JP4823798B2 - Map alignment apparatus and method - Google Patents

Description

  The present invention is used to convert map information into digital information. It is particularly suitable for use when managing a house map and a public map in association with each other.

  With the progress of digitalization of map information, map information has been widely used in the electronic information society. There is a strong demand that it would be convenient if a public map could be searched from a house map in the flow of digital information of such map information, and there was a proposal to manage the house map and the public map in full correspondence. is there.

  However, traditional public maps are often created for each character, and the scale and orientation are not constant, and the accuracy is often low. Confirming the correspondence between such a public map and a residential map without error is important in managing the residential map and the public map in a completely corresponding manner.

  In such a confirmation process of the correspondence between the house map and the public map, the worker is forced to repeat the trial and error while fitting the public map into the house map while checking the public map and the house map. In many cases, this work relies on technology based on abundant experience, and the actual situation is that the work is carried out by the hands of a very small number of experienced workers.

  However, the digitalization of map information is an urgent need, and there is an urgent need for the development of a technology that can automate the above-described fitting work without relying on the skills of a small number of experienced workers.

  The present invention has been performed under such a background. For example, the present invention provides a map alignment apparatus and method that can semi-automate the work of fitting a public map into a house map and increase work efficiency. The purpose is to do.

The present invention is a map alignment apparatus that superimposes a predetermined region (a shape in which the outer periphery is divided by a line segment) in the second diagram on a region (a shape in which the outer periphery is divided by a line segment) on the first map. there, the present onset Ming, display means for displaying the area and the second predetermined area of the map on the first map, the operator causes the display of the two regions on the display unit, the two Two or more points that are likely to correspond to a region are designated as feature points, and the same number r (2 ≦ r ≦ (n, m smaller)) points are further selected and extracted from the designated feature points. And, for each of the _n C _r combinations of _r feature points extracted from one region having _n feature points, _{r r} extracted from the other region having m feature points. _{R in} each of the _m C _r combinations of feature points. Corresponding point generation means for generating corresponding points that are pairs of feature points having _n C _r × _m P _r pairs obtained by associating (r factorial) permutations with each other, and the correspondence Corresponding point extracting means for extracting the corresponding points satisfying a predetermined condition indicating that the shapes of the two areas can be superimposed from among the corresponding points of the two areas generated by the point generating means, and Superimposing means for superimposing the two regions so that the sum of squares of the distance between corresponding points, which is a set of feature points having a correspondence relationship between the two sections extracted by the corresponding point extracting means, is minimized. Error determining means for obtaining an average value of distances between corresponding points, which is a set of feature points having a correspondence relationship between two regions superimposed by the overlapping means, and an average value of the distances obtained by the error determining means From the smaller one Position Ru Tei and means for displaying a plurality of overlapping patterns to.

  In the following description, the first map will be described as a house map, and the second map will be described as a public map. However, the scope of the present invention is not limited to only a house map and a public map.

When an operator who handles the map alignment apparatus of the present invention has two regions on a house map and a public map , that is, a shape in which the outer periphery is divided by a line segment, and the feature points are set on the outer periphery , the present invention device The feature points on the two areas are automatically associated with each other. The process takes the step of excluding the apparently inappropriate association after the brute force association for the time being. Furthermore, the device of the present invention superimposes the two regions so that the sum of squares of the distances between the corresponding points of the two regions that have been associated is minimized, and between the corresponding points of the two superimposed regions. An average value of the distances is obtained, and a plurality of overlapping patterns are displayed from a smaller average value of the distances to a predetermined order.

  Thereby, the operator can compare and contrast a plurality of overlapping pattern candidates. The average value of the distance corresponds to an error at the time of superposition, and generally speaking, the smaller the error, the better. However, for example, even if the error is relatively large, the error is evenly distributed. In some cases, an overlay pattern having a visually good appearance may be found, and the determination is left to the operator's judgment.

For example, the predetermined condition is that attention is paid to one set of corresponding points of the plurality of corresponding points generated by the corresponding point generating means, and each of the areas is in the same area as the corresponding corresponding point. The lengths of the line segments that connect to other feature points are obtained, and the line segments obtained in each of the above regions are associated with each other in a one-to-one relationship between the two regions, and the ratio of the lengths of all the line points is determined. This is a condition for excluding the pair of corresponding points obtained for the correspondence between the minutes, and the standard deviation of the ratios obtained for the correspondence between all the line segments deviating from a predetermined range.

  That is, if the selected pair of corresponding points is an appropriate corresponding point, the positional relationship between the corresponding point and other feature points is almost similar. Using this fact, the length of the line segment connecting the corresponding point and other feature points is compared in two regions, and if the comparison of the length ratio includes little variation, It is determined that the corresponding point is appropriate.

At this time, as a predetermined condition , when an average value of angles formed by the line segments corresponding to one-to-one deviates from a predetermined angle, the line is calculated prior to obtaining the length ratio. Conditions for excluding minutes in advance can also be included.

  That is, when the corresponding points are appropriate, the line segments corresponding to one to one are almost parallel to each other. On the other hand, when the corresponding points are inappropriate, the line segments corresponding one-to-one form different angles. As a result, an inappropriate pair of corresponding points can be removed in advance. However, in this case, it is required that the directions of the areas in the two different maps coincide with each other with a certain degree of accuracy.

For example, the superimposing means includes means for performing affine transformation on one of the two regions, x and y are coordinates before conversion, X and Y are coordinates after conversion, s is an expansion / contraction ratio, and θ is When the rotation angle is set and c and d are parameters related to movement,
X = a * x−b * y + c
Y = b * x + a * y + d
a = s * cos θ
b = s * sin θ
Means for converting and superimposing the coordinates.

  The superimposing means may include means for restricting expansion / contraction or rotation in accordance with an operation input when superimposing. According to this, superposition can be performed according to a certain rule. It is possible to perform superposition without performing any one or both of expansion and contraction and rotation, or to limit the amount of expansion and contraction or rotation. Therefore, it is possible to perform an overlaying operation in consideration of the nature (character) of the map.

The present invention can also be viewed from the viewpoint of a map alignment method performed by the map alignment apparatus of the present invention. That is, the map alignment method of the present invention displays areas on two maps, a house map and a public map, and designates two or more points that are likely to correspond to the two displayed areas as feature points, From the designated feature points, the same number r (2 ≦ r ≦ n, smaller one of m) feature points are selected and extracted, and r features extracted from the region having n feature points. For each of the _n C _r combinations of points, r! In each of the _m C _r combinations of _r feature points extracted from the region having _m feature points. Generates a corresponding point is a set of feature points having _n C _r × _m P _r sets of correspondence with one another which is obtained by matching the number of permutations to each other, the two from a plurality of the corresponding points the generated The corresponding points satisfying a predetermined condition indicating that the shapes of the two regions can be overlapped are extracted, and the two sums of the distances between the corresponding points of the two extracted regions are minimized. The two regions are overlapped, an average value of the distances between the corresponding points of the two overlapped regions is obtained, and a plurality of overlapping patterns are displayed from the smaller of the average values of the distances to a predetermined order. .

  For example, the predetermined condition focuses on one set of corresponding points of the plurality of corresponding points generated, and in each of the regions, the corresponding point of interest and other feature points in the same region The lengths of the connecting line segments are obtained, the line segments obtained in each of the regions are made to correspond one-to-one, and the ratio of the lengths is obtained for the correspondence between all the line segments. This is a condition for excluding the pair of corresponding points where the standard deviation of the ratio obtained for the correspondence between them deviates from a predetermined range.

At this time, the predetermined condition is that, when the average value of the angles formed by the line segments corresponding to one-to-one deviates from the predetermined angle, the line is prior to obtaining the length ratio. Conditions for excluding minutes in advance can also be included.

For example, when x and y are coordinates before conversion, X and Y are coordinates after conversion, s is an expansion / contraction ratio, θ is a rotation angle, and c and d are parameters related to movement, respectively,
X = a * x−b * y + c
Y = b * x + a * y + d
a = s * cos θ
b = s * sin θ
The coordinates are transformed by affine transformation and superimposed. Further, in superposition, expansion / contraction or rotation can be restricted according to an operation input.

  The present invention can also be viewed from the viewpoint of a program for realizing the map alignment apparatus of the present invention. That is, the present invention is a program that, when installed in a general-purpose information processing apparatus, causes the general-purpose information processing apparatus to realize functions corresponding to the map alignment apparatus of the present invention. By recording the program of the present invention on a recording medium, the general-purpose information processing apparatus can install the program of the present invention using this recording medium. Alternatively, the program of the present invention can be directly installed in the information processing apparatus via a network from a server holding the program of the present invention.

  Thereby, the map alignment apparatus of this invention is realizable using a general purpose information processing apparatus.

  According to the present invention, for example, the work of fitting a public map into a house map can be semi-automated to improve work efficiency.

  A map alignment apparatus according to an embodiment of the present invention will be described with reference to FIGS. FIG. 1 shows the overall configuration of the map alignment apparatus of this embodiment. The map alignment apparatus of the present embodiment is a map alignment apparatus 4 that superimposes a predetermined area of a public map on an area on a house map. As shown in FIG. A device that newly creates a housing map for searching for a map by referring to the housing map information stored in the housing map DB) 1 and the map information stored in the public map database (public map DB) 2. is there. The created public map search residential map is stored in the public map search residential map database (public map search residential map DB) 5. A display unit 3 for displaying map information, and a keyboard 6 and a mouse 7 for an operator to input operations are provided. Further, the housing map DB 1 or the public map DB 2 or the public map search housing map DB 5 and the map alignment device 4 may be connected via a network (not shown).

In FIG. 3, the example of a display of the housing map and public map of the display part 3 is shown. In the present embodiment , one area in the house map is referred to as map # 1, and the corresponding area in the public map is referred to as map # 2. Although contains multiple divided land in the area of public view as in the example of the map ♯2 displaying only the outer peripheral regions to you ease of explanation in the following. In addition, this area | region is a shape by which the outer periphery was formed with the straight line as FIG. 3, FIG. 4 shows, and this straight line has an intersection (vertex).

As shown in FIG. 4, feature points on map # 1 and map # 2 are designated in advance by the operator. In the example of FIG. 4, six feature points are designated on map # 1, and four feature points are designated on map # 2. As this feature point, as shown in FIG. 4, the intersection of the outer straight lines of the map # 1 and the map # 2 is designated. FIG. 5 shows the concept of alignment. As shown in FIG. 5, the map # 1 (solid line) and the map # 2 (dashed line) are overlapped by overlapping the feature points. At this time, the map # 1 and the map # 2 have different accuracy, so they do not always match. Conventionally, when the map # 1 and the map # 2 do not match, as a result of trial and error based on abundant experience of operators, they are matched, but with the assistance of the map alignment device 4 of this embodiment Even an operator with little experience can make map alignment possible.

  In addition, since the final purpose of the said operation | work is to enable it to search a public map from a housing map without an error, it should just confirm the correspondence between a housing map and a public map without an error. Therefore, it is not the final goal to make the figure shape on the house map and the figure shape on the public map completely coincide with each other with high accuracy, and the map of the public map only has to overlap the house map.

  Here, a process in which the map alignment device 4 according to the present embodiment and the operator cooperate to perform the map overlaying operation will be described. FIG. 6 shows a flowchart of the map overlay process. A map # 1 as a background is called from the housing map DB 1 and displayed on the screen of the display unit 3 (S1). Subsequently, a map # 2 to be overlaid on the map # 1 is called from the public map DB2 and displayed side by side on the map # 1 on the screen of the display unit 3 (S2). On the map # 2, several feature points that are likely to be corresponding points are input (S3). Similarly, several feature points that are likely to be corresponding points are input on the map # 1 (S4). If the size of the figure to be overlaid differs due to unknown scale or incorrect scale, etc., it should be expanded / contracted when it should be expanded / contracted, otherwise it is specified that there is no expansion / contraction (S5). When the figure to be overlapped is rotated and overlapped due to unknown orientation or the like, the rotation range is unlimited, and when the rotation is allowed even if the orientation is known, an allowable rotation range (angle) is designated (S6). Up to this point, the process is performed by an operator.

  Next, the map alignment device 4 automatically calculates and displays the overlay method candidate (s) (S7). The registration method candidates automatically calculated by the map alignment device 4 are sorted and displayed in ascending order of registration error so that the optimum solution can be easily selected. Usually, the candidate with the smallest registration error calculated by the map alignment device 4 is the optimal solution.

  7 and 8 illustrate a specific display method. The example of FIG. 7 is an example in which graphics are displayed in duplicate by performing color coding in ascending order of error. In FIG. 7, instead of color coding, a broken line (candidate # 1), a one-dot chain line (candidate # 2), and a two-dot chain line (candidate # 3) are shown. The example of FIG. 8 is not a graphic display, but is an example in which text is displayed as a list in ascending order of error, and the graphic is displayed when the operator designates an arbitrary order. In addition, various display methods conventionally used can be employed.

  If the optimum solution is found (S8), the process ends. However, if the optimum solution is not found (S9), it may be because the designation of the feature point in step S3 or S4 is not appropriate. Since there are many, the operator manually inputs a set of corresponding points (S10). The map alignment device 4 automatically performs registration so that the registration error is minimized with respect to the set of corresponding points manually input (S11). Specifically, the overlapping is performed by the least square method so that the sum of squares of the distances between corresponding points is minimized. Further, at this time, it is assumed that the operator can set whether or not to perform expansion / contraction and rotation of the figure.

The above is the entire process of the overlapping operation. Below, the process which the alignment apparatus 4 of this invention performs automatically is demonstrated. FIG. 2 shows a block configuration of the map alignment device 4. Designated operator two or more points as the feature points match each map ♯1 and ♯2 Then, each the same number r (2 ≦ r ≦ (n , smaller m)) from the specified feature points feature points For each of the _n C _r combinations of _r feature points extracted from the map having _n feature points, the _r feature points extracted from the map having m feature points. _{r in} each of the _m C _r combinations. Corresponding point generation unit 10 that generates corresponding points that are sets of feature points having a corresponding relationship of _n C _r × _m P _r sets obtained by making each permutation correspond to each other, and corresponding point generation unit 10 A corresponding point extracting unit 11 that extracts the corresponding points satisfying a predetermined condition from the plurality of generated corresponding points, and between the corresponding points of the maps # 1 and # 2 extracted by the corresponding point extracting unit 11 And the corresponding points of the regions of the maps # 1 and # 2 superimposed by the overlapping unit 12 so that the maps # 1 and # 2 are superimposed so that the sum of squares of the distances of the two is minimized. An error determination unit 13 for obtaining an average value of the distances between them, and a candidate sorting unit 14 for displaying a plurality of overlapping patterns from a smaller average value of the distances obtained by the error determination unit 13 to a predetermined rank. Prepare.

  In addition, the predetermined condition focuses on one set of corresponding points of the plurality of corresponding points generated by the corresponding point generation unit 10, and the same area as the corresponding corresponding point on the maps # 1 and # 2 The lengths of the line segments that connect to other feature points are obtained, and the line segments obtained in the maps # 1 and # 2 are made to correspond to each other on a one-to-one basis, and the ratio of the lengths is determined for all the line segments. This is a condition that excludes the pair of corresponding points that are obtained from the correspondence between each other and the standard deviation of the ratio obtained from the correspondence between all the line segments deviates from a predetermined range. Furthermore, when the average value of the angles formed by the line segments corresponding to one-to-one deviates from a predetermined angle, a condition for excluding the line segments in advance prior to obtaining the length ratio is set. Can be included. The predetermined condition can be specified as an extraction condition by the operator, and is stored in the extraction condition setting unit 15.

The superposition unit 12 can perform affine transformation, x and y are coordinates before transformation, X and Y are coordinates after transformation, s is an expansion / contraction ratio, θ is a rotation angle, c, When d is a parameter related to movement,
X = a * x−b * y + c
Y = b * x + a * y + d
a = s * cos θ
b = s * sin θ
The coordinates are transformed by using and superimposed. Furthermore, the overlapping unit 12 can limit expansion / contraction or rotation in accordance with an operation input when overlapping. This superposition condition can be designated by the operator and is stored in the superposition condition setting unit 16.

Next, the operation of the corresponding point generation unit 10 will be described. FIGS. 9 and 11 show examples in which three feature points are designated on maps # 1 and # 2, respectively. As shown in FIGS. 9 and 11, three feature points are extracted from maps # 1 and # 2, respectively, the total number of feature points in map # 1 is six, and the total number of feature points in map # 2 is four. Therefore, for each of ₆ C ₃ combinations of ₃ feature points extracted from map # 1 having ₆ feature points, ₃ extracted from map # 2 having 4 feature points. R in each of ₄ C ₃ combinations of feature points! Corresponding points are generated which are ₆ C ₃ × ₄ P ₃ sets of feature points having a corresponding relationship obtained by associating the permutations with each other.

Specific examples of corresponding point generation will be described with reference to FIGS. 10-13, for each of the ₆ C ₃ combinations of ₃ feature points extracted from map # 1, ₄ C ₃ combinations of ₃ feature points extracted from map # 2 are shown. R in each! An example in which the permutations correspond to each other is shown.

  As shown in FIG. 10, a combination of feature points “2”, “4”, “6” is extracted from map # 1, whereas feature points “1”, “2”, Permutations of “4” are extracted and associated with each other. Similarly, as shown in FIG. 12, a combination of feature points “1”, “3”, “5” is extracted from map # 1, whereas feature points “1”, “5” are extracted from map # 2. Permutations of “3” and “4” are extracted and associated with each other.

As a result of extracting the feature points in this way, as shown in FIG. 13, the feature points (“1”, “2”, “3”) to (“4”, “5”, “ 6 ”) ₆ C ₃ = 20 combinations are extracted. On the other hand, feature points (“ 1 ”,“ 2 ”,“ 4 ”), (“ 1 ”,“ 3 ”, “4”), (“2”, “3”, “4”), (“2”, “3”, “1”) ₃ for each of ₄ C ₃ = 4 combinations! = 6 permutations are extracted. As a result, ₄ P ₃ = 24 permutations are extracted and associated with each other. FIG. 14 shows an example in which the correspondence is generalized.

  Next, the operation of the corresponding point extraction unit 11 will be described. As described above, the corresponding point generation unit 10 generates corresponding points in such a way that all feature points are hit. The corresponding point extraction unit 11 extracts appropriate corresponding points from the corresponding points. 15 to 17 show a procedure for extracting corresponding points in the corresponding point extracting unit 11. In the example of FIG. 15, the feature point “2” of the map # 1 and the feature point “1” of the map # 2 are selected as a set of corresponding points. In the map # 1, line segments having lengths g1, g2, and g3 are generated from the feature point “2” to the other feature points “4”, “5”, and “6”. In map # 2, line segments having lengths h1, h2, and h3 are generated from feature point “1” to other feature points “2”, “3”, and “4”. Subsequently, f1 = h1 / g1, f2 = h2 / g2, and f3 = h3 / g3 are calculated. Similarly, as shown in FIG. 16, f1, f2, and f3 are calculated for each pair of corresponding points.

FIG. 17 shows standard deviations of f1, f2, and f3. As shown in FIGS. 17 (a) and 17 (b), the standard deviation is small if it is an appropriate corresponding point. That is, since map # 1 and map # 2 are similar in most cases, if a pair of corresponding points is appropriate, the distance from the corresponding point to other feature points is also determined in maps # 1 and # 2. Cases with very different values are rare. Therefore, it can be determined that a corresponding point having a large standard deviation of f1, f2, and f3 is not an appropriate corresponding point. FIG. 16 is an example of inappropriate corresponding points. In order to prove this, calculation was performed by substituting actually measured numerical values for h1, h2, h3, g1, g2, and g3. Result is,
In FIG. 15, f1 = 1.3, f2 = 1.3, f3 = 1.5 (average value is 1.3)
In FIG. 16, f1 = 4.2, f2 = 1.7, f3 = 1.25 (the average value is 2.4).
Met. In the example of FIG. 16, the variation in the values of f1, f2, and f3 is clearly larger than that of the example of FIG.

  Thus, when the set of corresponding points is appropriate, the average value of the above f1 to f3 is usually close to 1, but when the size of the overlapping portion is different due to an unknown scale or an incorrect scale, It becomes the ratio of the size. For example, if the size is twice different, the average value is in the vicinity of 2 or 0.5, and the standard deviation increases or decreases depending on the average value. A value obtained by multiplying the threshold value σ when the average value is 1 by the average value of fi (i is a natural number) is adopted as the threshold value.

Threshold value of standard deviation = average value of threshold value σ × fi when average value is 1 (Expression 1)
The threshold σ when the average value of fi is 1 is, for example, 0.1 (about 68% of fi falls within the range of 0.9 to 1.1, and about 95% falls within the range of 0.8 to 1.2. Set to ON. This setting can be changed, but once set, it does not need to be changed.

  When trying to superimpose shapes that are close to regular polygons such as regular triangles and squares, the standard deviation may be small even for combinations with incorrect orientations. It is necessary to extract combinations with small deviations.

  Even when the scale is unknown or the direction is unknown, if the superposition is close to a similar shape, there is a set of corresponding points with small standard deviation of fi, and there is a correct set among them. On the other hand, when there is no pair with a small standard deviation, the operator tries to superimpose a pair that is not similar but does not resemble, so in that case, the operator manually inputs the corresponding pair. Need arises.

  Furthermore, prior to the calculation of the above f1 to f3, the corresponding point extraction unit 11 is inappropriate using the angles formed by the line segments of the lengths h1 to h3 and g1 to g3 shown in FIGS. 15 and 16. Corresponding points can be excluded in advance. FIG. 18 shows a procedure for excluding inappropriate corresponding points in advance using the angles formed by the line segments having the lengths h1 to h3 and g1 to g3. As shown in FIG. 18, when the average value of the angles formed by the line segments having the lengths g1 to g3 in the map # 1 and the line segments having the lengths h1 to h3 in the map # 2 is calculated, the corresponding points are appropriate. If so, the average value should fall within a small value range (for example, ± 15 °). By using this fact and excluding inappropriate corresponding points in advance, the calculation amount can be reduced by calculating f1 to f3. However, in this case, it is a condition that the directions of the maps # 1 and # 2 are matched to some extent in advance. Therefore, the range of the average value can be arbitrarily set by the operator.

Next, the operation of the overlapping unit 12 will be described. FIG. 19 shows the concept of superposition. As shown in FIG. 19, in order to superimpose map # 2 on map # 1, map # 2 is subjected to affine transformation. The affine transformation formula is such that x and y are coordinates before conversion and X and Y are coordinates after conversion.
X = a * x−b * y + c (Formula 2)
Y = b * x + a * y + d
Given in. The affine transformation is a transformation that performs movement, rotation, and expansion / contraction at the same time. When s is an expansion / contraction ratio and θ is a rotation angle,
a = s * cos θ
b = s * sin θ
It is. C and d are parameters relating to movement. When there is no expansion / contraction, s = 1, and when there is no rotation, θ = 0. The presence / absence of expansion / contraction or rotation is set as an overlapping condition by the operator in the overlapping condition setting unit 16.

  In order to perform the affine transformation, it is necessary to determine a, b, c, and d in Expression 2, which are the positions after conversion of r corresponding points on the map # 2, as shown in FIG. Then, a, b, c, and d that minimize the sum of squares of the distances between those corresponding points on the map # 1 are obtained by the least square method.

  Since the azimuth and scale are known and the map to be overlaid may be stretched or not rotated, it may be possible to specify the presence or absence of the stretch and rotation. If you do not want to extend or rotate, as described in the above-mentioned parameters of the affine transformation equation, if there is no expansion / contraction, the expansion / contraction rate s = 1, and if there is no rotation, the rotation angle θ = 0. Find the affine transformation parameters.

  When the affine transformation of Expression 2 is applied to the coordinate values of all points on the map # 2 using the obtained affine transformation parameters a, b, c, d, the map # 2 becomes the map # 2, as shown in FIG. Overlapping with 1. Then, the error determination unit 13 obtains an average value of the distance between the points of the corresponding points after conversion, and sets it as a superposition error e.

  Finally, in order for the operator to select an optimum overlay pattern and to overlay the maps, it is necessary to find the optimum one from the above-described overlay pattern candidates with a small overlay error.

  In order to help this, the candidate sorting unit 14 obtains several overlay patterns including a set of corresponding points, an affine transformation parameter, and an overlay error e in ascending order of error, and stores them in the storage area. Specifically, a storage area is prepared for memorizing i overlay patterns, and when the storage area is filled with i overlay patterns, the maximum overlay error in the storage area is started. When a new overlay pattern with a smaller error comes, the overlay pattern with the largest error is discarded, and the new overlay pattern is stored in the storage area instead. Find the overlay pattern.

  As described with reference to FIG. 7 or FIG. 8, the candidates for the superposition pattern having a small superposition error are sorted in the order of the small superposition error and displayed on the screen of the display unit 3. Present.

  FIG. 20 is a flowchart showing the overlapping pattern candidate automatic calculation process described above. The corresponding point generation unit 10 first extracts two (r = 2) corresponding points from the feature points set on the map # 1 by the operator (S20 → S21), and performs the first combination process (S20 → S21). S22). Subsequently, two (r = 2) corresponding points are first extracted from the corresponding points set on the map # 2 by the operator (S20 → S23), and a permutation combination of the extracted r corresponding point candidates is performed. By searching for and generating a pair of corresponding points (S24), the first permutation combination process is performed (S25).

  Subsequently, when the orientation is not unknown, the corresponding point extraction unit 11 determines whether or not the average value of the angle difference between the feature points is within a predetermined value (S26). Exclude (S38). For the corresponding point candidates not excluded in step S26, the standard deviation of the ratio of the distances between the feature points is obtained (S27). It is determined whether or not the standard deviation is within the threshold (S28), and those exceeding the threshold are excluded from the corresponding point candidates (S38).

  The overlapping unit 12 obtains affine transformation parameters a, b, c, and d for the corresponding point candidates not excluded in step S28 (S29). Here, s = 1 when there is no expansion / contraction, and θ = 0 when there is no rotation.

  The error determination unit 13 calculates a registration error e when the registration by the registration unit 12 is performed (S30). As a result, with respect to those where the overlay error e is from the smallest to the i-th (S31), the values of the affine transformation parameters a, b, c, d and the error e are recorded (S32).

  If all permutation combinations for extracting r corresponding point candidates from map # 2 have not been processed (S33), the next permutation combination is processed (S39). If not all combinations for extracting r corresponding point candidates from map # 1 are processed (S34), the next combination is processed (S40).

  Processing has been completed for all permutation combinations that extract r corresponding point candidates from map # 2 (S33), and processing has been completed for all combinations that extract r corresponding point candidates from map # 1. If (S34), r is replaced with 2 to 3 (S35), and Steps S21 to S32 are executed again. Steps S21 to S32 are repeatedly executed up to the maximum value of r (S36).

  The candidate sorting unit 14 sorts the combinations of corresponding points with small overlay errors in the record up to the i-th order in ascending order of errors and displays them on the screen of the display unit 3 (S37). The operator selects the optimum solution while confirming this display (S9 in FIG. 6). When the optimum solution is not found, the operator manually sets a pair of corresponding points (S10 in FIG. 6), and the map alignment device 4 automatically performs overlay so that the overlay error is minimized (FIG. 6). 6 (S11), this process is performed using the superposition function of the superposition unit 12.

  Further, the present embodiment can be implemented as a program that, when installed in a general-purpose information processing apparatus, causes the general-purpose information processing apparatus to realize functions corresponding to the map alignment apparatus 4 of the present embodiment. This program is recorded on a recording medium and installed in a general-purpose information processing apparatus, or installed in a general-purpose information processing apparatus via a communication line, so that the corresponding point generation unit 10, Corresponding functions can be realized in the corresponding point extraction unit 11, the superposition unit 12, the error determination unit 13, the candidate sort unit 14, the extraction condition setting unit 15, and the superposition condition setting unit 16, respectively. The general-purpose information processing apparatus is, for example, a general-purpose personal computer.

  According to the present invention, for example, it is possible to automate the work of fitting a public map into a house map and improve the work efficiency, so that it is possible to meet the strong demand that it is convenient if a public map can be searched from the house map. In addition, it is possible to contribute to the construction of a system that manages the housing map and the public map in a completely corresponding manner.

1 is an overall configuration diagram of a map alignment device of the present embodiment. The block block diagram of the map alignment apparatus of a present Example. The figure which shows the example of a display of the housing map and land-register map of a display part. The figure which shows a feature point. The figure which shows the concept of alignment. The flowchart of a map superimposition process. The figure which shows the example which performed the color classification in the order with small error, and displayed the figure redundantly. The figure which shows the example which displayed the text as a list in order with a small error. The figure which shows the example in which three feature points are each designated in map # 1 and # 2. The figure which shows the example which feature point "2", "4", "6" was extracted from map # 1, and feature point "1", "2", "4" was extracted from map # 2. The figure which shows the example in which three feature points are each designated in map # 1 and # 2. The figure which shows the example which feature point "1", "3", "4" was extracted from map # 1, and feature point "1", "3", "4" was extracted from map # 2. For each of 6C3 combinations of 3 feature points extracted from map # 1, r! In each of 4C3 combinations of 3 feature points extracted from map # 2! The figure which shows the example which makes a permutation correspond to each other. The figure which shows the example which represented the correspondence relationship in general. The figure which shows the example (appropriate corresponding point) in which the feature point "2" of map # 1 and the feature point "1" of map # 2 are selected as a set of corresponding points. The figure which shows the example in which the feature point "1" of map # 1 and the feature point "1" of map # 2 are selected as a set of corresponding points (an inappropriate corresponding point). The figure which shows the standard deviation of f1, f2, and f3. The figure which shows the procedure which excludes an inappropriate corresponding point beforehand using the angle | corner which the line segments of length h1-h3 and g1-g3 comprise. The figure which shows the concept of superimposition. The flowchart of a superposition pattern candidate automatic calculation process.

1 housing map database (DB)
2 Public database (DB)
3 Display unit 4 Map alignment device 5 Housing map database for public map search (DB)
6 Keyboard 7 Mouse 10 Corresponding Point Generation Unit 11 Corresponding Point Extraction Unit 12 Overlaying Unit 13 Error Determination Unit 14 Candidate Sorting Unit 15 Extraction Condition Setting Unit 16 Overlapping Condition Setting Unit

Claims (10)

A map alignment device for superimposing a predetermined area of a second map on an area on a first map,
Display means for displaying an area on the first map and a predetermined area on the second map;
The operator displays the two areas on the display means, designates two or more points likely to correspond to the two areas as feature points, and the same number r (2 ≦ r ≦ () from the designated feature points. n, the smaller of m))) input means for further selecting and extracting points;
For each of _n C _r combinations of _r feature points selected and extracted from one region having _n feature points, _r features selected and extracted from the other region having m feature points _{R in} each of the _m C _r combinations of points! Corresponding point generation means for generating corresponding points that are sets of feature points having _n C _r × _m P _r pairs that correspond to each other obtained by making (r factorial) number of permutations correspond to each other;
A corresponding point extracting unit that extracts the corresponding point satisfying a predetermined condition from a plurality of the corresponding points of the two regions generated by the corresponding point generating unit;
Superimposing means for superimposing the two regions so that the sum of squares of the distance between corresponding points, which is a set of feature points having a correspondence relationship between the two regions extracted by the corresponding point extracting unit, is minimized;
Error determining means for obtaining an average value of distances between corresponding points, which is a set of feature points having a correspondence relationship between the two regions superimposed by the overlapping means;
Means for displaying a plurality of overlapping patterns from a smaller average value of the distances determined by the error determination means to a predetermined order,
The predetermined condition focuses on any one set of corresponding points of the plurality of corresponding points generated by the corresponding point generation means, and in each of the regions, other points in the same region as the corresponding corresponding point The lengths of the line segments that connect the feature points are obtained, and the line segments obtained in each of the areas are associated with each other in a one-to-one correspondence between the two areas, and the ratio of the lengths of all the line segments is determined. A map alignment apparatus characterized in that a pair of corresponding points whose standard deviation of ratios obtained for correspondence between all line segments deviates from a predetermined range is a condition for excluding this correspondence . In the predetermined condition, when the average value of the angles formed by the line segments corresponding to each other deviates from the predetermined angle, the line segment is determined prior to obtaining the length ratio. The map alignment apparatus according to claim 1, which includes conditions to be excluded in advance. The superimposing means includes means for performing affine transformation on one of the two regions,
When x and y are coordinates before conversion, X and Y are coordinates after conversion, s is an expansion / contraction ratio, θ is a rotation angle, and c and d are parameters related to movement, respectively,
X = a * x−b * y + c
Y = b * x + a * y + d
a = s * cos θ
b = s * sin θ
The map alignment apparatus according to claim 1, further comprising means for performing coordinate conversion and superimposing with each other.   4. The map alignment apparatus according to claim 1 or 3, wherein said superimposing means includes means for restricting expansion / contraction or rotation in accordance with an operation input upon superimposing. A map alignment method performed by a map alignment apparatus that superimposes a predetermined area of a second map on an area on a first map,
The display means displays the area on the first map and the area on the second map,
The input means designates two or more points that are likely to correspond to the two areas displayed by the display means as feature points, and the same number r (2 ≦ r ≦ n, m) from the designated feature points. Whichever is smaller)) select feature points,
Corresponding point generation means selects and extracts from the other region having m feature points for each of _n C _r combinations of _r feature points selectively extracted from one region having _n feature points. R in each of the _m C _r combinations of the _r feature points generated. Generating a corresponding point that is a set of feature points having _n C _r × _m P _r pairs that correspond to each other obtained by causing each permutation to correspond to each other;
Corresponding point extraction means extracts the corresponding points that satisfy a predetermined condition indicating that the shapes of the two sections can be superimposed from among the plurality of corresponding points of the generated two sections ,
Superposing means, superposing two of said regions as the square sum becomes the minimum distance between corresponding points of the set of feature points having correspondence between the extracted two regions,
The error determination means obtains an average value of distances between corresponding points that are a set of feature points having a correspondence relationship between the two overlapping regions .
The display means displays a plurality of overlapping patterns from a smaller average value of the distances to a predetermined order ,
The predetermined condition focuses on one set of corresponding points of the plurality of generated corresponding points, and in each of the regions, a line connecting the corresponding corresponding point with another feature point in the same region The lengths of the minutes are obtained, the line segments obtained in each of the regions are made to correspond one-to-one between the two regions, and the ratio of the lengths is obtained for the correspondence between all the line segments, A map alignment method, characterized in that a set of corresponding points in which a standard deviation of a ratio obtained for correspondence between the line segments deviates from a predetermined range is a condition for excluding this . The predetermined condition is that, when an average value of angles formed by the line segments corresponding to one-to-one deviates from a predetermined angle, the line segments are preliminarily determined before the ratio of the lengths is obtained. The map alignment method according to claim 5 , including conditions to be excluded. The superimposing means performs affine transformation on one of the two sections, x and y are coordinates before transformation, X and Y are coordinates after transformation, s is an expansion / contraction ratio, θ is a rotation angle, c, When d is a parameter related to movement,
X = a * x−b * y + c
Y = b * x + a * y + d
a = s * cos θ
b = s * sin θ
The map alignment method according to claim 5, wherein the coordinates are converted and overlapped with each other.   The map alignment method according to claim 5 or 7, wherein expansion and contraction or rotation is restricted in accordance with an operation input when overlaying. The program which implement | achieves the function corresponding to the map alignment apparatus in any one of Claim 1 thru | or 4 in the general-purpose information processing apparatus by installing in the general-purpose information processing apparatus. A recording medium readable by the information processing apparatus on which the program according to claim 9 is recorded.

Images