JPH05181411A - Map information collation and update system - Google Patents

Abstract

PURPOSE:To add secular change and scenery information to map information or update it by automatically extracting the map information from a map or scenery image by utilizing the map information. CONSTITUTION:A drawing or photograph is inputted from a scanner by the indication of a processing processor 4, the input image is stored in an image memory 2, and the map stored on a magnetic disk and the input image are positioned. Then the processing processor 4 processes the input image on the image memory 2 to divide the area when the input image is, for example, a light-shade image or extract a line figure when a binary image. Further, the image processing result is collated with map information after coordinate conversion by the positioning, the result is analyzed, and new map information is stored on the magnetic disk 3. Thus, the area division result is collated with the figure data of the map information to extract the secure change and scenery information, thereby updating a map data base.

Description

Detailed Description of the Invention

[0001]

[Industrial application] This invention uses map information registered in advance to analyze maps and aerial photographs, extract secular changes and landscape information, and automatically add and update a database map information. Regarding the collation update method.

[0002]

2. Description of the Related Art Conventionally, when a map or an aerial photograph is analyzed, data is manually input or the image is analyzed only by using an image processing technique. However, it takes too much time and cost to manually input the data, and sufficient accuracy cannot be obtained only by the image processing technology. Also,
There is an example of correcting the image processing result using map data in the analysis of the remote sensing image,
There is no one that collates image processing results using map data, automatically extracts changes, and updates map information.

[0003]

As described above, conventionally, when it is attempted to additionally update map information from a map or an aerial photograph, it takes much time and effort to create the data. The present invention has been made to solve such a conventional problem, and the purpose thereof is to:
By automatically extracting secular change and landscape information from maps and landscape images using pre-registered map data,
It is to provide a map information collating and updating method capable of automatically adding and updating a map database.

[0004]

To achieve the above object, the present invention provides a map information collating and updating method for collating a newly obtained map image with a previously registered map image to update map information. Means for aligning the new map image with the registered map image, means for extracting a characteristic region from the new map image and the registered map image, the characteristic region and each map It is characterized in that it has a collating means for collating the image and a means for updating the registered map image based on the collation result.

[0005]

In the map information collating and updating method according to the present invention, the aerial photographic image and the map prepared in advance are aligned, the projective transformation coefficient for converting the coordinate of the map data on the image is obtained, and the image has the same characteristics. The map database is updated by extracting the secular change and the landscape information by dividing the area into areas and comparing the area division result with the graphic data of the map information.

Therefore, the database can be automatically updated, and the labor can be greatly saved.

[0007]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of a system to which the map information collation updating method according to the present invention is applied. In the figure, reference numeral 1 denotes an input device for inputting an image to be collated, such as a video camera, a scanner for inputting drawings or photographs, and the like. An image memory 2 stores an image input by the image input apparatus 1 and an image processing result. Reference numeral 3 is a magnetic disk in which map information is stored, and an input image, alignment information, collation information, etc. are stored as processing results. Reference numeral 4 is a processing processor that controls the image input device 1, the image memory 2, and the magnetic disk 3 and executes image processing and collation processing, and 1 to 4 are connected by a bus.

Next, referring to FIG. 2, a schematic operation of the system of the embodiment shown in FIG. 1 will be described.

First, in step 21, a drawing or photograph is input from a scanner or the like according to an instruction from the processor 4, and in step 22, the input image is stored in the image memory 2. In step 23, the map stored in the magnetic disk 3 and the input image are aligned. For example, the map of the area corresponding to the input image is searched from the magnetic disk 3, the coordinates of the same point are obtained from the input image and the map, and if the input image is a map, the document "Digital Image Processing for Image Processing" The correspondence between the input image and the map information is obtained by so-called affine transformation known in the art.

In step 24, the image processor 2 executes image processing on the input image in the image memory 2. For example, if the input image is a grayscale image, area division is performed, and if it is a binary image, a line drawing is performed. Is extracted. In step 25, collation is performed between the image processing result of step 24 and the map information coordinate-converted by the alignment of step 23, and the result is analyzed in step 26,
New map information is stored on the magnetic disk 3.

The step 25 is further composed of three steps, which will be described with reference to the flow charts shown in FIGS. First, steps 101 to 1
At 04, a candidate to be collated with the map data is selected from the elements divided by the image processing result at step 24, and collated at steps 105 and 106 to make a determination. Then, in steps 107 to 111, when a region that should correspond to one map data is divided into a plurality of elements, a candidate to be integrated is selected from the elements adjacent to the matching candidates selected in steps 101 to 104, The integration process is repeated until the criterion is satisfied.

Next, a specific operation will be described by taking as an example the case where the input data is a landscape image such as an aerial photograph.

In steps 21 and 22 in FIG. 2, an aerial photograph is input by a scanner or the like in shades or full colors and stored in the image memory 2. Then, in step 23, as shown in FIG. 6, the coordinates (X0, Y0, Z0) in the map coordinate system of the lens center of the camera, which is the camera parameter of the aerial photograph, and each of the map and the input image become a reference point. Four or more sets of coordinates of the same point are obtained, parameters of a conventionally known projective transformation formula are calculated in the literature “Modern Surveying (1), (2)” and the like, and thereafter, the coordinates of the input image (x, The coordinates are converted between y) and the coordinates (X, Y, Z) of the map information.

Next, step 24 will be described. First, the processing target is limited. The shadow area (a in FIG. 10A) is obtained from the pixel value or the lightness obtained from the pixel value and excluded from the subsequent processing targets. Next, as shown in FIG. 3, with respect to the non-shadow pixel A0, the eight non-shadow pixels Ai (i =
1, 2, ..., 8) and the difference in pixel value is obtained, and if | A0-Ai | <D (1) is satisfied for a certain threshold value D, Ai is a pixel belonging to the same area as A0. Further, the above process is recursively executed for Ai until all pixels belong to some region. For example, if the input image is a full-color image, RG
The difference in pixel value is compared by the Euclidean distance in B space.

Next, the collation with the map information in step 25 will be described with reference to the processing example of FIG. 10 according to the flowcharts shown in FIGS. First, in step 101, map data is sequentially searched from the map DB on the magnetic disk. For example, search the data of one house. Next, based on this map data, as shown in FIG. 6, the coordinates (Xi, Yi, Zi) of the house figure are converted into coordinates (xi, yi) on the image by the conversion coefficient obtained in step 23 of FIG. Then, the lower left coordinate (xL, yL) is (min (xi) −δ, min (yi) −δ) (2) The upper right coordinate (xH, yH) is (max (xi) + δ, max (yi) + δ ) (3) The rectangular area of (3) is set as the matching target range for the house, and only the segments within the matching target range among the area division results obtained in step 24 are set as the matching target. Figure 10 (a)
A region b surrounded by a dotted line is a collation target range, and outline polygons c to f included therein are collation target segments. Hereinafter, this will be referred to as a matching target segment. In the above equations (2) and (3), δ is determined in consideration of the alignment accuracy in step 23 and the size of the house figure.

In step 102, the representative segment to be collated is determined from among the segments to be collated, for example, the segment having the largest area. In FIG. 10, the segment f is selected as the representative segment. If there is no representative segment, it is considered that the map data being collated at present is the data that has changed over time or the region division did not go well, so move to step 112 and record that the map data cannot be collated automatically. Then return to step 101. After step 104, the representative segment is collated with the graphic data (h in FIG. 10A). The matching is performed based on the two evaluation functions Φ1 and Φ2. Φ1 represents the degree of coincidence between the representative segment and the graphic data, and Φ2 represents the degree of disagreement. Two thresholds THR1 and THR2 are set for these two evaluation functions, respectively, and the comparison is made by two judgment criteria: judgment 1: Φ1> THR1 (4) judgment 2: Φ2 <THR2 (5). To do. For example, a case of performing matching by template matching will be described with reference to FIG. Figure 1
In 1, a is a house figure and b is a representative segment.
As a result of collating a and b, supposing that a portion where the representative segment and the figure data match is c and a partial region which does not match the figure data is d, Φ1 = (area of c) / (area of a) ...... ( 6) An evaluation function such as Φ2 = (area of d) / (area of a) (7) can be considered.

In step 104, the figure data is moved in parallel by α (| α | <δ) so as to maximize Φ1 between the representative segment and the figure data, and matching is performed.
The position where Φ1 is maximum is found, and Φ2 is found at that position. In step 105, the determination 2 is evaluated, and if the determination 2 is not satisfied, it is determined that the selection of the representative segment is incorrect, the current representative segment is excluded from the matching target segment, and the process returns to step 102. Then step 106
Then, the determination 1 is evaluated, and if the determination 1 is satisfied, it is determined that the collation is successful, and the process proceeds to step 26. If the determination 1 is not satisfied, it is considered that the house is divided into several segments due to the shape of the roof as shown in FIG. In 10 (d), g) is selected and integrated processing (processing after step 107) is performed. First, in step 107, adjacent segments of the representative segment are set as similar segment selection candidates, and those that satisfy the following conditions are selected as similar segments.

(1) The pixel value at the boundary with the representative segment changes gently and reaches the peak of the roof edge.

(2) When both chromas are high, the hues do not differ greatly.

(3) Among the adjacent segments satisfying (1) and (2), the segment having the longest boundary portion is the segment.

If no similar segment exists, it is determined that the representative segment is incorrectly selected, the representative segment selected in step 102 is excluded from the matching target segments, and the process returns to step 102. If there is a similar segment, in step 109, the segment obtained by integrating the representative segment and the similar segment is newly set as a representative segment and collated with the graphic data by the above method (see FIG. 10 (e)). If the judgment 2 is evaluated in the step 110, and if the judgment 2 is not satisfied, it moves to step 114 as a similar segment selection error, the representative segment is returned to the previous state, and the similar segment selected earlier is selected as the similar segment candidate. And returns to step 107. If the determination 2 is satisfied, the determination 1 is evaluated in step 111,
If the determination 1 is satisfied, it is determined that the collation is successful, and the process proceeds to step 26 shown in FIG. 2 (FIG. 10 (f)). Otherwise,
Step 10 to further integrate adjacent segments
Return to 7.

By the above-mentioned processing, it is possible to perform accurate alignment for each house and extract the segment corresponding to each figure data.

Further, the map data recorded in step 112 for which the automatic collation has failed is later recalled as necessary and collated by a conventional manual operation. If necessary, the map information is updated. ..

Next, step 26 will be described. Analysis processing is performed for the segment obtained from the collation, and new map information is acquired. For example, statistics of pixel values are obtained for each segment to obtain a representative color of the area, or a texture is cut out and stored in the map information in association with each part of the graphic data as shown in FIG. .. At this time, the analyzed color or texture may be directly recorded, or the individually obtained matching position information may be recorded and appropriately extracted from the image when necessary.

Further, as another analysis example, for example, when the area corresponding to the roof is divided into a plurality of segments,
It is possible to estimate the shape of the roof from the way of dividing it. As shown in FIG. 9, a division pattern of a typical roof shape is prepared in advance, and the roof shape is registered by collating this pattern.

Further, by analyzing the segment remaining after the collation with all the map information is completed, it is possible to newly extract a building or extract a region such as a forest or a waterside from the color information. is there.

As shown in FIG. 8 as an application example of these landscape information, only the map information created from the map could only display the landscape by arranging blocks. It is conceivable to display the landscape with its own color, texture and shape.

Further, in the present embodiment, the area division is performed in the image processing of step 24. However, it is possible to change the edge extraction according to the target image and the comparison between the edge and the graphic data. Is.

Next, as another embodiment, the case of inputting a map will be described.

In steps 21 and 22, the scanner is used to input into the image memory in the same manner as in the case of aerial photography, and in step 23, three or more sets of reference points are obtained, and the parameters of the affine transformation formula are obtained. In step 24, conventionally known raster vector conversion is executed to extract lines, closed figures, characters and the like. In step 25, the comparison with the map information is performed, but unlike the case of the aerial photograph, the process of integrating the adjacent regions is not necessary, so only steps 101 to 106 are executed. Referring to FIG. 12 as an example, the collation between the extracted vector graphics a to d and the house graphics e to i of the map information will be described. First, in step 101, the matching target areas j to n are determined, and the respective representative figures are selected. In step 104, the figures are compared with each other, the determination criteria of steps 105 and 106 are used, and those that can be automatically determined are automatically updated, and the map data that cannot be compared is updated after visual confirmation. To do. In FIG. 12, it can be seen that the house figures h and i in the map information have changed over time, and the map information is changed to d.

In the case of inputting an underground buried pipe drawing or the like, in step 25, the road figure of the map information is collated with the vector figure extracted from the drawing, and in step 26, it is included in the collated road vector. Vectors are extracted and stored in the map information as underground buried pipe figure data.
If a map for classifying the regional attributes by color or pattern is input, the map is divided for each color or pattern in step 24, and in step 25, the degree of overlap with the segment extracted for each land use figure is obtained. , Register the percentage. Alternatively, the collation is omitted and the area graphic and its attribute are newly registered in the map information.

As described above, in this embodiment, by inputting various maps and landscape images based on the map information registered in advance, the map information can be updated and new map information can be acquired. It will be possible.

[0033]

As described above, according to the present invention, map information is automatically extracted from a map or a landscape image by using the map information, and secular change or landscape information is added or updated to the map information. The effect of being able to do is obtained.

[Brief description of drawings]

FIG. 1 is a configuration diagram showing an embodiment of a system to which a map information collation updating method according to the present invention is applied.

FIG. 2 is a flowchart showing the operation of this embodiment.

FIG. 3 is an explanatory diagram of area division.

FIG. 4 is a first partial diagram of a flowchart showing a collating operation of map information and an image processing result.

FIG. 5 is a second partial diagram of the flowchart showing the matching operation between the map information and the image processing result.

FIG. 6 is an explanatory diagram showing alignment between a map and an image.

FIG. 7 is an explanatory diagram when analyzing the information of each part of the house from the collation result of the map information and the image processing result.

FIG. 8 is a diagram showing a position application example in the system of the present invention.

FIG. 9 is a diagram showing an example of a dictionary of division results according to roof shapes.

FIG. 10 is an explanatory diagram showing a matching process of an aerial photograph.

FIG. 11 is an explanatory diagram showing criteria for determining a matching result.

FIG. 12 is an explanatory diagram showing a matching process when a map is targeted.

[Explanation of symbols]

 1 image input device 2 image memory 3 magnetic disk 4 processor

Claims (4)

[Claims] 1. A map information collating and updating method for collating a newly obtained map image with a previously registered map image to update map information, wherein the new map image and the registered map image are Positioning means, means for extracting a characteristic region from the new map image and the registered map image, collating means for collating the characteristic region with each map image, and based on the collation result Means for updating the registered map image, and a map information collating and updating method. 2. The map information collation updating method according to claim 1, wherein the means for extracting the characteristic region extracts, for a color image, a region having specific color information. 3. The matching means selects a matching candidate from partial areas extracted by image processing, means for judging matching between the matching candidate and map information, and matching candidates until a judgment criterion is satisfied. The map information collation updating method according to claim 1, further comprising means for selecting an integrated area from the adjacent partial areas to integrate the adjacent area and the adjacent partial areas. 4. The unit for selecting the integrated area selects, from the adjacent partial areas, a partial area that satisfies the condition of the same area and has the largest number of adjacent pixels as an area to be integrated. The map information collation updating method according to claim 3.