JPH0512342A - Retrieving method for image information data base - Google Patents

Abstract

PURPOSE:To realize the image information data base retrieving method for retrieving image information containing retrieval key information at a high speed with high precision by using a retrieval key indicating a position. CONSTITUTION:This retrieving method consists of a retrieval display part 910 which inputs the retrieval key and displays retrieval information, macro-object retrieval processing 970 which retrieves an image containing an object larger than a geometrical distortion error by using the retrieval key, local distortion correction processing 1140 which corrects the latitude conversion coefficient of the display image and registers the image, micro-object retrieval processing 1240 which retrieves an image containing an object smaller than the geometrical distortion error by using the corrected latitude conversion coefficient, a data file part 900 which is stored with a table used for the retrieval, an image analytic part 940 which enlarges the retrieved image and corrects the distortion, and a data registration assistance part 930 which registers a newly generated table.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention highly accurately retrieves a target image from an image file device storing a large amount of image data such as a remote sensing image (hereinafter referred to as image), or a display image. The present invention relates to an image information database search method for easily and highly accurately searching for information about an object inside.

[0002]

2. Description of the Related Art As a conventional image retrieval method, the meteorological satellite image (NOA) described in Technical Report IE 87-90, 1987, pp. 36-43.
A) There is a search method in the information database system. In the search method in this system, first, a keyword is input to the image file device, and the image index (or directory) corresponding to this keyword is searched and displayed on the display device. The index is identification information consisting of a file name, a device, a date, etc., which is represented by letters and numbers, and a plurality of indexes stored in the image file device are read, and an index that matches the keyword is selected from them. If the desired information cannot be retrieved, change the keyword or combination of keywords and repeat the above procedure.

The search method using the above-mentioned index is, for example, when a relational database is used for the data structure, SQL (Structured Query Languag) of the search language.
e) is provided as a computer software tool.

[0004]

The image stored in the image filing apparatus in the above-mentioned prior art uses, as an index, information unique to each image, for example, shooting data such as shooting date and sensor type. Therefore, it is easy to search for an image that matches the shooting specifications, but it is difficult to directly search for an image that includes a specific point or area.

The captured image obtained by the satellite is corrected so as to match the projected coordinates of the Mercator map.

To retrieve an image containing a particular object,
The image may be directly searched by the position information. Specifically, the latitude and longitude of the four corners of the captured (corrected) image (hereinafter referred to as latitude and longitude) are registered in advance, and the corresponding image is searched depending on whether the search position is included in the latitude and longitude of the four corners. To do.

As shown in FIG. 1 (a) (also in FIG. 2 (a)), an index polygon (indicated by a dotted line) formed by latitude and longitude at four corners of an image formed by the search key 10 (a, b) in the index space. Rectangular area) 20, A (Lo1, La
1), B (Lo2, La2), D (Lo4, La4),
It is determined whether or not it is included in C (Lo3, La3), and a search is performed.

As an improvement of the conventional method using the image index, S which is a relational database search language is used.
The method shown in (Equation 1) using the judgment condition function of the QL command can be applied.

[0009] La defines a coordinate axis in the vertical direction (latitude direction) and Lo defines a horizontal axis (longitudinal direction). SELECT image identification number INTO DA FROM image table WHERE (a ≥ Lo1 AND (OR) a ≥ Lo3) AND (a ≤ Lo2 AND (OR) a ≤ Lo4) (number 1) AND (b ≥ La3 AND (OR) b ≥ La4) AND (b ≤ La1 AND (OR) b ≤ La2) (Equation 1) is an example of searching an image including a search key of latitude and longitude coordinates (a, b). This means that an image identification number satisfying the area condition represented by the WHERE clause as shown in FIG. 1A is retrieved from the image table for managing images and stored in the data area DA. Below, data area D
The description about A is omitted.

Since the index polygon formed by the latitude and longitude of the four corners of the photographed image is a convex quadrangle of an arbitrary shape, it is erroneously rejected or erroneously detected. This occurs because the polygons in the index space are quadrangles of arbitrary shape.

The conventional method represented by (Equation 1) can perform a highly accurate search only when the index polygon is a right-angled quadrangle, but when the index polygon is a convex quadrangle having an arbitrary shape, erroneous rejection or erroneous search occurs. Therefore, it is effective to use the position information of the four corners of the image as the index. However, when the above SQL command is used as the search language, the SQL command can accurately determine whether or not the search key exists in an arbitrary quadrangle. There is a problem that cannot be judged.

FIG. 1 (a) shows an example of erroneous rejection, which is the case where the AND condition is searched in (Equation 1).
The difference between the polygon 30 by the command and the polygon 20 by the index becomes the false rejection 40. In addition, FIG.
This is an erroneous detection example in the case of searching under the OR condition in (Equation 1), and the difference between the polygon 60 by the SELECT command and the polygon 50 by the index becomes the erroneous detection 70.
As can be seen from FIGS. 1 (a) and 2 (a), SEL
The ECT command has a problem that false rejection or false detection occurs.

As shown in FIG. 1 (b) (also in FIG. 2 (b)), the search key 10 (a, b) is an index polygon (dotted line) composed of the latitude and longitude of the four corners of the image formed in the index space. Rectangular area) 20, A (Lo
1, La1), B (Lo2, La2), D (Lo4, L
a4), C (Lo3, La3) It is determined whether or not it is included and searched. The polygon 30 by the SQL command shown in FIG. 1B is the polygon 20 by the index.
1 and the coordinate components of the four corners of the polygon 20 by the index (in FIG. 1B, Lo1, Lo4,
It is a rectangle formed by using La1 and La4), and is hereinafter referred to as an inner polygon. Further, the polygon 60 by the SQL command shown in FIG. 2 includes the polygon 50 by the index and the polygon 50 by the index.
Coordinate components of the four corners (Lo2, Lo3,
It is a rectangle formed by using La2 and La3), and is hereinafter referred to as an outer polygon.

As an improvement of the conventional method using the image index, S which is a relational database search language is used.
Using the search condition specification function of QL (Formula 2) to (Formula 4)
The method shown in can be applied.

La defines a coordinate axis in the vertical direction (latitude direction) and Lo defines a horizontal axis (longitudinal direction). DECLARE CR1 CURSOR FOR SELECT Image identification number FROM Image table WHERE (a ≥ Lo1 AND (OR) a ≥ Lo3) AND (a ≤ Lo2 AND (OR) a ≤ Lo4) (Number 2) AND (b ≥ OR3) b ≧ La4) AND (b ≦ La1 AND (OR) b ≦ La2) OPEN CR1 (Equation 3) FETCH CR1 INTO DA (Equation 4) (Equation 2) includes a search key for latitude and longitude coordinates (a, b) Specify the image as the search set, and use the specified set as the cursor CR1
It is an example to correspond to. (Equation 3) is an example of opening the cursor CR1. (Equation 4) is an example in which the position of the cursor CR1 is moved to the next record and the data of that record is taken out to the data area DA. The series of SQL commands of (Equation 2) to (Equation 4) retrieves an image identification number satisfying the area condition represented by the WHERE clause of (Equation 2) from the image table for managing the image and stores it in the data area DA. Means that. Hereinafter, description regarding the data area DA will be omitted. Further, when the AND condition is used for the search in the WHERE clause of (Equation 2), the image data including the search key (a, b) in the inner polygon area can be searched for, and the OR
When the search is performed under the condition, the image data including the search key (a, b) in the outer polygon area can be searched.

Since the index polygon formed by the latitude and longitude of the four corners of the photographed image is a convex quadrangle of an arbitrary shape, it is erroneously rejected or erroneously detected. This occurs because the polygons in the index space are quadrangles of arbitrary shape.

The conventional methods represented by (Equation 2) to (Equation 4) can perform highly accurate retrieval only when the index polygon is a rectangle. However, if the index polygon is a convex quadrangle having an arbitrary shape, false rejection or erroneous retrieval occurs. Therefore, it is effective to use the position information of the four corners of the image as the index. However, when the SQL command is used as the search language, the SQL command accurately determines whether or not the search key exists in an arbitrary quadrangle. There is a problem that cannot be done.

FIG. 1 (b) shows an example of erroneous rejection, which is a case of searching under the AND condition in (Equation 2). The difference between the polygon 30 (inner polygon) by the SQL command and the polygon 20 by the index is erroneous rejection 40. Become. See also FIG.
(B) shows an example of erroneous detection, which is a case of searching with the OR condition in (Equation 2), and the polygon 6 by the SQL command is used.
The difference between 0 (outer polygon) and the polygon 50 due to the index is the false detection 70. As can be seen from FIG. 1B and FIG. 2B, the SQL command has a problem that erroneous rejection or erroneous detection occurs.

Estimating from the Landsat satellite Thematic Mapper sensor image, about 6% of false rejection and false detection occurs. In the case of spot satellite HRV (High Resolution Visible) sensor images, there is perspective imaging, so the imaging area is a square that is more lazy than the thematic mapper sensor image, and it is thought that false rejection and false detection will be even greater. .

The above erroneous rejection and erroneous detection lead to deterioration of detection accuracy, which is a serious problem in the image database.

Further, when the position information is simply used as an index, the above-mentioned corrected image has motion characteristics that cannot be estimated by the satellites and sensors, and undulations of the ground surface for which a mathematical model cannot be determined by realistic computer processing capability. There is a geometrical distortion caused by. Therefore, the object on the image is not located at the correct latitude and longitude on the corrected image. In addition, when the target to be searched is smaller than the geometric distortion correction accuracy, there is a problem that the target cannot be searched accurately. Similarly, when searching for information related to an object on the display image, because of geometric distortion of the image, there is a problem that the object cannot be accurately searched when the target to be searched is smaller than the geometric distortion correction accuracy. .

An object of the present invention is to provide a method for searching information with high accuracy even if geometric distortion is superposed on information such as accumulated images, and to use position information as a search key to store information in a database. The object is to provide a method for retrieving captured images with high precision and easy operation.

[0023]

[Means for Solving the Problems]

(1) The problem of retrieving an image including a specific position or a specific region of an image is solved by adding an index related to shooting data for each image and using position information as an index. That is, the shooting specifications and position information are used as the search key. As the position information, for example, latitude and longitude of map coordinates, lines and pixels of image coordinates (hereinafter referred to as line pixels) are used. When map coordinates are used as the position information, a conversion formula from (latitude, longitude) to (line, pixel) (including a reverse conversion formula is referred to as a latitude-longitude conversion formula) is used in order to associate it with image coordinates.

(2) Two alternatives have been devised for the problem of searching by determining whether or not the search key is included in a quadrangle of arbitrary shape formed in the index space.

(I) Polynomial search method As shown in FIG.
Represented by l2, l3, and l4, it is determined by (Equation 5) to (Equation 8) whether the search key 80 at the o point is included in the quadrangle ABCD 90. When (Equation 5) to (Equation 8) are satisfied, the search key 80 exists inside the quadrangle. The coordinates of each vertex are registered in advance for each image. In FIG. 3, LA
T represents latitude and LONG represents longitude.

The coordinates of each point are A: (Lo1, La
1), B: (Lo2, La2), C: (Lo3, La)
3), D: (Lo4, La4), and o: (a, b).

[0027] In addition to the above-mentioned first-order polynomial, it can be expressed by a higher-order polynomial according to the property of the image.

(Ii) Compound search method As shown in FIG. 4, the candidate image information search 110 is performed on the search key 100, and after the search condition is determined, the candidate information that may cause false detection or false rejection, that is, The detailed image information search 130 is performed only for the fuzzy candidate 120, and the search information 150 is obtained by the two-stage search method for searching the fixed candidate 140. That is, the candidate image information search 110 is a search by the SQL command represented by (Equation 4), and the detailed image search 130 is a linear search represented by (Equation 5) to (Equation 8).

There are four search cases by the SQL command as shown in FIG. The inner index 160 of the search by the SQL command means an inner judgment area according to the AND judgment condition, and the outer index 170 means an outer judgment area according to the OR judgment condition.

When the search key satisfies the conditions of including both the internal and external index spaces, it means that there is the corresponding information in the database, and it can be said that the search 180 has been established. When the both inclusion conditions are not satisfied, it means that there is no corresponding information, and it can be said that the search 180 is still confirmed. If the inner index condition is not satisfied and the outer index condition is satisfied, false rejection or false detection may occur, which can be said to be a fuzzy search 190. Therefore, fuzzy search requires more detailed search. Here, in principle, there is no case where the inner index condition is satisfied but the outer index condition is not satisfied, and therefore it is not necessary to consider it in the determination condition.

(3) For the problem that the target image and related information cannot be searched due to geometric distortion, the following method was devised.

(Iii) Small objects are managed as a group belonging to large objects located in the vicinity.

(Iv) The latitude / longitude conversion formula of the area including the object forming the group is obtained by comparing a reference image with a person from the display device to obtain the position error, and the latitude / longitude conversion formula is obtained based on the position error. Modify the conversion factor of.

(4) The following was devised for the problem of searching by determining whether the search key is included in a quadrangle of arbitrary shape formed in the index space. However, the fixed candidate used below means information that has been determined to be a search target candidate, and the fixed non-candidate means information that has been determined not to be a search target candidate.

As shown in FIG. 4, a candidate image information search 110 is performed on the image data 75 and the search key 80,
After determining the search condition and searching the confirmed candidate 140, the search information 150 is obtained by the two-step search in which the detailed image information search 130 is performed only on the fuzzy candidate 120, that is, the candidate information that may cause false detection and false rejection.

As shown in FIG. 45, the candidate image information search 110 is performed for the image data 75 and the search key 80.
First, by executing the SQL outside polygon inclusion determination search 1400 by the SQL command using the OR condition in (Equation 2) to (Equation 4), the fixed non-candidate 1500 is discarded and the search target is limited. Next, search for inclusion determination in inner polygon 16
00 to determine the information satisfying the determination condition as the confirmation candidate 14
When 0 or the determination condition is not satisfied, the detailed image information is searched for as the fuzzy candidate 120. Here, the coordinates (a, b) of the search key and the latitude and longitude coordinates of the four corners of the image are respectively upper left (Lo1, La1), upper right (Lo2, La2), lower left (Lo3, La3), lower right (Lo4, La4). Then, the determination sentence of the inner polygon inclusion determination search 1600 is (Equation 9)
Can be expressed as

IF (a ≧ Lo1 AND a ≧ Lo3) AND (a ≦ Lo2 AND a ≦ Lo4) (Equation 9) AND (b ≧ La3 AND b ≧ La4) AND (b ≦ La1 AND b ≦ La2) 4 FIG. In the detailed image information search 130, a quadrangle having an arbitrary shape as shown in FIG.
4 is a polynomial search for determining whether or not the search key 80 (o) is included in the quadrangle ABCD 90 by using (Equation 10) to (Equation 13). From (Equation 10) to (Equation 1)
When 3) is satisfied, the search key 80 (o) exists inside the rectangle. The coordinate value of each vertex is registered in advance for each image. In FIG. 3, LAT indicates latitude and LONG indicates longitude. The coordinate value of each point is A: (Lo1, La1),
B: (Lo2, La2), C: (Lo3, La3),
D: (Lo4, La4) and o: (a, b).

11: a ≧ (Lo1-Lo3) (b-La1) / (La1-La3) + Lo1 (Equation 10) 12: a ≦ (Lo2-Lo4) (b-La4) / (La2-La4) + Lo4 ( (Equation 11) 13: b ≤ (La2-La1) (a-Lo2) / (Lo2-Lo1) + La2 (Equation 12) 14: b ≥ (La4-La3) (a-Lo3) / (Lo4-Lo3) + La3 ( (13) The search case by the SQL command is 4 as shown in FIG.
There is an individual. Index 1 in search by SQL command
Reference numeral 60 denotes an inner judgment area based on the AND judgment condition,
The outer index 170 means an outer judgment area according to the OR judgment condition. If the search key satisfies the conditions of including both the internal and external index spaces, it means that there is corresponding information in the database, and it can be said that the definite search 180 is performed. If both inclusion conditions are not satisfied, it means that there is no corresponding information, and it can be said that the definitive search 180 is still obtained. If the inner index condition is not satisfied and the outer index condition is satisfied,
There is a possibility of false rejection and false detection, and ambiguous search 1
I can say 90. Therefore, fuzzy search requires more detailed search. Here, if the inner index condition is satisfied and the outer index condition is not satisfied, it is not possible in principle, so it is not necessary to consider it in the determination condition.

[0039]

(A) By using the position information as the index, the object area can be directly searched regardless of the image type and the photographing time. For example, when searching the same area of different types of sensor-captured images such as a semmatic mapper image and an HV image (HRV) image, the target image can be searched with the same search key, that is, one piece of position information. If latitude and longitude are used, the map and image can be handled with the same search key.

If a conversion equation for associating latitude and longitude with line pixels is obtained at the time of search, conversion from latitude and longitude to line pixels or vice versa can be freely obtained. Therefore, it is possible to uniquely obtain the relationship between the latitude and longitude that are map coordinates and the line pixel that is image coordinates. A polynomial may be used for the conversion formula.

(B) The inventions of (i) and (ii) above operate as follows.

(I) Polynomial search method A quadrangle having an arbitrary shape is represented by a line format, and the inclusive relationship between the search key and the index space is directly determined. Therefore, it is possible to perform a spatial position search in which false detection and false rejection do not exist in principle.

(Ii) Compound search method A detailed search is performed only on ambiguous candidate information that may cause false detection and false rejection. Therefore, detailed image information retrieval is not performed for all cases,
It may be performed only when it cannot be determined by the candidate image information search. When the search positions are random, the frequency of cases that cannot be determined is proportional to the area, and therefore geometric determination may be performed for about 10% of the number of searches, which leads to a reduction in calculation time.

(C) The inventions of (iii) and (iv) above operate as follows.

(Iii) Small objects can be searched by managing objects that are located close to each other and are larger and smaller than the geometric distortion correction accuracy as a group. That is, first, the position information is used to search for a large object. A large object can be searched for an image including the object because the object is large even if geometric distortion is superimposed. Then, the searched object is compared with the reference object information to obtain the positional deviation amount. By using the obtained positional deviation amount, the local geometric distortion near the large object is corrected. In other words, the geometric distortion of the area forming the group has been corrected. The conversion coefficient of the latitude-longitude conversion formula is modified in consideration of the search error of the target object, and the conversion coefficient is used to search for a small target object at the end. In this way, it is possible to search for small objects by managing the objects as a group of large objects and small objects and performing a two-stage search process.

(Iv) To detect the search error, the object is displayed on the display device, the reference image is registered in advance as a reference image, and the reference information file
The reference image of a large object is searched using the object name or the like as a search key, and the positional error between the reference image and the search object is measured to obtain the correction amount of the conversion coefficient, and the latitude / longitude conversion coefficient is corrected. As a result, the geometric distortion of the area forming the group is corrected and a small object can be searched.

(D) The device of (4) above operates as follows.

The detailed image information retrieval is performed only for ambiguous candidate information that may cause false detection and false rejection. Therefore, detailed image information search need not be performed for all cases. It may be performed only when the determination cannot be made by the candidate image information search. If the search position is random,
Since the frequency of undecidable cases is proportional to the area, it is sufficient to perform geometrical judgment about 10% of the number of searches, which leads to a reduction in calculation time.

[0049]

Embodiments of the present invention will be described below with reference to the drawings.

[Embodiment 1] An image information database system, which is an embodiment of the present invention and which carries out a survey (analysis) for updating a map using satellite images or the like, will be described with reference to FIGS. 6 to 16.

FIG. 6 is a block diagram of the image information database system. The system includes an image, an object, various related tables such as latitude and longitude line pixel conversion coefficients, and a data file unit 200 for accumulating image data, input of a search request, and data is searched from the data file unit 200 and displayed on the display 640. Search display unit 210
A search support unit 220 that efficiently searches data and a display support unit 230 that supports image cutout display and the like. The arrows in the figure show the flow of information. Further, FIG. 7 shows the hardware configuration. The user 240 uses the workstation 250 to search various information stored in the disk device 260. The minicomputer 270 uses the image processing dedicated device 280 and the image display device 290 to display and analyze the retrieved image information.

The data structure of this system uses the relational database (RDB), and the search language uses the SQL language described above.

The details of the search process will be described below.

(1) Latitude / Longitude Retrieval Processing 300 FIG. 8 shows a flowchart of processing procedure (hereinafter referred to as processing flow).

By this processing, the object name is input from the workstation 310 in FIG. 6 as shown in FIG. 9 (block 320), and the latitude and longitude of the object are automatically calculated using the object table shown in FIG. (330). The column of the object table, that is, the column, includes the object name, the latitude and longitude of the object (for example, the true latitude and longitude based on the map of the Geographical Survey Institute (hereinafter abbreviated as latitude and longitude)).
There are as many rows, or records, as there are objects in the object table. It should be noted that the values shown in the table in the figure are not accurate. Approximate values are entered to explain the examples.

For example, when retrieving the latitude and longitude of Haneda Airport, the SQL command is as follows.

SELECT ALL (latitude, longitude) FROM Object table WHERE Object = "Haneda Airport" (Equation 14) If the object name does not exist, "No applicable object name" is displayed on the workstation. (340), this processing ends (350).

If the object name exists, the information (object name, latitude, longitude, size, etc.) of the retrieved candidate object is displayed on the workstation (360). The user selects the target object (370), outputs the latitude and longitude of the selected target object (380), and the present process ends.

Here, it is assumed that the latitude and longitude coordinates of the retrieved Haneda Airport are (a, b).

(2) Detailed Image Information Retrieval Processing 390 FIG. 11 shows a processing flow. This process uses the latitude and longitude output from the latitude / longitude search process 300 as a key, and
This is a process of searching for an image identification number using the image table of No. 2 (when the latitude and longitude are known in advance, this process is started without passing through the latitude / longitude search process 300 as shown in FIG. ). In the image table, the sensor name, the shooting date, the latitude and longitude of the four corners of the shot image, the passrow (corresponding to the satellite shooting location identification address), the satellite name, the area name, and the image identification number for identifying the image data Is listed.

First, enter the latitude and longitude (400),
A detailed image information search 410 is performed. In the detailed image information search 410, whether or not the latitude / longitude of the object is included in the latitude / longitude of the four corners of the captured image is checked to find the corresponding image identification number and image information by checking the inclusion relation.

The image identification number including the latitude / longitude coordinates (a, b) of Haneda Airport and other information are the latitude / longitude coordinates of the four corners registered in the table at the upper left (Lo1, L).
a1), upper right (Lo2, La2), lower left (Lo3, La)
3) and the lower right (Lo4, La4), (Equation 15)
You can search with the SQL command represented by.

[0063] According to (Equation 15), the image identification number and the image information corresponding to the latitude and longitude input from the image table are searched, and finally these are output (420), and this processing is ended.

As described above, one or more sets of information such as the image identification number are searched.

(3) Image selection processing 430 Next, a target image is selected from the searched candidate images.

FIG. 13 shows an image selection processing flow. First, some image information obtained as a result of the detailed image information search 390 is input (440), and the information is displayed on the workstation (450) like the intermediate output of FIG. The user looks at the image information and selects an image (460), and the identification number and information of the selected image is output to the workstation 450 (470).

(4) Image cut-out processing 480 Next, the selected image is centered on the input object 51.
The image display device 2 shown in FIG. 7 is obtained by cutting out 2 × 512 pixels.
Processing for displaying on the screen of 90 is performed.

FIG. 14 shows an image cutout processing flow.
First, the image identification number selected by the image selection process 430 and the latitude / longitude of the object selected by the latitude / longitude search process 300 are input (490). An image including the target object is read out from the image data stored in the data file unit 200 of FIG. 6 using the image identification number as a key (5
00), and then again using the image identification number as a key in FIG.
The conversion coefficient of the image is searched from the latitude / longitude line pixel conversion coefficient table of (510).

Here, the latitude / longitude line pixel conversion coefficient is defined as follows.

(L, P) = f (Lo, La) (Equation 16) Here, La represents latitude, Lo represents longitude, L represents a line, and P represents a pixel.

(Equation 16) is a conversion formula from latitude and longitude to line pixels, and f is expressed by a polynomial. The coefficients of the polynomial are
It is calculated in advance at the time of geometric distortion correction processing of the image.

Next, the input object latitude / longitude is converted into line pixels by the above conversion coefficient, and is displayed on the screen of the image display device 290 which is cut out at 512 × 512 pixels centered around Haneda Airport as shown in FIG. (520).

Although the first embodiment has been described above, maps other than satellite images can be handled in the same manner by the search method of the present invention. In that case, the map information is stored in the map table as shown in FIG. Further, in the image selection process 370 in FIG. 8, a method of overlaying the candidate image on the image display device in addition to the candidate image information and allowing the user to visually select the image is also effective.

[Embodiment 2] Like Embodiment 1, Embodiment 2 is an image information database system for conducting a survey for map update using satellite images and the like. This embodiment is a speed-up of the first embodiment.

This embodiment will be described with reference to FIGS. 17 to 19.

FIG. 17 is a block diagram of the image information database system. Similar to the first embodiment, the search display unit 53
0, a search support unit 540, a data file unit 550, and a display support unit 560. Unlike the first embodiment in that the candidate image information search process 580 is added in the search support unit 540, as shown in FIG. 5, after the search in the candidate image information search process 580, there is a possibility that erroneous detection or false rejection may occur. Image information only Detailed image information retrieval processing 63
This is a two-step search method in which 0 is performed.

The hardware configuration of the present invention is the same as that of FIG. 7, and the screen display example is the same as that of FIG. Further, similar to the first embodiment, the data structure of this system is RDB and the search language is SQL.

Details of the search will be described below. However, description of the same processing as that of the first embodiment is omitted.

(1) Latitude / Longitude Search Process 570 First, the name of the object to be searched is input from the workstation 670 to search the latitude and longitude of the object. The processing content is the same as the latitude / longitude search processing in the first embodiment, and when the latitude / longitude of the object is known in advance, this processing is omitted and the next candidate image information search processing 580 is performed.

(2) Candidate image information retrieval processing 580 Next, the candidate image information is searched.

FIG. 18 shows a candidate image information search processing flow. This processing uses the latitude and longitude searched by the latitude / longitude search processing 570 as a key, and uses the image table shown in FIG. 12 to determine fixed candidate information and ambiguous candidate information. The inner polygon and the outer polygon used below are the same as those defined in FIGS. 1 and 2, respectively.

As shown in FIG. 18, the search key, that is, the latitude and longitude of the object is input (590). It is determined whether or not the coordinate (a, b) of the latitude / longitude, which is a key, exists in the inner polygon formed by the index.
00 is performed.

The latitude and longitude coordinates of the four corners of the image forming the polygon are set to the upper left (Lo1, La1) and upper right (Lo), respectively.
2, La2), lower left (Lo3, La3), lower right (Lo
4, La4).

SELECT ALL Latitude and longitude of four corners Image identification number Sensor name Satellite name Date of imaging FROM image table WHERE (a ≥ Lo1 AND a ≥ Lo3) AND (a ≤ Lo2 AND a ≤ Lo4) (Equation 17) AND (b If ≧ La3 AND b ≧ La4) AND (b ≦ La1 AND b ≦ La2) (Equation 17) is satisfied, the image selection processing 660 is performed. If not satisfied, the outer polygon search process 620 of (Equation 18) is performed.

SELECT ALL Latitude and longitude of four corners Image identification number Sensor name Satellite name Imaging date FROM image table WHERE (a ≥ Lo1 OR a ≥ Lo3) AND (a ≤ Lo2 OR a ≤ Lo4) (Equation 18) AND (b If ≧ La3 OR b ≧ La4) AND (b ≦ La1 OR b ≦ La2) (Equation 18) is satisfied, detailed image information search processing 630 is performed. If the information is not satisfied, the corresponding information is displayed on the workstation 670.

(3) Detailed Image Information Retrieval Processing 630 Next, according to the result of the candidate image information retrieval processing 580,
A detailed search is performed for search information that does not satisfy the inner polygon but does satisfy the outer polygon. This is called a detailed image information search 630.

This processing consists of the linear format retrieval shown in (Equation 5) to (Equation 8), and FIG. 19 shows the detailed image information retrieval processing flow. First, the search key (a, b) and the search information in the candidate image information search processing 580 are input (6
40). Next, a search process 650 is performed using a polynomial.

The latitude / longitude coordinates of the four corners of the image searched in the candidate image information search processing 580 are set to the upper left (L
o1 ', La1'), upper right (Lo2 ', La2'), lower left (Lo3 ', La3'), lower right (Lo4 ', La
4 '). Actually, there are a plurality of search images, and the following processing is repeated for the number of search images.

[0089] If (Equation 11) is satisfied, the image selection search processing 660 is performed on the search information, and if it is not satisfied, the corresponding information is displayed on the workstation 670. (4) Image selection process 660 Next, a process of selecting a target image from the searched candidate images is performed.

This processing is the image selection processing 430 of the first embodiment.
13 and the processing flow is shown in FIG. Display information about candidate images on workstation 670,
Let the user select from the candidate images.

(5) Image cut-out processing 680 Next, the selected image is displayed centering on the input object.
2 × 512 pixels are cut out and displayed on the screen of the image display device.

This processing is the image cutting processing 4 of the first embodiment.
Similar to 80, the processing flow is shown in FIG. The object of the selected image is cut out with 512 × 512 pixels centered and displayed on the display 690.

As described above, the search key is the data file 5
It is possible to determine at a high speed whether or not it is registered in 50 and is included in the display image. However, detailed image search process 6
In addition to the detailed search at 30 as described above, a method of thinning out the candidate image information and displaying it directly on the image display device for visual selection by a human is also effective.

[Third Embodiment] A third embodiment will be described below with reference to FIGS.

FIG. 20 is a block diagram of the image information database system of this embodiment. This system has a function of marking a position corresponding to an object input by the user on the image displayed on the image display device screen 705. The hardware configuration is the same as in FIG. 7. FIG. 21 shows a screen display example.

As shown in FIG. 20, the system is an object,
A data file section 700 for accumulating various related tables such as latitude / longitude line pixel conversion, a search display section 710 for inputting a search request and searching for and displaying data from the data file section, and a search support section 72 for efficiently searching for data.
0, a display support unit 7 for marking an object on an image
It is composed of 30. The arrows in the figure show the flow of information.

The details of the search will be described below. In this embodiment, an example is shown in which marking is performed on the thematic mapper image obtained by photographing the Kanto region displayed on the image display device 705. Further, the description of the same processing as in the first and second embodiments will be omitted.

(1) Latitude / Longitude Line Pixel Conversion Coefficient Search Process 740 This process is performed with the image identification number of the display image as a key in FIG.
This is a process of searching the latitude / longitude line pixel conversion coefficient of the display image using the latitude / longitude line pixel conversion coefficient table.

FIG. 22 shows a latitude / longitude line pixel conversion coefficient search processing flow. First, the image identification number of the display image is input 750 and the latitude / longitude line pixel conversion coefficient search 760
I do. Finally, the latitude / longitude line pixel conversion coefficient retrieved is output 770, and this processing is terminated.

For example, the latitude / longitude line pixel conversion coefficient of the image with the image identification number 0031 can be retrieved by the following SQL command from the latitude / longitude line pixel conversion coefficient table of FIG.

SELECT ALL Latitude / Longitude Line Pixel Conversion Coefficient FROM FROM Latitude / Longitude Line Pixel Conversion Coefficient Table (Equation 20) WHERE Image Identification Number = “0031” (Equation 20) You can search the latitude / longitude line pixel conversion coefficient corresponding to. However, this processing may be performed when the image is displayed.

(2) Line Pixel / Latitude / Longitude Conversion Process 780 This process is performed by line pixels at four corners of an image, for example,
(1,1), (1,512) (512, 1), (51
2, 512), the latitude and longitude line pixel conversion coefficient search 7
Using the conversion coefficient f retrieved by 40, (Equation 16)
Is a process of converting to latitude and longitude. Here, the converted latitude and longitude coordinates of the four corners are set to the upper left (Lo1, La
1), upper right (Lo2, La2), lower left (Lo3, La2)
3) and the lower right (Lo4, La4), the conversion formula is (Equation 2)
It can be represented by 1).

(Lo1, La1) = (1 / f) (1,1) (Lo2, La2) = (1 / f) (1,512) (Equation 21) (Lo3, La3) = (1 / f) (512, 1) (Lo4, La4) = (1 / f) (512, 512) (3) Object search processing 790 Next, the four corners of the display image of the object input by the user from the workstation 800. The latitude and longitude of the object within the longitude and other object information are searched using the object table.

FIG. 23 shows a flow chart of the object search processing. First, enter the latitude and longitude of the four corners of the image and the object name (81
0). In the next object search 820, the name of the object and the other information in which the input object name is the same as the registered name in the table and are within the latitude and longitude of the four corners of the display image are searched. Finally, output the latitude and longitude of the searched object and information (83
0) This process ends.

Here, (Equation 14) is the description by the SQL command when marking "airfield" and "airport" in the image. However, the latitude and longitude coordinates of the four corners of the display image are set to the upper left (Lo1, La1) and upper right (Lo2, La2), respectively.
La2), lower left (Lo3, La3), lower right (Lo4, L
Let (a, b) be the object latitude / longitude coordinates retrieved as a4).

[0106] By using (Equation 22), it is possible to retrieve the latitude and longitude and the information of the target object within the latitude and longitude of the four corners of the image using the target object table.
In this process, in determining the inclusion relation of latitude and longitude, only the polynomial as shown in (Equation 15) is used.
As in the candidate image information search and the detailed image information search in the second embodiment, the determination may be made by a two-step search using both the SQL command and the polynomial.

(4) Latitude / Longitude / Line Pixel Conversion Processing 840 Next, the latitude / longitude of the object searched by the object searching processing 790 is converted into the latitude / longitude line pixel conversion coefficient searching processing 74.
This is a process of converting into a line pixel as shown in (Equation 16) using the latitude / longitude line pixel conversion coefficient searched for by 0.

(5) Marking Process 850 Next, a process of marking the retrieved object position on the image is performed.

FIG. 24 shows a marking processing flow. First, the lines and pixels of the object converted by the latitude / longitude / line pixel conversion processing 740 are input (86
0), marking the image display device screen as in the output of FIG. 21 (870) and displaying the information of the marked object on the workstation 880.

As described above, as shown in the output of FIG. 21, the positions of Haneda Airport, Narita Airport and Atsugi Airfield in the display image and their information can be obtained.

[Embodiment 4] An image information database system according to a fourth embodiment of the present invention will be described with reference to FIGS. 25 to 39. This system is a system that local governments use satellite images to conduct surveys for map updates.

FIG. 25 is a block diagram of the image information database system. The present system includes a data file unit 900 that stores images, reference information, and latitude / longitude conversion coefficients, and a search request input / data file unit 9
Search registration display section 910 for searching and displaying data from 00
A search support unit 920 that efficiently searches data, a data registration support unit 930 that prevents double registration of data, and an image analysis unit 940 that analyzes a search image. The arrows in the figure show the flow of information.

The data structure of this system is the above-mentioned RDB,
The above-mentioned SQL is used as the search language. All command examples in the embodiments of the present invention conform to the above-mentioned manual specifications. However, the present invention does not depend on the search language given as an example.

The search method of the present system treats a large object (hereinafter referred to as a macro object) and a small object (hereinafter referred to as a micro object) as a set of groups as compared with the geometric distortion amount on the satellite image. The target micro object can be searched for by a multi-step search in which a macro object is searched and then a micro object is searched. Below, spots (SPO
T) An example of searching the XX bank near Shinjuku Station from the satellite H.R.V. (HRV) image and enlarging the building will be described.

From the workstation 950 of the search registration display unit 910, enter a search request such as "Shinjuku Station, XX bank" (the data input device is not limited to the workstation as long as it can input and output numbers and characters). . Here, Shinjuku Station and the bank belong to the same group. The method of creating a group will be described in the registration process.

HRV accumulated in the data file 960
It is assumed that geometric distortion of about 200 meters square remains in the image even in the corrected image. Here, the size of Shinjuku Station is 300 meters square and the size of the bank is 30 meters square.

(1) Macro Object Retrieval Processing 970 FIG. 7 shows a macro object retrieval processing flow. An image including Shinjuku Station, which is a macro object, is searched by this processing flow. The latitude / longitude search 980 of the object corresponding to the initially input object name is performed. The latitude / longitude (La, Lo) of Shinjuku Station is obtained using the object table shown in FIG. La represents latitude and Lo represents longitude. The record (horizontal row) of the object table is composed of the object name and the latitude and longitude of the object (hereinafter abbreviated as latitude and longitude). Columns (vertical columns) are present only for the object (note that the numerical values and the like described in the table in the figure are not accurate. They are entered to explain the embodiment).

For example, when searching for the latitude and longitude of Shinjuku station,
The SQL command is as follows.

SELECT ALL (latitude, longitude) INTO DA FROM object table WHERE object = 'Shinjuku station' (Equation 23) This is to search the latitude and longitude where the object name corresponds to Shinjuku station from the object table. Means

If the object name does not exist, "No corresponding object name exists" is displayed on the workstation 950 (1000), and this processing ends. If the object name exists, the next image search 1010 is performed.

Next, an image search 1010 is performed. Using the image table of FIG. 28, as shown in FIG.
Whether or not Shinjuku station is included in 0 is checked for inclusion relation, and a corresponding image to be included is searched. The image table 1020 includes sensor types, pass / low (corresponding to satellite image capturing location identification addresses), latitude and longitude of four corners of a captured image, image capturing date such as image capturing date, cloud amount, sensor image capturing angle, and image data. The stored image identification number is described.

Here, the data file section 900 is shown in FIG.
As shown in FIG. 5, a table 1050 in which each information is related to each other and a data file 1060 in which a large amount of data is stored are stored in a pointer (identification number) 10
Related at 70. For example, the data file section 90
The portion related to the image of 0 is composed of an image table in which image specifications are described and an image file in which the image data is stored, and both are related by an image identification number.

The image identification number of the image including the object of latitude and longitude (a, b) is searched as follows.

SELECT ALL image identification number INTO DA FROM image table WHERE (a ≦ upper right latitude OR a ≦ upper left latitude) AND (a ≧ lower right latitude OR a ≧ lower left latitude) AND (b ≦ upper right longitude OR b ≦ lower right longitude ) AND (b ≧ upper left longitude OR b ≧ lower left longitude) (Equation 24) where a ≧ b is a or more than b, a ≦ b is a or less than b, and OR and AND are logical sums and logicals, respectively. Means product.

Further, upper right latitude, upper left latitude, lower right latitude, lower left latitude, upper right longitude, lower right longitude, upper left longitude, lower left longitude: each means the latitude and longitude of the four corners of one scene of the photographed image shown in FIG.

When there is no image identification number satisfying (Equation 24), "there are no search image candidates" as shown in the workstation display example of FIG. 31 (a).
Message is displayed on the workstation 950 and the macro object search process 970 ends. The following processing after the local distortion correction processing is not performed. If there is one, the corresponding image is transferred from the image file to the image display 1095 and displayed (1085). When there are a plurality of images, candidate image identification numbers and image data are displayed on the workstation 950 as shown in FIG. 31B (1100). The user visually judges this display (112
0), the corresponding image is selected, transferred from the data file 1060 and displayed (1130).

(2) Local Distortion Correction Processing 1140 This processing corrects the conversion coefficient for position correspondence from the latitude / longitude to the line pixel, or from the line pixel to the latitude / longitude in the area including the group of objects. Consists of processing. This conversion coefficient is called a latitude / longitude conversion coefficient. The method of calculating the latitude-longitude conversion coefficient before correction is calculated in advance in the pre-processing of the satellite ground station, which will be described later, and stored in the latitude-longitude conversion coefficient file.

The image and the latitude / longitude conversion coefficient are related by the latitude / longitude conversion coefficient identification number.

According to the local distortion correction processing flow of FIG. 32, the latitude / longitude conversion coefficient is corrected. First, the reference table shown in FIG. 33 is searched for a reference image serving as a reference for Shinjuku Station (1160).

FIG. 34 shows a search image 1170 of the macro object.
Indicates. The image 1170 and the reference image 1180 retrieved by the macro object retrieval process 970 are displayed side by side on the image display 1095.

The search for the reference image 1180 is performed as follows.

SELECT ALL Reference image identification number INTO DA FROM reference table WHERE object = 'Shinjuku station' (Equation 25) If the reference image 1180 exists, the image display 109 searches for the corresponding image.
5 is displayed (1190). If it does not exist, "No corresponding reference image exists" is displayed on the workstation (1200), and the local distortion correction processing 11
40 ends.

When the reference image 1180 exists, the reference image 1180 and the search image 1170 are displayed side by side on the display device, and the error amount (dL between the reference point 1210 of the target visually instructed and the corresponding point 1220 instructed by the search (dL , DP) is measured (1230). The amount of error g corresponds to the distance between the Shinjuku station 1040 that exists at the true position on the macro object search image 1170 and the search corresponding point 1220. dL is the line direction and dP is the error amount in the pixel direction.

Here, the latitude / longitude conversion formula is defined as follows.

(L, P) = f (Lo, La) (Equation 26) (Lo, La) = g (L, P) (Equation 27) La represents latitude and Lo represents longitude. Then, L represents a line and P represents a pixel.

(Formula 26) is a conversion formula from latitude / longitude to line pixels, (Formula 27) is a conversion formula from line pixels to latitude / longitude, and f and g represent polynomials.

The coefficient of the polynomial is calculated in advance at the time of the geometric distortion correction processing of the image.

The present invention does not depend on polynomial order. In this embodiment, the following affine transformation formula is used.

L = f _L (Lo, La) = a ₀ + a ₁ Lo + a ₂ La (Equation 28) P = f _P (Lo, La) = b ₀ + b ₁ Lo + b ₂ La (Equation 29) Lo = g _Lo (Equation 29) L, P) = c ₀ + c ₁ L + c ₂ P (Equation 30) La = g _La (L, P) = d ₀ + d ₁ L + c ₂ P (Equation 31) Using the error amount (dL, dP), the coefficient of the polynomial Fix
A corrected latitude / longitude conversion coefficient is calculated (1225).

F _L (Lo, La) + dL = F _L (Lo, La) (Equation 32) f _P (Lo, La) + dP = F _P (Lo, La) (Equation 33) g _Lo (L + dL, P + dP) = G _Lo (L, P) (Equation 34) g _La (L + dL, P + dP) = G _La (L, P) (Equation 35) The coefficients of the polynomials F and G are called modified latitude / longitude conversion coefficients.

The corrected latitude / longitude conversion coefficient is managed by the identification number in the latitude / longitude conversion coefficient table of FIG. 35 together with the latitude / longitude conversion coefficient before correction. Coefficients are stored in the latitude / longitude conversion coefficient file.

These corrected latitude / longitude conversion coefficients are created for each of the photographed images of the small areas forming the group. The size of the small area is
For example, it is defined as 512 pixels square in accordance with the image size of the display device. The storage period of the modified latitude / longitude conversion coefficient depends on the capacity of the data file, but is stored during the analysis period of the target image to avoid reprocessing.

(3) Micro Object Retrieval Processing 1240 The latitude and longitude of the micro object XX bank 1250 (a,
b) is obtained using the object table of FIG.

First, from the object table, XX bank 125
Search 0 latitude and longitude (a, b) in SQL (126
0).

SELECT ALL (a, b) INTO DA FROM object table WHERE object = '○ × bank' (Equation 36) Next, using the modified latitude-longitude conversion coefficient F, the latitude-longitude conversion 1270 is used to calculate the value of ○ × bank 1250. Position on display image (L, P)
Ask for.

[0146] _{L = F L (a, b} ) ( number _{37) P = F P (a} , b) ( number 38) coordinates (L, P) corresponds to the exact position of the micro-object,
The corresponding coordinates on the image display 1095 are marked with a colored cursor 1280 (1290).

As described above, the micro object can be searched and displayed.

(4) Image analysis processing 1300 Image processing is performed on the detected micro object. The image processing command of the image analysis processing unit 940 is sent to the workstation 950.
More input, and the result is displayed on the image display 1095. Here, the enlargement processing is performed mainly on the XX bank 1250. The enlargement process may be performed by the cubic convolution process described on pages 137 to 141 of “Image Data Processing and Analysis (1)” edited by the Resource Observation and Analysis Center, March 1989.

As described above, H with ambiguity in position has occurred.
From the RV image, it was possible to search the ◯◯ bank 1250 and perform the desired image processing.

The method of creating the latitude / longitude conversion coefficient and the method of registering the object data will be described below.

(5) Creation of Latitude / Longitude Conversion Coefficients The latitude / longitude conversion coefficients f and g before correction can be calculated in advance when an image is received by the satellite ground station and geometric distortion correction processing is performed. For example, since the latitude / longitude coordinates at the position of one line and one pixel can be specified during the geometric distortion correction processing, it is possible to associate the line pixel coordinates with the latitude / longitude coordinates.
If the coordinates of the corresponding points at the four corners are calculated, (Equation 28) to (Equation 3)
The latitude / longitude conversion coefficient which is the coefficient of 1) can be obtained.

The geometric distortion correction processing is described in, for example, Koichi Ihara and Yasunari Yamamoto, "Precision Correction Technology for Earth Observation Satellite," Journal of the Institute of Electrical Engineers, Vol. 101, No. 4, pp. 317 to 324. Method may be used.

Although the latitude and longitude are used as the search key or the index in the above embodiment, line pixels of the image may be used instead of the latitude and longitude. in this case,
Correspondence between the search key and the index may be performed directly by line pixels.

When the line pixel is used as the search key or the index, the conversion from (latitude, longitude) to (line, pixel) becomes unnecessary.

Further, the position coordinates of the macro object are represented as "parent" and the position coordinates of the micro object are represented as "child", the position coordinates of the parent are represented by absolute coordinate values of the image or map, and the position coordinates of the child are parent. It may be expressed by a deviation amount from the coordinates of. this is,
Affects the construction method of the object table. There are four alternatives to the object table of FIG. 27, as shown in FIG.

A (1315) is the macro object, o ₁ (1
316) and o ₂ (1317) are micro objects having A as a parent. The image coordinates and map coordinates of each object are (L,
P), (Lo, La), (L1, P1) (Lo1, La
1), a (L2, P2) (Lo2, La2), polarization of the child to the parent is o _1, o _2, each image coordinates, map coordinates, (ΔL1, ΔP1) (ΔLo1 , ΔL
a1) and (ΔL2, ΔP2) (ΔLo2, ΔLa2).

Each deviation amount of the micro object o ₁ (1316) is ΔL1 = L−L1 (Formula 39) ΔP1 = P−P1 (Formula 40) ΔLo1 = Lo−Lo1 (Formula 41) ΔLa1 = La−La1 ( The deviation amount of the micro object o ₂ (1317), which is the equation (42), can be similarly defined. Method 1 is represented by map coordinates for both the parent and child.
In method 2, both parent and child are represented by image coordinates. In Method 3, both the parent and the child are represented by map coordinates, but the child is represented by the deviation from the parent. In Method 4, both the parent and the child are represented by image coordinates, but the child is represented by the deviation from the parent.

The method adopted in this embodiment is method 1. Other methods can also be realized according to the processing configuration of FIG. In method 2 and method 4, the latitude / longitude conversion process from (latitude, longitude) to (line, pixel) is not necessary. Further, in the methods 3 and 4, when registering the latitude and longitude of the child in the table, it is necessary to register the deviation amount.

Although the search process has been mainly described according to the system configuration of FIG. 25, the data registration also has a problem of position ambiguity. For example, in the data registered in the past and the data to be registered now, the accuracy of the latitude / longitude conversion coefficient may change over time (various information can be assumed to be available. This improves the position accuracy, but improves it). This may result in double registration.

If double registration data occurs in the data file and the user cannot manage it, the information reliability of the database will be deteriorated. This is solved by the following method. FIG. 38 shows a block diagram of the image information database system. The difference from FIG. 25 is that the registration support unit 93
There is a point where 0 is detailed.

The registration support unit 930 uses the double registration judgment processing 1
320 and registration processing 1330. FIG. 39 shows a processing flow of the registration support unit.

(1) Double registration determination processing 1320 (b) Input The object name and group code to be registered are input from the workstation. Further, the position information of the object is input by instructing it on the image display 1095. Here, as input information, for example, the object name is a □□ department store, the line pixels are (L3, P3), and the latitude and longitude are (Lo).
3, La3), the quality flag is 1 (meaning that the reliability is low), and the group code (01) belongs to Shinjuku Station.

The quality flag Q registered in the object table is, for example, Q = 10: quite high accuracy, Q =
5: Suspicious, Q = 0: The accuracy is considerably low. For example, register according to the certainty of the information, and input it when registering the information. The reliability of the information is entered by the database user (information reader). In the latitude / longitude conversion 1380, the line pixel is converted into the latitude / longitude by the number (30) and the number (31).

Lo3 = G _Lo (L3, P3) (Equation 43) La3 = G _La (L3, P3) (Equation 44) Next, double registration of data occurs using the information of the object name and position. Check if there is no.

(B) Allowable range setting 1390 Using the quality flag, the search range value (ΔLo, Δ
La) is calculated. The search range value is variable depending on the value of the quality flag.

As a method of variably setting, for example, there is a method of considering the quality flag in the reference search range value (ΔLo _b , ΔLa _b ). For example, the reference search range (Δ
Lo _b , ΔLa _b ) is set to a value corresponding to 5 times the spatial resolution of the HRV image, which is about 3 seconds (100 meters). When the quality flag is α, ΔLo = ΔLo _b · β (Equation 45) ΔLa = ΔLa _b · β (Equation 46) β = 1 / α (Equation 47) When reliability is low, (ΔLo, ΔLa) is large. Set the value.

(C) Object name search 1400 Information about objects that may be double-registered, which are located in the vicinity of objects to be registered next, is displayed as object names, group codes, and quality flags. Search from the table. SELECT ALL object name, group code, quality flag INTO DA FROM object table WHERE (Lo3-ΔLo ≤ latitude ≤ Lo3 + ΔLo AND La3-ΔLa ≤ longitude ≤ La3 + ΔLa) (number 48) (d) group code determination 1410 Next searched. It is determined whether or not the group code and the input group code of the registration object are the same.

(A) When the group codes are the same The positions are close and the group codes are the same, so there is a risk of double registration. Therefore, the message "There is a risk of double registration, please examine the possibility of registration" is displayed on the workstation. Then, the registrant determines whether or not to register. If not registered, the registration process ends. When registering, there is a risk of double registration, so the quality flag is reset to a value with low reliability. Quality flag change process 14
Make 40.

(B) When the group codes are different When the group codes are different but the positions are close to each other, it can be determined that the same object may be double-registered in different groups.
“Similar data will be registered in different groups. Please examine whether registration is possible ”is displayed on the workstation (145
0). If not registered, the registration process ends. When registering, it is determined whether the registered data is modified (1460). In the case of correction, it is updated using the SQL change command (1470). If not corrected, the group code of the data to be registered is changed (148).
0).

(2) Registration Process 1330 The registration process 1330 may be executed by SQL as follows.

INSERT INTO object table (CODE object name, latitude / longitude, group, quality flag) VALUES (○ □ bank, Lo3, La3, 01, 1) (Equation 49) Due to the above, position ambiguity has occurred. You can register the object.

In the above embodiment, as an alternative to the local distortion correction processing 1140 in FIG. 25, instead of the processing of detecting the reference point of the reference image and the corresponding point of the search image and visually determining the error amount, shape recognition is performed. The reference point and the corresponding point may be automatically detected by the processing. For shape recognition, SSDA (Sequential Similarity Detection A) described in “Introduction to Computer Image Processing” by Tamura, published by Soken Shuppan 1985
lgorithm) method, which may be applied.

In this embodiment, first the macro object is searched,
This is a method of performing local strain correction by comparing with reference information serving as a reference, and performing a multi-stage search for micro objects. As a result, the position of the object in the image file can be directly specified by the image having the position ambiguity and the search key, and the image containing the object at the arbitrary position can be searched with high accuracy and easily. In addition to searching, there is an effect that marking on the corresponding target object can be performed with high accuracy. In addition, when registering an object with position ambiguity in the database, there is an effect that the reliability of registration information can be improved by automatically detecting data that may be double-registered.

[Fifth Embodiment] A fifth embodiment will be described below with reference to FIGS. 40 to 44.

FIG. 40 shows the image information database system of this embodiment. The present embodiment is a system for searching the position corresponding to the object name input by the search key and marking the image. This system includes a search registration display unit 910, a search support unit 920, and a data file unit 90.
It consists of zero.

In this embodiment, two search examples will be described.

(A) Retrieval processing of target object information The related information of the instructed target object is searched by inputting an arbitrary position on the image.

(B) Marking process of object Only a specific object is highlighted in the displayed image.

The above (A) and (B) will be described in order.

(A) Retrieval process of object information On the image display 1095, the object on the displayed image is pointed by the cursor, the information related to the pointed object is retrieved from the data file 960, and is sent to the workstation 950. indicate. FIG. 22 shows a search processing flow of object information.

(1) Input of position coordinates of target object 1490 By pointing the target object displayed on the image display 1095 with a cursor, the coordinates (L,
Read P).

(2) Group code input 1500 Enter the group code of the display image.

(3) Search for modified latitude / longitude conversion 1510 A modified latitude / longitude conversion coefficient is searched from the latitude / longitude conversion coefficient table of FIG. 35 using a group code.

(4) Latitude / longitude coefficient conversion 1520 The image coordinates (L, P) are converted into map coordinates (Lo, La).

Lo = G _Lo (L, P) (Equation 50) La = G _La (L, P) (Equation 51) (5) Retrieval of Related Information 1530 By SQL as in the fourth embodiment, the object to be instructed is indicated. Related information is searched from the position information (Lo, La) and the object table and reference table shown in FIGS. 27 and 33.

SELECT ALL Object Name INTO DA FROM Object Table WHERE (Lo-ΔLo ≤ Latitude ≤ Lo + ΔLo AND La-ΔLa ≤ Longitude ≤ La + ΔLa) (Numeric 52) SELECT ALL Image ID Reference Number TO Text Identification Number, Text Identification Number TO WHERE (Lo-ΔLo ≤ Latitude ≤ Lo + ΔLo AND La-ΔLa ≤ Longitude ≤ La + ΔLa) (Equation 53) (6) Display of Search Information 1540 The searched information is displayed on the workstation 950 and the image display 1095. For example, candidate information is displayed as shown in FIG. Also, make it possible to retrieve more detailed information as needed.

(B) Marking process of target object According to the processing flow shown in FIG. 43, the target object of interest in the displayed image is marked. The name of the object of interest (name of the object to be searched) is input by the database user (1570).

(1) Search for latitude / longitude conversion coefficient The corrected latitude / longitude conversion coefficient identification number is searched from the group code of the display image to obtain the modified latitude / longitude conversion coefficient G.

SELECT ALL Modified Latitude / Longitude Conversion Coefficient Identification Number INTO DA FROM Latitude / Longitude Conversion Coefficient Table WHERE Group Code (Equation 54) (2) Latitude / Longitude Conversion of Four Corners 1560 The latitude / longitude of the coordinates of the four corners of the display device is obtained. For example, 51
If the display device has a size of 2 pixels, the coordinates of the four corners are
(1,1), (1,512), (512,1), (51
2, 512), and the latitude and longitude are obtained by the equations (39) and (40). Let each coordinate be (Lo1, La1), (Lo
2, La2), (Lo3, La3), (Lo4, La
4).

Lo = G _Lo (L, P) (Equation 55) La = G _La (L, P) (Equation 56) (3) Input of Search Object Name 1570 Input Search Object Name to Workstation 250 .

(4) Coordinate search of relevant information 1580 The park position is searched using the data calculated by the latitude / longitude conversion processing of the four corners.

[0192]   SELECT ALL Latitude / Longitude INTO DA     FROM target table   Where object name = 'park'     AND ((Latitude ≤ Lo2 OR Latitude ≤ Lo1)     AND (Latitude ≥ Lo4 OR Latitude ≥ Lo3) (Equation 57)     AND (Longitude ≤ La2 OR Longitude ≤ La4)     AND (longitude ≤ La1 OR longitude ≤ La3)) (5) Latitude / longitude conversion 1590 Image coordinates are calculated for all searched latitudes and longitudes.

For example, when the searched latitude and longitude are (a, b), L = F _l (a, b) (Equation 58) P = F _p (a, b) (Equation 59) (6) Marking display 1600 Image coordinates (L, P) converted by the above latitude / longitude conversion processing
The coordinate position on the image display 1095 corresponding to is displayed in a superimposed manner with a colored cursor 1610, as shown in FIG. 44, to emphasize the target position.

(7) Display of object information At the same time as marking, related information is displayed on the workstation 950 as shown in FIG.

As described above, it is possible to highly accurately mark an object on an image in which position ambiguity has occurred and to easily search for related information.

Finally, a sixth embodiment of the present invention will be described based on FIG. 17, FIG. 45 and others.

In this embodiment, the search support unit 540 shown in FIG.
45, the detailed image information search process 6 is performed only on the candidate image information that may be erroneously detected or rejected after being searched by the candidate image information search process 580 as shown in FIG.
Do 30.

Details of each search process in FIG. 17 will be described below.

(1) Latitude / longitude search processing 570 FIG. 8 shows a latitude / longitude search processing flow.

By this processing, the object name is input from the workstation 680 of FIG. 17 as shown in FIG. 9 (320), and the latitude and longitude of the object are automatically calculated using the object table shown in FIG. Search (330). The column of the object table, that is, the column, includes the object name, the latitude and longitude of the object (for example, the true latitude and longitude based on the map of the Geographical Survey Institute (hereinafter abbreviated as latitude and longitude)). There are as many rows, or records, as there are objects in the object table. It should be noted that the values shown in the table in the figure are not accurate. Approximate values are entered to explain the examples.

For example, when searching the latitude and longitude of Haneda Airport, the SQL command is as follows.

DECLARE CR1 CURSOR FOR SELECT Latitude / Longitude FROM Object table WHERE Object name = 'Haneda Airport' (number 60) OPEN CR1 (number 61) FETCH CR1 INTO DA (number 62) (number 60) is the object table In the table, a record satisfying the condition (the object name is “Haneda Airport”) indicated in the WHERE clause is designated as the search set, and is associated with the cursor CR1.

(Equation 61) opens the cursor CR1. In (Equation 62), the position of the cursor CR1 is moved to the next record, and the data of that record is taken out to the data area DA.

If the corresponding object name is not found by the SQL command of (Equation 60) to (Equation 62),
"There is no corresponding object name" is displayed on the workstation (340), and this processing is terminated (350).

When the object name is found, information on the searched candidate object (object name, latitude, longitude, size, etc.)
Is displayed on the workstation (360). The user selects a target object (370), the latitude and longitude of the selected target object is output (380), and this processing ends.

Here, it is assumed that the latitude and longitude coordinates of the retrieved Haneda Airport are (a, b).

(2) Candidate image information retrieval processing 580 FIG. 46 shows a candidate image information retrieval processing flow. This process uses the latitude and longitude searched by the latitude / longitude search process 570 as a key and uses the image table shown in FIG.
This is a process of searching the image information of the confirmed candidate information. Information similar to that of the first embodiment is described in the image table of FIG. The inner polygon and the outer polygon used below are the same as those defined in FIGS. 1 and 2, respectively.

First, input latitude and longitude 4300 and SQ
The L outside polygon search processing 4400 is performed. In the non-SQL polygon search processing 4400, first, an SQL command is used to specify a record satisfying the WHERE clause in the image table as a search set, and data is extracted from the search set one by one (4500). The condition shown in the WHERE clause is that the outer polygon formed by the index includes the latitude / longitude coordinates (a, b) serving as a search key. Next, using the extracted data, the inner polygon inclusion relation determination is performed (4600). The conditional expression in the judgment sentence indicates a condition that the inner polygon formed by the index includes the search key (a, b). When the judgment condition indicated by the conditional expression is satisfied, the definite search (corresponding to) 180 in FIG.
Then, the data of the current record (latitude / longitude of four corners, image identification number, etc.) is output (4700). If the determination condition is not satisfied, the area of the fuzzy search 190 in FIG. 5 is entered, so the detailed image information search 630 is performed.
The non-SQL polygon search processing 4400 is executed from 4500 to 4500 for all the records of the search set specified by the WHERE clause.
The processes up to 700 and 630 are repeated.

The latitude / longitude coordinates of the four corners of the image forming the polygon are set to the upper left (Lo1, La1) and upper right (Lo), respectively.
2, La2), lower left (Lo3, La3), lower right (Lo
4, La4), the information of the image including the search key (a, b) in the inner polygon can be searched by the SQL command represented by (Equation 63) to (Equation 65) and the determination statement represented by (Equation 66).

DECLARE CR1 CURSOR FOR SELECT Latitude / longitude image identification number of four corners Sensor name Satellite name Imaging date FROM image table WHERE (a ≥ Lo1 OR a ≥ Lo3) AND (a ≤ Lo2 OR a) ≤ Lo4 AND (b ≥ La3 OR b b ≥ La4) AND (b ≤ La1 OR b ≤ La2) OPEN CR1 (Equation 64) FETCH CR1 INTO DA (Equation 65) IF ((a ≥ Lo1 AND a ≥ Lo3) AND (Lo3) AND AND a ≤ Lo4) (Equation 66) AND (b ≥ La3 AND b ≥ La4) AND (b ≤ La1 AND b ≤ La2)) (Equation 65) to (Equation 65) are input to the outer polygon. Specify the image table records to be included in the search set (Latitude and longitude of the four corners, the image such as an identification number) data of the specified retrieval data set is a SQL command string to retrieve the one record. (Equation 66) is calculated from (Equation 63) to (Equation 6)
This is a sentence for determining whether or not the input latitude / longitude is included in the inner polygon using the latitude / longitude data of the four corners extracted in 5). SQ from (Equation 63) to (Equation 65)
If the corresponding record is not found by the L command, "There is no corresponding search target" is displayed on the workstation, and this processing ends. If there is even one record satisfying the conditions of (Equation 63) to (Equation 66), the data of the corresponding record (latitude / longitude of four corners, image identification number, etc.) is output (4700).

(3) Detailed Image Information Retrieval Processing 630 Next, detailed image information retrieval 630 is performed for retrieval information that does not satisfy the inner polygon but satisfies the outer polygon.

This processing consists of a judgment sentence using a polynomial. FIG. 47 shows the detailed image information search processing flow.
First, the search key (a, b) and the record data (the latitude and longitude of the four corners, the image identification number, etc.) extracted by the SQL outside polygon search 4400 of the candidate image information search processing 580 are input (5000). Next, it is determined whether or not the search key input using the polynomial is included in the latitude and longitude of the four corners of the captured image (5100). If the determination condition is satisfied, the data (image identification number, other image information) of the current record is output (5200), and this processing ends. If the determination condition is not satisfied, this process ends.

If the input search key and the latitude and longitude coordinates of the four corners are the same as those in the candidate image information search 580, the record data of the image including the search key (a, b) in the inner polygon is expressed by (Equation 67). You can search by sentence.

[0214] (Equation 67) is a judgment sentence that makes a polynomial judgment using input data. When (Equation 67) is satisfied, the data of the current record is output.

(4) Image selection processing 660 Next, a target image is selected from the searched candidate images.

FIG. 48 shows an image selection processing flow. First, some image information obtained as a result of the detailed image information search 630 is input (5400), and the information is displayed on the workstation (5500) like the intermediate output of FIG. 9. The user looks at the image information and selects an image (5600), and as a result, the identification number and information of the selected image is output on the workstation 670 (5).
700).

(5) Image cut-out display processing 680 Next, the selected image is displayed centering on the input object 51.
The image display device 2 shown in FIG. 7 is obtained by cutting out 2 × 512 pixels.
Processing for displaying on the screen of 90 is performed.

FIG. 49 shows an image cutout processing flow.
First, the image identification number selected by the image selection process 660 of FIG. 17 and the latitude / longitude of the object selected by the latitude / longitude search process 570 are input (6000). Using the image identification number as a key, the data file section 55 in FIG.
An image including the target object is read from the image data stored in 0 (6100), and then the conversion coefficient of the image is searched again from the latitude / longitude line pixel conversion coefficient table of FIG. 49 using the image identification number as a key. (6200).

Here, the latitude / longitude line pixel conversion coefficient is defined as follows.

(L, P) = f (Lo, La) (Equation 68) Here, La represents latitude, Lo represents longitude, L represents a line, and P represents a pixel.

(Equation 68) is a conversion formula from latitude and longitude to line pixels, and f is represented by a polynomial. The coefficients of the polynomial are
It is calculated in advance at the time of geometric distortion correction processing of the image.

Next, the input object latitude / longitude is converted into line pixels by the above conversion coefficient, and 512 × 512 pixels centered around Haneda Airport are displayed on the screen of the image display device as shown in FIG. (6300).

As described above, it is possible to quickly determine whether or not the search key is registered in the data file section 550 and is included in the display image. However, in the detailed image information search processing 630, other than searching for detailed image information as described above, it is also effective to thin out the candidate image information and display it directly on the image display device so that a human can visually select it.

[0224]

According to the present invention, since position information is used, there is an effect that a photographed image including a retrieval object can be detected with high accuracy, high speed, and easy operation. Further, there is an effect that various information such as images and maps can be integrated and managed by using the position information.

As described above, since specific point or area information can be directly managed, it is possible to construct a multimedia information processing system in which a database and analysis processing are integrated.

[Brief description of drawings]

[Figure 1] Illustration of false rejection in satellite image retrieval

[Fig.2] Explanatory diagram of erroneous retrieval in satellite image retrieval

FIG. 3 Schematic diagram of polynomial search method

[Fig. 4] Schematic diagram of compound search method

FIG. 5 is an explanatory diagram of search conditions.

[Fig. 6] Image information database system configuration diagram

FIG. 7 is a hardware configuration diagram of the image information database system.

FIG. 8: Flow chart of latitude / longitude search processing

[Figure 9] Image information database system screen display example

FIG. 10 is an explanatory diagram of an object table configuration.

FIG. 11 is a flowchart of detailed image information search processing.

FIG. 12 is an explanatory diagram of an image table configuration.

FIG. 13 is a flowchart of image selection processing.

FIG. 14 is a flowchart of image cutout processing.

FIG. 15 is an explanatory diagram of a latitude / longitude line pixel conversion coefficient table configuration.

FIG. 16 is an explanatory diagram of a map table configuration.

FIG. 17 is a block diagram of an image information database system using a composite search method.

FIG. 18 is a flowchart of a candidate image information search process.

FIG. 19 is a flowchart of detailed image information search processing.

FIG. 20: Image information database system configuration diagram with object marking function

FIG. 21 is a diagram showing a display example of an object marking screen.

FIG. 22 is a flowchart of latitude / longitude line pixel conversion coefficient search processing.

FIG. 23 is a flowchart of an object search process.

FIG. 24 is a flowchart of marking processing.

FIG. 25: Image information database system configuration diagram

FIG. 26 is a flowchart of a macro object search process.

FIG. 27 is an explanatory diagram of an object table configuration

FIG. 28 is an explanatory diagram of an image table configuration.

FIG. 29 is an explanatory diagram of the inclusion relationship between the captured image and the search target.

FIG. 30 is an explanatory diagram of a relationship between a table and a data file.

[Fig. 31] Example of workstation display

FIG. 32 is a flowchart of local distortion correction processing.

FIG. 33 is an explanatory diagram of a reference table configuration.

FIG. 34 is a display example diagram of a search image and a reference image.

FIG. 35 is an explanatory diagram of a latitude / longitude conversion coefficient table configuration.

FIG. 36 is a flowchart of a micro object search process.

FIG. 37 is an explanatory diagram of an alternative plan of a target table

FIG. 38 is an image information database system configuration diagram.

FIG. 39 is a flowchart of registration support processing.

FIG. 40: Image information database system configuration diagram

FIG. 41 is a flowchart of a search process for object information.

42] Image display of search information and example of workstation display

FIG. 43 is a flowchart of a marking process of an object.

[Fig.44] Image display of search information and example of workstation display

FIG. 45 is an explanatory diagram of a candidate image information search method.

FIG. 46 is a flowchart of a candidate image information search process.

FIG. 47 is a flowchart of detailed image information search processing.

FIG. 48 is a flowchart of image selection processing.

FIG. 49 is a flowchart of image cutout processing.

[Explanation of symbols]

10 ... Search key, 20 ... Polygon by index,
30 ... Polygon by SELECT command, 31 ... S
Polygon by QL command, 40 ... False rejection, 110 ...
Candidate image information search, 130 ... Detailed image information search, 530 ... Search display section, 540 ... Search support section, 550
... Data file part, 560 ... Display support part, 570 ... Latitude / longitude search, 580 ... Candidate image information search, 630 ...
Detailed image information search, 660 ... Image selection processing, 680
… Cut display processing.

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Tetsuka Shouyu             5-2-1 Omika-cho, Hitachi City, Ibaraki Prefecture             Ceremony company Hitachi Ltd. Omika factory (72) Inventor Fuminobu Furumura             1099 Ozenji, Aso-ku, Kawasaki City, Kanagawa Prefecture             Ceremony company Hitachi Systems Development Laboratory (72) Inventor Kuniaki Matsumoto             1099 Ozenji, Aso-ku, Kawasaki City, Kanagawa Prefecture             Ceremony company Hitachi Systems Development Laboratory

Claims (32)

[Claims] 1. A search keyword is specified from a search / display section for inputting / outputting data, and the search keyword is used from the data file section storing image data, an image table, a coordinate conversion coefficient table, and an object table. Read the index that represents the position information
Using the index, the candidate image information is searched first, and then the detailed image information is searched. Then, the information corresponding to the index is searched from the data file section by the two-step search process, and the searched information is searched as described above. Image information database search method to be displayed on the display. 2. The process of searching the candidate image information,
2. A process for determining whether or not the search keyword is present in a region surrounded by the coordinates of the four corners of the image data of the quadrangle stored in the data file portion by a comparison calculation process in the SQL command and performing a search process. Image information database search method. 3. The above-mentioned determination processing is a comparison calculation processing by AND and OR logical expressions of the four corners of the image data and SQL commands, and when the search keyword satisfies the AND condition, the information corresponding to the search keyword is 3. The image information database searching method according to claim 2, further comprising a process of determining whether or not the difference information between the search information of the OR condition and the search information of the AND condition that is definitely present cannot be determined by the SQL command alone. . 4. The process of retrieving the detailed image information comprises:
A process of expressing the area surrounded by the coordinates of the four corners of the quadrangle image data stored in the data file section by a polynomial, and determining whether or not the search keyword is included in the area expressed by the polynomial by a polynomial determination and searching. The image information database searching method according to claim 1, further comprising processing. 5. The process of retrieving the detailed image information comprises:
The image information database searching method according to claim 1, further comprising the step of displaying the image information searched for by the processing for searching the candidate image information on a search display unit and visually confirming it by a human. 6. The process of retrieving the detailed image information,
2. The image information database searching method according to claim 1, which is directly executed for a search keyword without performing a process of searching for the candidate image information. 7. The image information database searching method according to claim 1, wherein the image data includes at least one of an image, a map and a document. 8. The image information database searching method according to claim 1, wherein the search keyword is a line or pixel which is an image coordinate, or a latitude and a longitude which are map coordinates. 9. A search keyword is specified from a search / display section for inputting / outputting data, image data, an image table, an object table, and a coordinate conversion coefficient table are stored in a data file section, and the specified search keyword is specified. The latitude and longitude information of the target object is searched from the target object table by using, and the searched latitude and longitude information is converted into the image coordinate position by the coordinate conversion coefficient table, and the target position is marked on the search display unit. An image information database search method for confirming and displaying the position of an object. 10. A search keyword relating to a position on an object is designated from a search / display section for inputting / outputting data,
The image data, the reference image data, the related information table, and the latitude / longitude conversion coefficient are stored in the data file section, an index similar to the above search keyword is read from the data file section, and the geometric distortion error of the image data is read by the index. Specifying a macro object that is larger than that, searching for image data including the macro object and reference image data including the macro object, displaying the searched image data and reference image data in parallel, and displaying on each display. Input corresponding positions of the same object respectively, measure the amount of error between each input position, correct the latitude-longitude conversion coefficient registered in the above related information table based on the amount of error, and correct latitude-longitude conversion Re-register the coefficient in the above data file section and re-register the latitude Degrees by the conversion coefficient to correct the local strain of the image data, the image information database search method for correcting the image data displayed on the search display unit. 11. The correction processing uses the corrected latitude-longitude conversion coefficient by designating a small object with the search keyword as compared with the geometric distortion error of the image data stored in the data file section. 11. The image information database searching method according to claim 10, further comprising a micro object searching process for searching an accurate position of the small object. 12. The image information database search according to claim 10, wherein both latitude and longitude which are map coordinates and lines and pixels which are image coordinates are used as position information for searching the image data stored in the data file section. Method. 13. The image information database searching method according to claim 10, wherein the data file section is composed of a related information table for correlating stored data and an image file for storing image data. 14. The data file section comprises an image table relating to image data, an image file, a reference image table, a reference image file, a latitude / longitude conversion coefficient table and a latitude / longitude conversion coefficient relating to conversion formulas of an image coordinate system and a map coordinate system. 11. The image information database searching method according to claim 10, which comprises a file. 15. A search keyword relating to a position on an object is designated from a search / display unit for inputting / outputting data,
Macro object search process for storing image data, reference image data, related information table and latitude / longitude conversion coefficient in the data file part, and searching macro object position using the above search keyword, and macro object search process By measuring the search error amount by comparing the image data of the position searched by and the reference image data, using the local distortion correction process of correcting the latitude-longitude conversion coefficient by the error amount, and the corrected latitude-longitude conversion coefficient An image information database searching method comprising a micro object searching process for searching a small object position compared to the geometric distortion error of the stored image data. 16. The objects in the macro object searching process and the micro object searching process are registered as one group on the related information table, and the same coefficient is used as the latitude / longitude conversion coefficient. Image information database search method. 17. The image information database searching method according to claim 16, wherein the group is registered with a macro object as a parent and a micro object as a child, and after the parent is searched, a two-step search process of searching the child is performed. . 18. The image information database searching method according to claim 17, wherein the position information of the parent and child objects is represented by absolute coordinates having a common origin. 19. The position information of the parent is expressed in absolute coordinates,
18. The image information database searching method according to claim 17, wherein the position information of the child is represented by relative coordinates that are deviations from the position of the parent. 20. The storing process is performed by inputting relevant information such as an object name, a group code, and position information.
11. The image information database searching method according to claim 10, further comprising the step of creating a new related information table and registering the table in the data file section. 21. In the registration process, the presence / absence of double registration is determined by searching the similarity of the name and position information of the data to be registered and the registered data. If there is a possibility of double registration, 21. The image information database searching method according to claim 20, further comprising: double registration determination processing for issuing an alarm. 22. The image information database searching method according to claim 10, wherein the related information table stores quality flags graded according to the reliability of accumulated data. 23. The image information database searching method according to claim 22, wherein the allowable error when determining the similarity between the index and the search keyword is made variable by the quality flag. 24. The image information database searching method, wherein the double registration judging process includes a process of changing a permissible error when judging the similarity between the index and the search keyword by the quality flag. 25. The image information database searching method according to claim 23, wherein when the reliability indicated by the quality flag is low, the allowable error is wide, and when the reliability is high, the allowable error is narrow. 26. The process of inputting the instruction comprises a process of automatically measuring the error amount by detecting each position of the same object on the display by performing image recognition. Image information database search method. 27. In the process of inputting the instruction, the object name is input, the latitude / longitude information is used as a search keyword, and image data of a partial area including the object is searched from the data file section and displayed on the screen. 11. The image information database searching method according to claim 10, which comprises a process of displaying at least one position of the object from which geometric distortion has been removed on the screen. 28. The process of inputting an instruction comprises a process of searching and displaying related information related to the position of the object from the data file section when the position of the object on the displayed image is instructed. The image information database search method of item 10. 29. The process of searching the candidate image information is SQL which is a relational database search language.
Outer polygon determination processing that determines data that does not definitely satisfy the search keyword by a command, and inner polygon determination processing that determines data that definitely satisfies the search keyword and other ambiguous data for the output data of the determination processing The image information database searching method according to claim 1. 30. In the outer polygon determination process, a search keyword is included in a rectangular area including the area surrounded by the coordinates of the four corners of the quadrangle image information stored in the data file section and formed by the coordinate values of the four corners. 30. The image information database searching method according to claim 29, which comprises a process of determining whether or not it is. 31. In the inner polygon determination processing, the search keyword is included in a rectangular area formed by coordinate values of four corners included in an area surrounded by coordinates of four corners of image information, with respect to output data of the outer polygon determination processing. 30. The image information database searching method according to claim 29, comprising a process of determining whether or not the image information is included. 32. The image information database searching method according to claim 29, wherein the processing for discriminating ambiguous data includes processing for discriminating between data that definitely satisfies the search keyword and data that does not satisfy the search keyword in the detailed image information search.