position of the object (object ID = E) included in the space area (Morton code value = 16, number of divisions = 2) is “(0, 2) to (2, 2)”.
Then, referring to FIG. 2A, the object “E” is included in the space area of the lower left coordinates (4, 0) to the upper right coordinates (7, 3).
The lower left coordinate of the object “E” is calculated as (4, 2) by adding the lower left coordinate (0, 2) of the in-region object table 22 to the lower left coordinate (4, 0) of the contained spatial region. it can.
The upper right coordinate of the object “E” is calculated as (6, 2) by adding the upper right coordinate (2, 2) of the in-region object table 22 to the lower left coordinate (4, 0) of the contained spatial region. it can.

Here, the disk capacity can be reduced by determining in advance the minimum number of bits necessary for expressing each parameter.
For example, the minimum number of bits required for expressing the Morton code value is 16 (2 to the 4th power), which is 8 bits because the range of the spatial region is 16 × 16 in the example of FIG. That is, the capacity of the variable for storing the Morton code value can be specified based on the size of the space area indicating the map range.
On the other hand, the minimum number of bits necessary for expressing the coordinates of the position in the area is, for example, an 8 × 8 area in the case of a spatial area (Morton code value = 128, number of divisions = 1). Since each y coordinate is 3 bits, the total is 6 bits. By storing the position of the object in the area in the form of a difference (relative coordinates) from the representative point (lower left coordinate) of the contained spatial area, the position of the area in the area can be stored compared to the data capacity required by the PMR quadtree. Can reduce the amount of disk space required. The effect of reducing the disk capacity is particularly effective when applied to an embedded device having a very small secondary storage device depending on the hardware components and applications to be applied.
In this way, the number of bits of the variable can be defined according to the size of the area to be handled. Instead of the number of bits, the number of bytes may be used.

Table 2 shows the in-region object table 22. Compared with Table 1, a “type ID” column is added to the object information.
The type ID indicates the type of object. For example, the type ID includes bank “0”, gas station “1”, restaurant “2”, school “3”, and the like. These type IDs are specified in search conditions from the user input via the input device 94. For example, the user inputs a search condition for selecting only “gas station” from facilities within a radius of 500 m from the vehicle.
The minimum number of bits necessary for expressing the type ID is set based on the range of values that the type ID can take. For example, when there are 100 type IDs, the type ID is stored in a variable that can store 2 8 = 128 values, so 8 bits is the minimum number of bits.

Table 3 shows the object detail table 23. The object detail table 23 stores object ID, type ID, and actual shape in association with each other as detailed information about each object.
The type ID in the object detail table 23 has the same meaning as the type ID in the in-region object table 22.
The actual shape of the object detail table 23 indicates the actual shape of the object. For example, the shape of the object can be represented by a sequence of coordinates. The actual shape of the object is used for highly accurate search in the object search process. For example, even if the bounding rectangle of the object is within the search range, if the object is a circle, the circle of the object does not fit within the search range if the corners of the bounding rectangle just overlap the search range. become.

One or more sets of the spatial data (region index tree 21, region object table 22, object detail table 23) described above are stored in the secondary storage device 93. The data switching unit 60 switches a set of spatial data that is a target of each object processing (search processing, registration processing, and deletion processing).
FIG. 3 shows a region index tree 21 and a developed view thereof in which a rectangular region from the lower left coordinate value (3, 3) to the upper right coordinate value (10, 10) in FIG. 2 is a local region.
Table 4 shows an example of the in-region object table 22 set with the region index tree 21 of FIG. Thus, by localizing a set of spatial data, the range handled by each data structure is narrowed, and the amount of data to be searched is reduced. As a result, the number of nodes (pages) read from the secondary storage device to the main storage device by the range search is reduced, and the search processing time can be shortened.

  FIG. 4 shows the region index tree 21 of FIG. 2 when the search range executed by the object search unit 30 is a rectangular lower left coordinate (5, 2) to upper right coordinate (7, 5) surrounded by a thick dotted line. Shows how to search. The procedure in this case will be described below with reference to FIGS. 4 (a) and 4 (b).

First, the first search process (leftmost dotted arrow) in FIG. 4B will be described. The search initial setting unit 31 converts the lower left coordinates (5, 2) of the search range into a Morton code value “28” and sets it to the search code value M. When searching the region index tree, a spatial region including coordinates corresponding to the search code value M is traced. Due to the nature of the Morton code value, a subtree including the maximum Morton code value below the search code value M is traced. In this case, referring to the root node of the region index tree 21, the link to the left child node of the Morton code value “60” is traced. This is because this link points to a subtree having the maximum Morton code value below the search code value M “28”.
Similarly, in the child node of the root node, the grandchild node indicated by the left pointer of the Morton code value “32” is traced. Further, in this grandchild node, the search code value M “28” is included in the spatial region whose representative point is the Morton code value “16” of the spatial region information (the Morton code value of the upper right coordinate of the spatial region is 31). Therefore, the object node “E” which is the link destination from the node is put in the rough determination list.

The search start point setting unit 33 updates the search code value M after tracing the region index tree 21 to the leaf node (and the object node below it). The next search code value M is the coordinates of the remaining search range after the space area searched from the current search code value M is excluded (cut out as an area) from the search range corresponding to the current search code value M. The minimum value among Morton code values corresponding to.
For example, if the range (4, 0) to (7, 3) of the searched spatial region is excluded from the current search range (5, 2) to (7, 5), the remaining search range is (5 , 4) to (7, 5). Therefore, the search start point setting unit 33 sets the Morton code value “49” at the coordinates (5, 4) as the next search target M.

Next, the second search process (center dotted arrow) in FIG. 4B will be described. Referring to the root node of region index tree 21, the link to the left child node of Morton code value “60” indicating the subtree having the maximum Morton code value below search code value M “49” is traced. In the child node, the link to the second child node from the left which is smaller than “52” and has Morton code value “32” or more of the spatial region information is traced. In the child node (the grandchild node of the root node), the search code value M “49” is included in the spatial region whose representative point is the Morton code value “48” of the spatial region information, so the object that is the link destination from that node The node “A” is put in the rough judgment list.
If the searched spatial region ranges (4, 4) to (5, 5) are excluded from the current search ranges (5, 4) to (7, 5), the remaining search ranges are (6, 4). ) To (7, 5). Therefore, the search starting point setting unit 33 sets the Morton code value “52” at the coordinates (6, 4) as the next search target M.

The third search process (rightmost dotted arrow) in FIG. 4B will be described. Referring to the root node of the region index tree 21, the search code value M “52” is smaller than the Morton code value “60” of the spatial region information of the root node, and therefore the link to the child node at the lower left is followed. In the child node, the search code value M “52” is equal to or greater than the Morton code value “52” of the spatial region information, and the link to the third child node from the left is further traced. In the child node (the grandchild node of the root node), the search code value M “52” is included in the spatial region whose representative point is the Morton code value “52” of the spatial region information, so the object that is the link destination from that node The node “F, G” is put in the rough judgment list.
If the searched spatial region ranges (6, 4) to (7, 5) are excluded from the current search ranges (6, 4) to (7, 5), the remaining search ranges disappear. Therefore, the search code value M is not updated, and the search for the region index tree 21 is terminated.

  As a result, the object “E, A, F, G” is searched by following the link from the area index tree 21. Subsequent detailed determination (refer to the processing of the intersection determination unit 32 to be described later for details) excludes the object “A” as out of the search range, and the object “E, F, G” in FIG. The search result is output from the output device 95 as a search result that matches the search range (5, 2) to (7, 5) surrounded by the thick dashed frame.

FIG. 5 is an explanatory diagram showing that the objects H and I are registered from the state of the area index tree 21 of FIG.
In FIG. 5A, objects H (6, 11) to (8, 12) and objects I (2, 15) to (6, 15) are sequentially arranged in the space regions (0, 8) to (7, 15) When inserted in the vicinity, four objects C, D, H, and I are included in the space area, which exceeds the allowable amount of three.
Therefore, the region dividing unit 42 divides one spatial region (0, 8) to (7, 15) whose allowable amount has been exceeded into four, and obtains four spatial regions (Morton code values = 128, 144, 160, 176). ). As a result, the number of objects included in each of the four spatial regions can be within the allowable amount of three.
The area index tree 21 shown in FIG. 5 (b) has one space area (Morton code value = 128, number of divisions = 1) in FIG. Are divided into four spatial regions (Morton code values = 128, 144, 160, 176). As a result of this division, the leaf node including the spatial region information of Morton code value = 128 in the region index tree 21 also exceeds the allowable three spatial region information, so the leaf node of the region index tree 21 is also divided into two. Has been.
Further, as the leaf node of the area index tree 21 is updated, the object node that is the link destination of the leaf node is also updated.

FIG. 6 is an explanatory diagram showing that the object A is deleted from the state of the area index tree 21 of FIG.
In FIG. 6A, by deleting the objects A (3, 4) to (4, 7), a set of four spatial regions in the vicinity where the object A was located (Morton code value = 48, 52). , 56, 60) each has a margin for the allowable amount (three) of the object.
Therefore, the region merging unit 51 merges the four space regions into one space region (Morton code value = 48, number of divisions = 2). Thereby, the number of nodes of the area index tree 21 can be reduced, and the capacity of the secondary storage device necessary for storing the area index tree can be suppressed.
The area index tree 21 shown in FIG. 6B has four spatial areas (Morton code values = 48, 52, 56, 60) in FIG. 2 as one spatial area in accordance with the area merging process by the area merging unit 51. (Morton code value = 48). Along with this merge, the object node that becomes the link destination of the merged leaf node in the area index tree 21 is also updated to the object node indicating the objects D, G, and F in the merged space area.

  FIG. 7 is a flowchart showing a search process for objects existing in the search range designated by the range shown in FIG. This flowchart is executed by the object search unit 30.

In S <b> 101, the search initial setting unit 31 receives an input of a search range serving as a search condition via the input device 94. The shape of the search range is arbitrary such as a circle, a rectangle, or a polygon. For example, when “search for a gas station existing within a radius of 500 m around the current value” is selected, the following processing is performed.
・ A rectangle circumscribing a circle with a radius of 500m centered on the current value is obtained, the lower left coordinate value of the rectangle is converted to Morton code value “M_min”, and the upper right coordinate value of the rectangle is converted to Morton code value “M_max” To do.
The Morton code value “M_min” is set as the initial value of the search code value M.

In S102, the object search unit 30 searches the area index tree 21 for a spatial area at least partially intersecting the rectangular area starting from the search code value M (details will be described later with reference to FIG. 8). .
In S103, the intersection determination unit 32 compares the shape of the object in the space area searched in S102 (obtained from the object table 22 in the area and the object detail table 23) with the shape of the search range specified in S101. Thus, an object that intersects at least partially within the search range is searched (details will be described later with reference to FIG. 9).

In S104, the search starting point setting unit 33 sets the search code value M that is initialized in S101 and updated in S104 to the search code value M indicating the search range after excluding the spatial region searched in S102. (The details will be described later with reference to FIG. 10).
In S105, the object search unit 30 determines whether or not the search code value M updated in S104 is larger than the Morton code value “M_max” input in S101. If the determination condition is satisfied, the process ends. If the determination condition is not satisfied, the process returns to S102.

  FIG. 8 is a flowchart showing details of the process of S102 executed by the object search unit 30. In this flowchart, a spatial area including coordinates corresponding to the search code value M is searched using the area index tree. At this time, a spatial region having a maximum Morton code value equal to or less than the search code value M is searched.

In S201, the root node of the area index tree 21 is set as the search target node N.
In S202, it is determined whether the search target node N is a non-leaf node. When the determination condition is satisfied, the process proceeds to S203, and when the determination condition is not satisfied, the process proceeds to S211.

In S203, one pointer of a non-leaf node is acquired. At this time, the pointers of the non-leaf nodes are acquired in order from the top.
In S204, the range of the Morton code value in the space area corresponding to the node pointed to by the pointer is calculated. The range of the Morton code value can be calculated from the key value (Morton code value, number of divisions) included in the non-leaf node.
In S205, it is determined whether or not the search code value M is included in the range of Morton code values calculated in S204. S205 is a process of searching for the maximum Morton code value that is equal to or less than the search code value. When the determination condition is satisfied, the process proceeds to S206, and when the determination condition is not satisfied, the process proceeds to S203.
In S206, the node pointed to by the pointer is set as the search target node N, and the process returns to S202.

In S211, one leaf node element is acquired. At this time, the leaf node elements are acquired in order from the top.
In S212, the range of the Morton code value of the spatial region corresponding to the element acquired in S211 is calculated. The range of Morton code values in the spatial domain can be calculated from the key value (Morton code value, number of divisions) of the leaf node.
In S213, it is determined whether or not the search code value M is included in the range of Morton code values calculated in S212. When the determination condition is satisfied, the process proceeds to S214, and when the determination condition is not satisfied, the process proceeds to S211.
In S214, the element of the leaf node is held as the searched space area, and the process returns to S103.

  FIG. 9 is a flowchart showing details of the process of S103, which is mainly executed by the intersection determination unit 32 (S304 is the absolute coordinate value conversion unit 34).

In S301, the object node pointed to by the element of the spatial area held in S102 is acquired.
In S302, it is determined whether all the elements of the object node have been acquired. When the determination condition is satisfied, the process returns to S104, and when the determination condition is not satisfied, the process proceeds to S303.
In S303, one object node element is acquired. At this time, the elements of the object node are acquired in order from the top. Hereinafter, the element of the acquired object node is referred to as “selected object”.

In S303b, with reference to the “type ID” of the in-region object table 22, it is determined whether or not the type of the selected object matches the type of the search condition. When the determination condition is satisfied, the process proceeds to S304, and when the determination condition is not satisfied, the process proceeds to S302.
The determination process of S303b is executed when the type ID is included in the intra-region object table 22 (see Table 2), but when the type ID is not included in the intra-region object table 22 (Table). 1) can be omitted.

In S304, the absolute coordinate value conversion unit 34 converts the absolute coordinate value of the circumscribed rectangle of the selected object according to the following procedure.
The Morton code value in the in-region object table 22 is converted into the lower left coordinate value (a, b) of the spatial region obtained in S102.
-The lower left coordinate value (a, b) of the spatial area is added to the lower left coordinate value (x'_min, y'_min) of the position in the area of the selected object, and the absolute coordinate value (x_min, y_min) is obtained (for example, x_min = x′_min + a, y_min = y′_min + b).
-The upper left coordinate value (a, b) of the spatial area is added to the upper right coordinate value (x'_max, y'_max) of the position in the area of the selected object, and the absolute coordinate value (x_max, y_max) is obtained (for example, x_max = x′_max + a, y_max = y′_max + b).

In S305, it is determined whether or not the circumscribed rectangles (x_min, y_min) to (x_max, y_max) of the selected object obtained in S304 intersect the spatial region obtained in S102. When the determination condition is satisfied, the process proceeds to S306, and when the determination condition is not satisfied, the process proceeds to S302.
Note that the intersection determination process in S305 is a process for determining whether or not there is an overlapping portion between two rectangles (the first rectangle and the second rectangle), for example, between the first rectangle and the second rectangle. This is realized by preparing in advance a judgment formula indicating the positional relationship of and substituting it into the judgment formula. As an example of the determination formula, if “(the x coordinate of the right end of the first rectangle) <(the x coordinate of the left end of the second rectangle)”, the determination formula that the first rectangle and the second rectangle do not overlap is exemplified. .

In S306, the actual shape of the selected object obtained in S304 is read from the object detail table 23, and it is determined whether or not it intersects with the actual shape of the space area input in S101. When the determination condition is satisfied, the process proceeds to S307, and when the determination condition is not satisfied, the process proceeds to S302.
Note that the intersection determination process of S306 is a detailed intersection determination performed based on the “real shape” data in the object detail table 23. In this intersection determination process, for example, scanning is performed so as to cover the absolute position of the actual shape of the selected object in order, and one or more scanning points of the selected object existing in the spatial region among the scanning points of the selected object. There is a method of determining that it intersects when it is discovered.
Here, since the intersection determination process of S306 is a determination process for a large data amount as an actual shape, the amount of calculation becomes large. Therefore, the user may optionally select whether or not to execute the intersection determination process of S306. If the user does not select, the intersection determination process of S306 may be omitted, and if S305 is satisfied, the process may proceed to S307.

  In S307, the selected object determined to intersect with the space area is output as a search result via the output device 95, and the process returns to S302.

  FIG. 10 is a flowchart showing details of the search code value M update processing of S104 executed by the search start point setting unit 33. In this flowchart, among the Morton code values included in the rectangle of the search range, the smallest Morton code value in the search range excluding the space area searched in S102 is obtained.

In S401, the search code value M is converted into a coordinate value (a, b).
In S402, the upper right coordinates (x_max, y_max) of the space area searched in S102 are set.
In S403, it is determined whether or not a line parallel to the y-axis, which is larger by 1 in the x-axis direction than the right side of the space area searched in S102, intersects the search range rectangle. When the determination condition is satisfied, the process proceeds to S404, and when the determination condition is not satisfied, the process proceeds to S405.
In S404, the Morton code value of the proximity point (x_max + 1, b) of the spatial region searched in S102 is inserted into the search code value list. The search code value list is implemented, for example, as a priority queue with Morton code values as priorities.

In S405, it is determined whether or not a line parallel to the x-axis that is larger by 1 in the y-axis direction than the upper side of the spatial area searched in S102 intersects the search range rectangle. When the determination condition is satisfied, the process proceeds to S406, and when the determination condition is not satisfied, the process proceeds to S407.
In S406, the Morton code value of the proximity point (a, y_max + 1) of the spatial region searched in S102 is inserted into the search code value list.
In S407, the minimum Morton code value of the Morton code values in the search code value list inserted in S404 and S406 is set as the search code value M.

  FIG. 11 is a flowchart showing processing for registering an object in the spatial data, which is executed by the object registration unit 40.

In step S <b> 501, the object registration unit 40 acquires detailed information on an object to be registered.
The detailed information is, for example, information (position of object on map, type ID, actual shape) registered in object detail table 23. Furthermore, arbitrary information (address, telephone number, etc.) provided as facility information for car navigation may be added as detailed information.

  In S502, the object registration unit 40 registers the detailed information of the object input in S501 in the spatial data (object detail table 23). If the object to be registered has already been registered in the object detail table 23, S502 may be omitted to avoid double registration, or if the detailed information input in S501 has changed, it has been registered. The record of the object detail table 23 may be rewritten with the changed value.

In S503, the object registration unit 40 creates registration data in other related spatial data (region index tree 21, in-region object table 22) based on the registration data in the object detail table 23 in S502. .
For example, the object registration unit 40 divides an object newly registered in the object detail table 23 along each spatial region registered in the in-region object table 22, and the absolute position of the divided object on the map Is converted into a relative position from the representative point of each spatial region.
In the process of dividing along each spatial area, for example, the object A exists for one record in the object detail table 23, but is divided into three spatial areas in the in-area object table 22.

  In step S504, the object registration unit 40 registers the registration data to the created spatial data (region index tree 21, in-region object table 22) into each spatial data (details are shown in FIG. 12).

  FIG. 12A is a flowchart showing details of the registration process (S504) to the spatial data (region index tree 21, in-region object table 22) executed by the object registration unit 40.

In S601, the object registration unit 40 calls the process of the object search unit 30 using the circumscribed rectangle of the registration target object created in S503 (refer to FIG. 7) as a search target circumscribed rectangular object. The intersecting space area (intersection area) is obtained and inserted into the intersection area list.
In step S602, the object registration unit 40 determines whether all the elements of the intersection area list have been acquired. When the determination condition is satisfied, the present process is terminated, and when the determination condition is not satisfied, the process proceeds to S603.

In step S603, the object registration unit 40 extracts one space area in order from the top of the intersection area list. Hereinafter, the extracted space area is referred to as a selection area.
In step S <b> 604, the object registration unit 40 acquires a registered object node linked from the leaf node corresponding to the selected region of the region index tree 21.
In S605, it is determined whether or not the sum of the number of registered objects and the number of objects to be registered this time exceeds the capacity of objects in each space area (for example, a maximum of three in the case of FIG. 2). judge. When the determination condition is satisfied, the process proceeds to S606, and when the determination condition is not satisfied, the process proceeds to S607.

In S606, the region dividing unit 42 divides one selected region into four spatial regions. Specifically, first, Morton code values of the representative points are calculated for four spatial regions when the selection region to be divided is divided into four equal areas. Then, the spatial area information of the selected area registered in the leaf node of the area index tree 21 is deleted, and the spatial area information of the four spatial areas calculated in S611 is added to the leaf node.
Here, when the allowable amount of leaf nodes of the area index tree 21 (up to three space area information in the case of FIG. 2) is exceeded, as shown in FIG. By complying with the above, the allowable amount of leaf nodes is observed. As a result of the processing in S606, the number of space areas has increased, so that there is room in the capacity of objects in each space area, and the determination condition in S605 can be easily satisfied. Then, the process returns to S503.

In S607, the relative coordinate value conversion unit 41 converts the position of the circumscribed rectangle of the object indicated by the registration target object node from the absolute coordinate value to the relative coordinate value (see FIG. 12B).
In step S608, the object registration unit 40 registers the object to be registered in the selection area. Specifically, the object to be registered is added to the object node linked from the leaf node of the selected area of the area index tree 21 and the object to be registered is added to the record indicating the selected area of the in-area object table 22 To do.

  FIG. 12B is a flowchart showing details of the coordinate conversion processing (S607) executed by the relative coordinate value conversion unit 41.

In S611, the lower left coordinate value and the lower right coordinate value (x_max, y_max), which are absolute coordinate values of the circumscribed rectangle of the object, are set.
In S612, the Morton code value and the number of divisions of the representative point of the space area including the object of S611 are read from the in-area object table 22, and the lower left coordinate value (x'_min, y'_min) and the upper right coordinate value of the space area are read. (X'_max, y'_max) is calculated.

  In S613, by subtracting the lower left coordinate value (x'_min, y'_min) in S612 from the lower left coordinate value (x_min, y_min) in S611, the relative coordinate value (x '' _ min, y of the lower left coordinate value of the object). '' _min) is calculated. For example, “x ″ _min = x_min−x′_min”.

  In S614, by subtracting the lower left coordinate value (x'_min, y'_min) in S612 from the upper right coordinate value (x_max, y_max) in S611, the relative coordinate value (x '' _ max, y of the object's upper right coordinate value). '' _max). For example, “x ″ _max = x_max−x′_min”.

  FIG. 13A shows a flowchart of object deletion executed by the object deletion unit 50.

  In step S <b> 701, the object deletion unit 50 acquires information on an object to be deleted. The input data in S701 may be at least the ID of the object to be registered.

  In step S <b> 702, the object deletion unit 50 deletes the deletion target object input in step S <b> 701 from the object detail table 23. Specifically, the object detail table 23 is searched using the ID of the object to be deleted as a search key, and the corresponding record is deleted from the object detail table 23 (see Table 3).

  In S703, the object deletion unit 50 deletes the object to be deleted from the spatial data (region index tree 21, in-region object table 22) as shown in FIG. 6 (see FIG. 13B).

  FIG. 13B is a flowchart showing details of the deletion processing (S703) from the spatial data (region index tree 21, in-region object table 22) executed by the object deletion unit 50.

In step S711, information on a spatial area that intersects the circumscribed rectangle of the object to be deleted is inserted into the intersection area list. The information of the space area to be inserted is information obtained as a corresponding record by searching the intra-area object table 22 (see Table 1 or the like) using the ID of the object to be deleted as a key.
In S712, it is determined whether or not all elements (spatial regions) in the intersection region list have been acquired. When the determination condition is satisfied, this flowchart is ended, and when the determination condition is not satisfied, the process proceeds to S713.
In S713, one element is acquired from the intersecting area list and set as a selected area. An object that intersects with this selection area can be acquired from the column of objects corresponding to the selection area of the in-area object table 22.

  In step S <b> 714, the object to be deleted corresponding to the selected area in the in-area object table 22 is deleted from the in-area object table 22. Further, the object to be deleted is deleted from the object nodes linked from the leaf nodes in the selected area of the area index tree 21. By this deletion process, there is a margin for the capacity of the object in the selected area.

  In S715, the number of divisions including the selected region is equal to the adjacent four sets of spatial regions (for example, the four sets of Morton code values = 48, 52, 56, and 60 in FIG. 6). It is determined whether or not the total number of existing objects falls below the capacity number (for example, three). When the determination condition is satisfied, the process proceeds to S716, and when the determination condition is not satisfied, the process proceeds to S712.

In S716, the region merging unit 51 merges the four sets of spatial regions determined in S715 into one spatial region. Specifically, the following processing is performed.
Merging object nodes linked from each of the four spatial regions (for example, the four spatial regions of Morton code values = 48, 52, 56, 60 in FIG. 6) into one object node; The objects included in the object nodes before the merge (in FIG. 6, objects D, F, and G) are grouped into merged object nodes.
Of the four sets of spatial regions, information on the three spatial regions other than the lower left (in FIG. 6, the spatial regions of Morton code values = 52, 56, 60) is deleted from the leaf nodes of the region index tree 21. As a result, when a (empty) leaf node having no space area information is created, the tree structure is changed so as to delete the leaf node.
Update the space area information (lower area of Morton code value = 48 in FIG. 6) that exists in the lower left of the four space areas. Specifically, the object node linked from the space area information is merged into an object node, and 1 is subtracted from the number of divisions in the space area information (in FIG. 6, the number of divisions (the number in parentheses of the space area information) is “3” is subtracted from “2”).

According to the embodiment described above, the object registration unit 40 arranges the shape and position of the object in the multidimensional space, the object deletion unit 50 deletes the object, and the object search unit 30 is input. Search for objects that meet the search criteria. Then, the spatial data management device 1 manages the spatial data in which the objects are arranged with a balance tree data structure called a region index tree 21 as a spatial region. In the leaf node of the region index tree 21, information of one spatial region is stored for one spatial region, and in the object node, the circumscribed rectangle of the object is expressed by relative coordinates from the representative point of the spatial region that overlaps the object. Has been.
As a result, the disk capacity required for storing the area index tree 21 can be reduced, and the object search, registration, and deletion processes can be realized at high speed. Therefore, the high speed and small capacity associated with the effect of the area index tree 21 are particularly effective in embedded software in which use restrictions on hardware resources are severe.

DESCRIPTION OF SYMBOLS 1 Spatial data management apparatus 21 Area | region index tree 22 Area | region object table 23 Object detail table 30 Object search part 31 Search initial setting part 32 Intersection determination part 33 Search origin setting part 34 Absolute coordinate value conversion part 40 Object registration part 41 Relative coordinate value Conversion unit 42 Area division unit 50 Object deletion unit 51 Area merging unit 60 Data switching unit 91 Central processing unit 92 Main storage unit 93 Secondary storage unit 94 Input unit 95 Output unit A, B, C, D, E, F, G , H, I objects

Claims (10)

A spatial data management device for managing spatial data in which objects are arranged in a multidimensional space,
The spatial data management device includes a storage unit that stores spatial data, and an object search unit that searches the spatial data for objects that exist within a designated search range,
The storage unit stores a region index tree that stores the spatial data as a balance tree that limits the number of child nodes from each node;
Each node of the stored region index tree stores a predetermined number or less of spatial region information obtained by dividing the spatial data at least once, and the spatial region information is stored on the multidimensional space of the spatial region. Contains a key value that is the result of mapping to a scalar value that uniquely identifies the position,
A link to each child node is associated with a non-leaf node having a child node of the region index tree according to a range of key values that each child node can take, and the child node of the region index tree A leaf node that does not have a link from the spatial region information to an object where a circumscribed rectangle intersects the spatial region,
The object search unit
When the search range is specified through the input device,
In order from the root node of the region index tree, a search code value obtained by mapping a representative point of the search range to a scalar value and a key value of a spatial region of each node of the region index tree are collated, whereby a non-leaf node A space traced from the search range for each round with respect to a procedure for extracting a spatial region including a representative point of the search range and an object linked from the spatial region by tracing the link to the leaf node and the object Repeat by updating the representative points of the area excluding the area as a new search code value,
When there is no region excluding the spatial region traced from the search range, for each object extracted by the process of tracing each node of the region index tree, the circumscribed rectangle of the object is collated with the search range, and at least a part An object that intersects is output as an object that intersects the search range. The spatial data management device calculates a Morton code value as a scalar value on spatial data indicating a key value of the region index tree and a search code value indicating a representative point of the search range, respectively. The spatial data management device according to 1. The spatial data management device uses a multidimensional space of a spatial region as a memory capacity for storing a scalar value on spatial data indicating a key value of the region index tree and a search code value indicating a representative point of the search range. The spatial data management device according to claim 1, wherein a minimum capacity for expressing a possible value of a scalar value that uniquely indicates an upper position is determined. The object search unit further includes an actual shape of the object stored in the storage unit and an actual shape of the input search range for an object at least partially intersecting the circumscribed rectangle of the object and the search range The spatial data management device according to any one of claims 1 to 3, wherein an object that intersects at least a part is output as an object that intersects the search range. For each object linked from the region index tree, a type ID indicating the type of the object is associated with the storage unit that stores the spatial data,
When the search range and type ID are specified as a search condition, the object search unit selects an object that matches the type ID of the search condition from the objects extracted by tracing from the region index tree, The spatial data management device according to claim 1, wherein the spatial data management device outputs the spatial data. The storage unit for storing spatial data stores a plurality of the region index trees corresponding to a plurality of spatial regions,
When the search range is specified as a search condition, the object search unit selects one region index tree from a plurality of region index trees based on position information on spatial data indicated by the search range The spatial data management device according to claim 1, wherein the spatial data management device is a spatial data management device. The spatial data management device further includes an object registration unit for registering an object in the region index tree,
When the object to be registered is designated via the input device, the object registration unit,
Based on the position information of the registration target object, one or more spatial regions of the region index tree where the registration target object intersects are specified, and from each node storing the specified spatial region information to the registration target object Link
When the number of objects linked from one spatial region exceeds a predetermined number, the spatial region is equally divided to divide the region into a plurality of spatial regions and update the region index tree. The spatial data management device according to any one of claims 1 to 5. The spatial data management device further includes an object deletion unit that deletes an object registered in the region index tree,
The object deletion unit
When an object to be deleted is specified via the input device, the object to be deleted is deleted from the objects linked from each node of the region index tree,
For each node of the area index tree that is the link source of the deleted object, when the number of objects linked from a plurality of adjacent equally divided areas falls below a predetermined number, those spatial areas are converted into one spatial area. The spatial data management device according to any one of claims 1 to 6, wherein the region index tree is updated by performing region integration. A spatial data management method by a spatial data management device that manages spatial data in which objects are arranged in a multidimensional space,
The spatial data management device includes a storage unit that stores spatial data, and an object search unit that searches the spatial data for objects that exist within a designated search range,
The storage unit stores a region index tree that stores the spatial data as a balance tree that limits the number of child nodes from each node;
Each node of the stored region index tree stores a predetermined number or less of spatial region information obtained by dividing the spatial data at least once, and the spatial region information is stored on the multidimensional space of the spatial region. Contains a key value that is the result of mapping to a scalar value that uniquely identifies the position,
A link to each child node is associated with a non-leaf node having a child node of the region index tree according to a range of key values that each child node can take, and the child node of the region index tree A leaf node that does not have a link from the spatial region information to an object where a circumscribed rectangle intersects the spatial region,
The object search unit
When the search range is specified through the input device,
In order from the root node of the region index tree, a search code value obtained by mapping a representative point of the search range to a scalar value and a key value of a spatial region of each node of the region index tree are collated, whereby a non-leaf node A space traced from the search range for each round with respect to a procedure for extracting a spatial region including a representative point of the search range and an object linked from the spatial region by tracing the link to the leaf node and the object Repeat by updating the representative points of the area excluding the area as a new search code value,
When there is no region excluding the spatial region traced from the search range, for each object extracted by the process of tracing each node of the region index tree, the circumscribed rectangle of the object is collated with the search range, and at least a part A spatial data management method characterized in that an object that intersects is output as an object that intersects the search range. A spatial data management program for causing the spatial data management device, which is a computer, to execute a spatial data management method by a spatial data management device that manages spatial data in which objects are arranged in a multidimensional space ,
The spatial data management apparatus has a storage unit for storing spatial data, and object search unit objects present in the specified search range to search from the spatial data, and
The storage unit stores a region index tree that stores the spatial data as a balance tree that limits the number of child nodes from each node;
Each node of the stored region index tree stores a predetermined number or less of spatial region information obtained by dividing the spatial data at least once, and the spatial region information is stored on the multidimensional space of the spatial region. Contains a key value that is the result of mapping to a scalar value that uniquely identifies the position,
A link to each child node is associated with a non-leaf node having a child node of the region index tree according to a range of key values that each child node can take, and the child node of the region index tree A leaf node that does not have a link from the spatial region information to an object where a circumscribed rectangle intersects the spatial region,
The object search unit,
When the search range is specified through the input device,
In order from the root node of the region index tree, a search code value obtained by mapping a representative point of the search range to a scalar value and a key value of a spatial region of each node of the region index tree are collated, whereby a non-leaf node A space traced from the search range for each round with respect to a procedure for extracting a spatial region including a representative point of the search range and an object linked from the spatial region by tracing the link to the leaf node and the object Repeat by updating the representative points of the area excluding the area as a new search code value,
When there is no region excluding the spatial region traced from the search range, for each object extracted by the process of tracing each node of the region index tree, the circumscribed rectangle of the object is collated with the search range, and at least a part A spatial data management program for executing processing for outputting an object that intersects as an object that intersects the search range.

Images