JP2010272093A - Image connecting method, device and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a connecting method for further naturally connecting two ortho images. <P>SOLUTION: An approximation degree calculation program (12a) calculates an approximation degree within a search range of an overlap part of two ortho image data (30) stored in a hard disk device (24). An approximation degree binarization program (12b) binarizes the approximation degree within the search range with a maximum threshold, a labeling processing program (12c) performs labeling, and a thinning program (12d) performs thinning. An unspecified search line search program (12e) or a start point-specified connecting line search program (12f) sets, while searching a close label in a direction along a connecting line determination reference line, a connecting line within a searched label. With respect to a jump section, labeling processing and thinning are performed with a reduced threshold of approximate value, and it is repeated to set, while searching the close label in the direction along the connecting line determination reference line, the connecting line within the search label. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a method, apparatus, and program for joining two images having overlapping portions.

  The ground can be precisely photographed from an airplane or satellite. These aerial photographs or satellite images are projected on a certain horizontal plane after distortion caused by the shooting direction is eliminated. An image obtained by this projection, that is, a single image showing the surface of a vast area can be obtained by combining or joining the projection images. Individual projection images may be converted into orthographic projection images (ortho images) by a plotter in advance. Patent Document 1 describes a method of joining raster image data such as a map or aerial photograph data that displays adjacent regions partially overlapping at the partially overlapping portions.

  For the purpose of such composition or joining, usually, in photographing from an airplane or a satellite, the subject is photographed so as to overlap more than a certain amount. FIG. 15 shows an example of duplication of ortho images from which photographs or the like taken partially overlapping in this way are obtained. In general, each ortho image partially overlaps one or more other ortho images more randomly.

  Even in an ortho image generated from a photograph taken from the vertically upward direction, a slight fall occurs at the edge of a building or an elevated part. Of course, when shooting from an oblique direction, most buildings fall down. Therefore, at the time of composition, a joining line is set at an inconspicuous place, and two ortho images are joined.

  For example, as shown in FIG. 16, there may be a case where an ortho image in which the same building collapses in the opposite direction to each other in the image overlapping portion may be synthesized. FIG. 16A shows a left ortho image, and FIG. 16B shows a right ortho image. In this case, when the joint line is set so as to cut the building, the shape of the building is distorted on the composite image as shown in FIG. FIG. 17A shows the left ortho image shown in FIG. 16A, FIG. 17B shows the right ortho image shown in FIG. 16B, and FIG. 17C shows the composite image.

JP 2001-351113 A

  The conventional method does not consider the collapse of the building in the aerial photograph. As a result, as shown in FIG. 17, when a joining line is set on a building that has fallen down, joining displacement occurs at the building location, and the shape of the building is impaired, which is not preferable.

  An object of the present invention is to provide an image joining method, apparatus, and program for joining two images in an inconspicuous place efficiently by automatic processing by a computer.

  An image joining method according to the present invention is a method of joining two ortho image data, which are obtained by photographing from mutually different directions, at least partially overlapping by a computer, and the computing means of the computer includes the two An initial search range is set in the overlapping portion of the ortho image data, and an initial search range setting step for storing the initial search range in the storage unit of the computer, and the calculation unit calculates the degree of approximation of the two ortho image data in the initial search range Then, the approximation degree calculation step stored in the storage means, and the calculation means extract pixels whose approximation degree obtained in the approximation degree calculation step is in an approximation value range that is equal to or greater than a predetermined approximation value, Grouping for each pixel to be extracted, and extracting the junction line search target pixel, and the calculation means is connected to the junction line search target pixel A junction line search step of sequentially searching for a label close to a predetermined direction from a predetermined label of the junction line search target pixel, and setting a junction line to the predetermined label and each searched label; Then, the preprocessing step is executed with a smaller approximation value range for the jump section of the junction line, and the junction line search target pixel extracted in this manner is extracted from the predetermined label of the junction line search target pixel. A step of sequentially selecting a label close to a predetermined direction and setting a joining line to each selected label, and the computing means joining the two ortho image data according to the joining line, And a bonding step to be generated.

  An image joining device according to the present invention is an image joining device that joins two ortho image data obtained by photographing from different directions and overlaps at least a part thereof. An initial search range setting means for setting an initial search range; an approximation degree calculating means for calculating the degree of approximation of the two ortho image data in the initial search range; and storing in the storage means; A preprocessing means for reading out the degree of approximation, extracting pixels whose read degree of approximation is in a specified degree of approximation value range, grouping the pixels for each adjacent pixel, and extracting a junction line search target pixel; and Among the labels of the connection line search target pixels generated by the preprocessing unit, the labels are sequentially searched from the predetermined label in the predetermined direction, and the predetermined label and A joining line search means for setting a joining line for each searched label, and a control means for controlling the preprocessing means and the joining line search means. The initial search range is designated as the designated search range at the first time. Then, the upper limit value range is designated as the designated approximation value range, the preprocessing means and the joint line search means are executed, and after the second time, the jump section of the set joint line is designated as the designated search range. And the control unit that executes the preprocessing unit and the joint line search unit, and the two ortho image data are joined in accordance with the joint line, with the approximation value range reduced stepwise as the designated approximation value range. And combining means for generating combined image data.

  An image joining program according to the present invention is an image joining program for joining two ortho image data, which are obtained by photographing from different directions to each other, at least partially overlapping, to the computer. An initial search range is set in the overlapping portion of the ortho image data and stored in the storage means of the computer, and an approximation degree of the two ortho image data is calculated in the initial search range, and the storage means The approximation degree calculation function to be stored in the image, and the approximation degree obtained by the approximation degree calculation function are extracted pixels having an approximation value range that is equal to or greater than a predetermined approximation value, and the adjacent pixels are grouped and joined together. A pre-processing function for extracting a line search target pixel and a predetermined label of the joint line search target pixel according to the joint line search target pixel. In order to search for a label close to a predetermined direction sequentially, a joint line search function for setting a joint line to the predetermined label and each searched label, and a smaller approximation value range with respect to the jump section of the joint line The pre-processing function is executed, and for the junction line search target pixel extracted by this, a label close to a predetermined direction is sequentially selected from the predetermined label of the junction line search target pixel, and each selected label is selected. A repeat function for setting a joining line and a joining function for joining the two ortho image data according to the joining line to generate composite image data are realized.

  According to the present invention, a joining line can be set at a portion having a high degree of approximation between two ortho images, and both images can be joined more naturally.

1 shows a schematic block diagram of an embodiment of the present invention. It is an explanatory example of the overlapping range of three images. It is a part of operation | movement flowchart of the image composition by a present Example. It is a part of operation | movement flowchart of the image composition by a present Example. It is an example of pixel distribution of binarized approximation data. The detailed operation | movement flowchart of an entrance designation | designated junction line search program is shown. It is explanatory drawing of the process process by the entrance designation | designated junction line search program with respect to the binarization approximation data shown in FIG. The detailed operation | movement flowchart of a non-designated joint line search program is shown. It is explanatory drawing of the process process by the non-specified joining line search program with respect to the binarization approximation data shown in FIG. It is an explanatory example of the jump section with respect to the search range. It is an explanatory example of the search range set with respect to the jump area shown in FIG. 12 is an explanatory example of a search range set in the jump section shown in FIG. 11 and a jump section in the search range. 12 is an explanatory example of a search range set in the jump section shown in FIG. 11 and a jump section in the search range. FIG. 17 shows an example of a result obtained by joining two ortho images shown in FIG. 16 according to the present embodiment. It is a schematic diagram which shows the mode of duplication of an ortho image. It is an example of the ortho image in which the same building falls down in the mutually opposite direction in the image overlap part. It is an example of the synthetic | combination result of the image shown in FIG.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 shows a schematic block diagram of an embodiment of the present invention. The image joining apparatus 10 of this embodiment includes a CPU 12, a ROM 14, a RAM 16, a mouse 18, a keyboard 20, a display device 22, and a hard disk device (HDD) 24, which are connected to each other via a bus 26. The ROM 14 or the HDD 24 stores program software that operates on the CPU 12 in order to realize the functions of the image bonding apparatus 10. The CPU 12 executes a joining process described below using the RAM 16 and the HDD 24 in accordance with a user instruction from the mouse 18 and the keyboard 20 as a calculation unit of the image joining apparatus 10. For this purpose, the CPU 12 has functions as function means 12a to 12g described below. The image joining apparatus 10 can be realized by a general-purpose computer or a dedicated computer.

  The HDD 24 stores the ortho image data 30 to be joined in a bitmap format. As described above, each ortho image data 30 stored in the HDD 24 generally overlaps one or more other ortho images for joining. The HDD 24 also stores ground elevation (DTM) data 32 and surface elevation (DSM) data 34 for the target area. The ortho image data 30 stored in the HDD 24 is standardized so as to have pixel values on points on the same ground coordinate system, and the DTM data 32 and the DSM data 34 are also each pixel of the ortho image data 30. It consists of elevation values at the same point. This is because the processing described later is facilitated. Of course, even if the pixels of the ortho image data 30 and the elevation values of the DTM data 32 and the DSM data 34 are not normalized so as to indicate the same point, the values on the same point are calculated by interpolation when necessary. Obviously, this may be done.

  Basically, the joint line is determined for the heavy belly portions of the two orthoimages. However, in the orthoimages obtained from continuously taken aerial photographs, as shown in FIG. Orthoimages A, B, and C may partially overlap. In FIG. 2, an area AB indicates an overlapping part between the images A and B, an area AC indicates an overlapping part between the images A and C, an area BC indicates an overlapping part between the images B and C, and an area ABC indicates the image A, The overlapping part of B and C is shown. In this way, when there are portions where three or more images overlap, they may be joined in stages. That is, any two ortho images are joined, and the remaining one ortho image partially overlapping is joined to the composite image obtained by the joining. This is repeated to finally generate a large composite image. As a method of selecting two images to be joined first, for example, in a portion where the most images overlap (area ABC in FIG. 2), the degree of approximation is calculated by a combination of two arbitrary images, and the two closest approximations are calculated. The image is the target of the first joining process. Accordingly, here, the operation of determining the joining line of two images and joining them will be described. However, the present invention can also be applied to the case of synthesizing three or more images.

  3 and 4 generally show an operation flowchart of image composition according to the present embodiment.

  In accordance with the user's specification, the CPU 12 selects, from the ortho image data 30 stored in the HDD 24, ortho image data of two images for which a joint line is to be set as a target image (S1).

  CPU12 detects a building part from the difference of DSM data 34 and DTM data 32 about two object images selected at Step S1, and removes data which shows a building from ortho image data of two object images (S2). The pixel value corresponding to the building may be replaced with a constant value such as 0 or an invalid value from the two ortho image data itself. You must exclude the calculation result of the degree of approximation. In order to avoid this, for example, a flag table (building flag table 36) corresponding to each pixel of two ortho image data is prepared, and a building flag is set for a pixel in which a building exists. Then, when it is desired to exclude the building portion, the building flag table 36 is referred to and the pixel on which the building flag is set is excluded from the processing target.

  Next, the CPU 12 detects an overlapping portion of the two ortho images selected in step S1, and sets a joint line search range (initial search range) (S3). In that case, you may exclude a building part from a search range with reference to the building flag table 36. FIG.

  The approximation calculation program 12a of the CPU 12 compares the two target image data within the search range and calculates the approximation (S4). The calculated approximation data 38 is held in the HDD 24 (or RAM 16) until the process of determining the joint line within the search range set in step S3 is completed.

  For example, distance (for example, RMS (Root Mean Square) value), distance of green (G) value or absolute difference, or luminance distance or absolute absolute regarding red (R), green (G) and blue (B) A value is calculated, and it is evaluated that the degree of approximation is higher as the distance or the absolute difference value is smaller. In this embodiment, the degree of approximation is evaluated in the range of 0 to 255, and the value is increased as the degree of approximation is increased. For example, a result obtained by subtracting the distance or the absolute difference value from the maximum value (255) is set as the approximation degree. The degree of approximation is distributed within the search range, and the degree of approximation becomes higher as the difference is smaller between the two target image data, and the degree of approximation is 255 where there is no difference.

  A part having a higher degree of approximation can be joined with no unnaturalness when a joining line is used. In this embodiment, a joining line is first assigned to a connected portion of pixels having a high degree of approximation, and a portion where the joining line cannot be assigned (a jump section of the joining line) is an approximation value range in which the degree of approximation is lowered by one step. Assign. This can be realized by binarizing the degree of approximation and lowering the binarization threshold step by step. However, since it is necessary to exclude the pixel that is the target of the joint line search from the search target again, in this embodiment, the pixel is identified by setting a processed flag.

  It should be noted that the same processing can be realized by extracting pixels having a certain degree of approximation value range. For example, in the first joint line search, pixels having an approximation value of 240 or more are used as the target pixels for the joint line search, and in the second time, pixels having an approximation value of 230 to 239 are used as the target pixels for the joint line search. In the second time, pixels having an approximation value of 220 to 229 are set as the target pixels for the joint line search, and the same processing is executed thereafter. As described above, the target pixel of the joint line search is extracted while sequentially decreasing the approximation value range, and the joint line is assigned, so that an appropriate joint line is determined as a whole.

  First, the approximation degree binarization program 12b of the CPU 12 initializes an approximation threshold T for binarizing the approximation degree within the search range with an appropriate maximum value, for example, 240 to 250 (S5). Then, the CPU 12 executes binarization (S6), labeling (S7), and thinning (S8) using the approximation threshold T as preprocessing for determining the joining line. That is, the approximation degree binarization program 12b binarizes the approximation degree data 38 calculated in step S4 with the approximation degree threshold value T (S6), and extracts pixels whose approximation value is equal to or greater than the approximation degree threshold value T. For example, a flag 1 is set for a pixel having an approximation degree equal to or greater than the approximation threshold T, and a flag 0 is set for a pixel having an approximation degree less than the approximation threshold T. The labeling processing program 12c groups the pixels extracted by the binarization program 12b with adjacent or continuous pixels, and assigns a different label number to each group (S7). During the labeling process (S7), the label composed of isolated pixels or a small number of pixels is regarded as noise and removed. The thinning program 12d finds the center line of each label and thins it (S8). Such a thinning computer program itself is known and may be of any alogism. By the pre-processing consisting of steps S6 to S8, the pixel that is the target of the subsequent joint line search process, that is, the joint line search target pixel is extracted. A pixel distribution of binarization approximation obtained by binarization and thinning and its label value are temporarily stored as binarization approximation data 40 in the HDD 24 (or RAM 16). Thinning is useful for simplifying and speeding up the bonding line setting process in the label, but thinning may be omitted if such a load is acceptable.

  FIG. 5 shows a distribution example of the binarized approximation data 40 that has undergone such preprocessing (S6 to S8). Adjacent and continuous pixels form one group (label), and each group is identified by a label number. As can be seen from FIG. 5, the pixels to be searched for joint lines are divided by binarization (S6) and labeling processing (S7), and branching can occur on the same label by thinning (S8).

  In the example shown in FIG. 5, the images to be joined are positioned on the upper side and the lower side, and the joining line can be basically set in the left-right direction of the search range. Therefore, a reference line (joining line determination reference line) for determining a joining line is virtually set at the center of the search range. For example, when neither the start point (entrance point of the search range) nor the end point (exit point of the search range) of the joint line to be set is given as in the first search region, It is set at the center of the overlapping part of the target image. When the entry point and the exit point are given, the joining line determination reference line is a line connecting the entry point and the exit point, and has a predetermined width above and below the joining line determination reference line between the entry point and the exit point. The range is the search range. The entry point and the exit point may be temporary that can be changed or may not be changeable.

  When the entry point of the joining line is not designated for the search range set in step S3 (S9), the undesignated joining line search program 12e executes a joining line search process that does not designate the entry point (S10). ). When the entry point of the joining line is designated (S9), the entry designated joining line search program 12f executes the joining line search process of entry point designation (S11).

  For example, in the example shown in FIG. 2, the joining line can be determined from the images A and B to some extent freely for the region AB, and the joining line can be determined from the images A and C to some extent for the region AC. Can be determined. However, after that, when an attempt is made to determine a joining line for the area ABC or the area ABC, the entry point in the search range is determined from the joining line of the area AB or the area AC. On the contrary, in consideration of the joint line in the region ABC, it may be good to determine the entry point for determining the joint line between the region AB and the region AC. Therefore, in this embodiment, steps S9 to S11 are prepared for the initial search range. Of course, if there is no such possibility, only one of step 10 and step S11 may be used.

  Since the non-designated joint line search program 12e uses the entrance designated joint line search program 12f, the operation of the joint line search (S11) by the entrance designated joint line search program 12f will be described in detail first. The entrance designation joint line search program 12f requires a search range, a search direction, and an entrance point as its arguments. FIG. 6 shows a detailed flowchart of the junction line search process (S11) of entry point designation by the entry designated junction line search program 12f. FIG. 7 is an explanatory diagram of the process in step S11 for the binarized approximation data shown in FIG. In FIG. 7, both the entry point (indicated by a circle) and the exit point (indicated by a x) are located on the joining line determination reference line. The exit point may not be specified. Further, x is added to the pixels corresponding to the start point (S) and end point (E) of the joint line set on the corresponding label in the process of joint line search shown in FIG.

  The entry designated joint line search program 12f refers to the binarized approximation data 40 to determine whether or not the designated entry point is located on the label within the search range (S31), and the designated entry point is located on the label. When doing so (S31), the label is set as a target label to be searched for a joint line, and the designated entry point is set as the start point. In the example shown in FIG. 7, the designated entry point does not belong to any label, but if the designated entry point belongs to any label within the search range (S32), that label is designated as the target label and designated. The entry point is the starting point (S32). Then, on the target label, the pixel that is on the furthest side in the search direction (the right direction in FIG. 7) and is closest to the joint line determination reference line is set as the end point (S33).

  The intra-label joint line setting program 12g of the CPU 12 searches the target label for a pixel connecting the start point and the end point with the shortest distance, that is, the minimum number of pixels, and sets a joint line flag for each of these pixels. A joining line is set (S34). The joint line flag indicates that the pixel is employed as a joint line. Searching for a pixel that connects two points at the shortest distance can be easily achieved by adopting a known algorithm that connects the maze entrance and exit at the shortest distance, or by trying all possibilities. It can be easily realized. It should be noted that a processed flag indicating that the joint line search process has been completed is set for all pixels of the target label, and is thereby excluded from the target of the subsequent joint line search process.

  Then, a non-passed pixel closest to the end point on a different label is searched for in the search direction from the end point (S35). At this time, in order to expand the connection possibility, the pixels immediately above or below are also included in the search range. A non-passed pixel is a pixel of a label that has not yet been subjected to a joint line search, and is a pixel for which no processed flag is set. When a plurality of non-passed pixels are found, the pixel closest to the joint line determination reference line is adopted.

  When the entry point is not located on the label within the search range (S31), that is, in the case of the example shown in FIG. 7, the entry designated joint line search program 12f refers to the binarization approximation data 40. The non-passing pixel closest to the entry point is searched for in the search direction from the entry point (S36). Also in step S36, in order to expand the connection possibility, the pixel immediately above or below is also included in the search range.

  As a quantitative criterion for the proximity in steps S35 and S36, in principle, the square of the Euclidean distance between the pixel whose distance is to be calculated and the reference pixel, that is, the number of pixels in the horizontal direction between the former and latter pixels. For example, the sum of the square of (for example, the number of pixels along the connection line determination reference line) and the square of the number of vertical pixels (for example, the number of pixels along the axis perpendicular to the connection line determination reference line) can be considered. However, since this calculation includes square, it takes time to obtain the final result. Therefore, in this embodiment, the maximum number of horizontal and vertical pixels between the former and latter pixels or the number of pixels through which a straight line connecting the former and latter pixels passes is more simply adopted. Since the closeness determination in steps S35 and S36 does not require a strict comparison, the latter method, which requires only a simple calculation that does not include the square, is convenient.

  As a result of the search in step S35 or step S36, if a non-passed pixel can be found (S37), if the found pixel is on the upper side or the lower side of the search range, further joint line search becomes uncertain. Therefore, when the found pixel is not on the upper side or the lower side of the search range (S38), the junction line is searched for with the next label. That is, the label including the found pixel is set as the attention label, and the found pixel is set as the start point (S39). Then, a joining line is set in the target label (S33, S34), and a label for which a joining line is to be set is searched for (S35).

  If a non-passed pixel cannot be found by the search in step S35 or step S36 (S37), or if the found pixel is on the edge of the search range (S38), a joint line is no longer set in the search range. Since there is no label to be processed, the process is terminated. As will be described later, the interval where the joining line is not connected is reduced as the jump interval, and the approximation threshold T for binarization is lowered and the joining line is searched again.

  The joining line search process (S10) by the unspecified joining line search program 12e will be described. The unspecified junction line search program 12e formally requires a search range as its argument, but since it is executed only for the first time, a search range is set in a global variable and is referred to. Also good. FIG. 8 shows a detailed operation flowchart of the non-specified joint line search program 12e. FIG. 9 is an explanatory diagram of the process in step S10 for the binarized approximation data shown in FIG. X is added to pixels corresponding to the start point (S) and end point (E) of the joint line set on the corresponding label in the process of joint line search in step S10.

  The non-designated joint line search program 12e refers to the binarized approximation data 40, searches for a label with the maximum area (maximum number of pixels) within the specified search range, and sets it as a target label (label to be processed). (S51). This is because the label with the largest area is considered most preferable for setting the joining line. When there are a plurality of labels having the maximum area, the label whose center is closest to the joining line determination reference line is set as the attention label. If there are a plurality of such labels, a label close to the center as viewed in the left-right direction of the search area or an appropriate label is set as a target label.

  In order to facilitate understanding, it is assumed that the joining line determination reference line in the search area is horizontal as shown in FIG. The pixel located at the leftmost end of the target label and closest to the joint line determination reference line is set as the start point of the joint line in the target label (S52). The pixel located at the rightmost end of the target label and closest to the joint line determination reference line is set as the end point of the joint line in the target label (S53). The start point and end point are stored in the search range for the joint line search process for the label on the left side and the label on the right side of the target label (S54).

  On the target label, a pixel that connects the start point and the end point with the shortest distance, that is, the minimum number of pixels, is searched, and a bond line flag is set for these pixels to set a bond line (S55). The same algorithm as in step S34 can be adopted, and as in step S34, all the pixels of the target label are set with a processed flag indicating that they have been subjected to the joint line search process, and thereby the subsequent joint line search is performed. Exclude from processing.

  Next, among the labels present on the left side of the target label determined in step S51, labels for which a joint line is to be set are sequentially searched, and a joint line is set within those labels (S56). Specifically, the unspecified joining line search program 12e sets a constant width centered on the joining line determination reference line from the start point stored in step S34 to the left side of the initial search range as the search range, and stores it in step S34. The starting point is the entry point, the left direction is designated as the search direction, and the entry point designation joint line search program 12f is called. The entry point designating joint line search program 12f refers to the binarized approximation data 40, and follows the flow shown in FIG. 6 from the labels positioned on the left side of the target label (label of the maximum area) in step S51. The labels on which the joint lines are to be set are sequentially determined, and the joint lines are set in the respective labels.

  Similarly, among the labels present on the right side of the target label determined in step S51, labels for which a joint line is to be set are sequentially searched, and a joint line is set within those labels (S57). Specifically, the non-specified joining line search program 12e sets a constant width centered on the joining line determination reference line from the end point stored in step S34 to the right side of the initial search range as the search range, and stores it in step S34. The entry point designation junction line search program 12f is called by setting the end point as the entry point and specifying the right direction as the search direction. The entry point designation joint line search program 12f refers to the binarized approximation data 40, and follows the flow shown in FIG. 6 from the labels positioned on the right side of the target label (the label of the maximum area) in step S51. The labels on which the joint lines are to be set are sequentially determined, and the joint lines are set in the respective labels.

  By the above processing (S51 to S57), the labels that are close to each other suitable for setting the bonding line are selected for the labels having the approximation value equal to or greater than the approximation threshold T, and the bonding line is added to the selected label. Can be set.

  The first label of the unspecified joining line search process (S10) is the label of the maximum area, but the length along the joining line determination reference line (the length when projected onto the joining line determination reference line) is the longest. May be the first attention label.

  After step S10 or S11, it is checked whether or not there is a section (jump section) where the set joint line is discontinuous (S12). This can be found, for example, by tracing the connection of the end points (start point or end point) of the set joint line from right to left or from left to right within the initial search range.

  If there is no jump section (S13), a continuous joining line is designed within the initial search range, so manual editing / modification is executed (S23).

  When there is a jump section (S13), the approximation threshold T is decreased by a predetermined value (Δ) (S14). The first jump section is set as a search range, and an entry point and an exit point of the search range are set (S15).

  When the jump section is short enough not to search for a joint line, for example, when it is about several pixels (S16), a joint line directly connecting the entrance point and the exit point of the search range is defined between the entrance point and the exit point. (S21).

  When the jump section has a certain distance or more (S16), the approximation binarization program 12b of the CPU 12 uses the approximation data 38 calculated in step S4 for the search range set in step S15. Binarization is performed with the threshold value T (S17). The binarized approximation is labeled in the same manner as in step S7 (S18), and thinned by the same processing as in step S8 (S19). The entrance designated joint line search program 12f executes an entry point designated joint line search process (S20). The end point of the joint line obtained as a result of step S20 may not coincide with the exit point, but a jump section is provided between the end point and the exit point.

  After step S21 or S20, the next jump section is searched (S22). If the next jump section is found before reaching the other end of the initial search range (S22), the next jump section is set as the search range (S23), and steps S16 to S21 are repeated.

  If the other end of the initial search range is reached without finding the next jump section (S22), the process returns to step S12. That is, a jump section that may occur in the joint line search in step S20 is searched (S12). If it is found (S13), the approximation threshold T is decreased by Δ, and steps S15 to S21 are performed. Repeat the joint line search process.

  When the jump section finally disappears (S13), the joining line set by the above processing is displayed on the display device 22 for the two images to be synthesized, and the editing or correction by the operator is accepted (S24). The operator manually edits / corrects the joining line as appropriate using the mouse 18 or the like. The CPU 12 synthesizes two ortho images with the joint line thus determined (S25). The obtained composite image data 42 is stored in the HDD 24.

  When there is an ortho image or another composite image to be further combined with the composite image obtained in this way, in FIG. 3, these two images are selected as targets (S1), and steps S2 and after may be repeated. .

  In FIG. 3 and FIG. 4, the joint line is searched and set for all jump sections for the same approximation threshold value T, and then the approximation threshold value T is decreased by Δ. The approximation threshold may be reduced within the section, and the joining line may be searched and set. Such processing can be easily realized by a recursive program.

  The flow shown in FIGS. 3 and 4 corresponds to processing for extracting pixels indicating the ridgeline of the approximation distribution from the approximation distribution having a value of 0 to 255. Even if the ridge line is searched from the multi-value approximation distribution, a suitable joint line can be determined in the same manner.

  With reference to FIG. 10, FIG. 11, FIG. 12, and FIG. 13, the search for the joint line in the jump section will be schematically described.

  FIG. 10 shows a case where one jump section exists in the search range set in step S3. In FIG. 10, the solid line indicates the joint line searched in step S10 or step S11, and the broken line indicates a straight line connecting the end points (entrance point and exit point) of the jump section. The end point of the jump section is the start point or end point of the connection line on the same side as the end point. In step S15, a search range having a constant width W is set as shown in FIG. 11 in parallel with the line connecting the end points of the jump section shown in FIG. The width W is appropriately selected and set according to the images to be joined.

  As a result of searching for a joint line in steps S17 to S20 with respect to the search range shown in FIG. 11, a joint line (solid line) as shown in FIG. 12 is obtained, and a jump section indicated by a broken line remains. In this case, with respect to the jump section shown in FIG. 12, the approximation threshold T is further reduced by Δ (S14), a search range is set (S15), and a joint line is searched (S17 to S20). FIG. 13 shows the search result. Steps S14 to S20 are executed again for the remaining jump section. If the jump section is below a certain short distance (S16), the entry point and the exit point are directly connected (S21).

  When the jump section disappears (S13), the user manually edits / corrects the joining line manually using the mouse 18 or the like (S24). Two ortho images are synthesized with the corrected joint line (S25), and the obtained synthesized image data 42 is stored in the HDD 24.

  FIG. 14 shows an example in which the two ortho images shown in FIG. 16 are joined according to this embodiment. FIG. 14A shows the overlapping portion of the left ortho image, FIG. 14B shows the overlapping portion of the right ortho image, and FIG. 14C shows the joining result. A joint line is set to avoid the building and its collapse. That is, the images of the same building that are falling in opposite directions are not joined.

  Although the invention has been described with reference to specific illustrative embodiments, various modifications and alterations may be made to the above-described embodiments without departing from the scope of the invention as defined in the claims. This is obvious to an engineer in the field to which the present invention belongs, and such changes and modifications are also included in the technical scope of the present invention.

10: Image joining device 12: CPU
12a: Approximation degree calculation program 12b: Approximation degree binarization program 12c: Labeling processing program 12d: Thinning program 12e: Undesignated joint line search program 12f: Entrance designated joint line search program 12g: In-label joint line setting program 14: ROM
16: RAM
18: Mouse 20: Keyboard 22: Display device 24: Hard disk device (HDD)
26: Bus 30: Ortho image data 32: Ground elevation data (DTM) data 34: Surface elevation (DSM) data 36: Building flag table 38: Approximation data 40: Binary approximation data 42: Composite image data

Claims (28)

A method of joining two ortho image data obtained by photographing from different directions, at least partially overlapping, by a computer,
An initial search range setting step in which the computing means of the computer sets an initial search range in the overlapping portion of the two ortho image data, and stores it in the storage means of the computer;
The calculation means calculates an approximation degree of the two ortho image data in the initial search range, and stores the approximation degree in the storage means; and
The calculation means extracts pixels in the approximation value range in which the approximation obtained in the approximation calculation step is equal to or greater than a predetermined approximation value, groups the adjacent pixels, and determines the junction line search target pixels. A preprocessing step to extract;
The calculation means sequentially searches for a label close to a predetermined direction from a predetermined label of the bonding line search target pixel according to the bonding line search target pixel, and sets a bonding line to the predetermined label and each searched label. A connecting line search step,
The calculation means executes the preprocessing step in a smaller approximation value range for the jump section of the joint line, and for the joint line search target pixel extracted in this manner, Iterative steps for selecting a label close to a predetermined direction sequentially from a predetermined label and setting a joining line for each selected label;
An image joining method comprising: a joining step of joining the two ortho image data according to the joining line by the computing means to generate composite image data.   The image joining method according to claim 1, wherein the joining line search target pixel is thinned in the preprocessing step.   The computing means of the computer further comprises a step of setting a joint line connecting the jump section when the jump section of the joint line is within a predetermined distance. The image joining method described.   The image joining method according to any one of claims 1 to 3, wherein the predetermined label is a label having the largest area among the labels of the joining line search target pixels.   The predetermined label is a label having the longest length along a joint line determination reference line set at a predetermined distance from the two ortho images among the labels of the joint line search target pixels. The image joining method according to any one of claims 1 to 3.   The step of searching for a joining line sequentially searches for a label close to a predetermined direction in one direction along a joining line determination reference line set at a predetermined distance from the two ortho images from the predetermined label; 6. The image joining method according to claim 4, further comprising a step of searching for a label close to a direction opposite to one direction along the joining line determination reference line.   The predetermined label is any one of a label including an entry point set in the initial search range and a label of the joint line search target pixel closest to the entry point. 4. The image joining method according to any one of 3 above.   The image joining method according to claim 1, wherein the repetition step is executed every time the jump section is detected.   The repetition step is executed after completing the setting of the joint line to each label of the joint line search target pixel for the same approximation value range. Image joining method.   Further, the calculation means refers to the ground elevation data and surface elevation data stored in the storage means, and before the approximation calculation step, from the two ortho images to be processed in the approximation calculation step The image joining method according to any one of claims 1 to 9, further comprising a building excluding step for excluding the building. An image joining apparatus that joins two ortho image data obtained by photographing from different directions, at least partially overlapping,
Initial search range setting means for setting an initial search range in an overlapping portion of the two ortho image data;
An approximation degree calculating means for calculating an approximation degree of the two ortho image data in the initial search range, and storing the approximation degree in the storage means;
Read out the degree of approximation within the designated search range from the storage means, extract pixels whose read degree of approximation is within the designated degree of approximation value range, group them for each adjacent pixel, and select the junction line search target pixels. Pre-processing means to extract;
Among the labels of the joint line search target pixels generated by the preprocessing unit, a label near a predetermined direction is sequentially searched from a predetermined label, and a joint line is set to the predetermined label and each searched label. A joining line searching means for
A control means for controlling the preprocessing means and the joint line searching means. The initial search range is designated as the designated search range for the first time, and the upper limit value range is designated as the designated approximation value range. Then, the preprocessing means and the joint line search means are executed, and after the second time, the jump section of the set joint line is set as the designated search range, and the approximation value range reduced in stages is set to the designated approximation. As the degree value range, a control means for executing the preprocessing means and the joint line searching means,
An image joining apparatus comprising: combining means for joining the two ortho image data according to the joining line and generating composite image data.   12. The image joining apparatus according to claim 11, wherein the preprocessing means includes means for thinning the joining line search target pixels.   The image joining apparatus according to claim 11 or 12, further comprising means for setting a joining line connecting the inside of the jump section when the jump section of the joining line is within a predetermined distance.   The image joining apparatus according to any one of claims 11 to 13, wherein the predetermined label is a label having the largest area among the labels of the joining line search target pixels in the first time. .   The predetermined label is the label having the longest length along the bonding line determination reference line set at a predetermined distance from the two ortho images among the labels of the bonding line search target pixels in the first time. The image joining apparatus according to claim 11, wherein the image joining apparatus is provided.   In the first time, the control means searches for a label near the predetermined direction sequentially in one direction along the joining line determination reference line set at a predetermined distance from the two ortho images from the predetermined label. The image according to claim 14 or 15, wherein the processing for searching for a label close to a direction opposite to one direction along the joining line determination reference line is executed by the joining line search means. Joining device.   12. The predetermined label is any one of a label including an entry point set in the initial search range and a label of the joint line search target pixel closest to the entry point. 14. The image bonding apparatus according to any one of items 13. The storage means stores ground elevation data and surface elevation data,
Furthermore, it comprises a building excluding unit for excluding a building from the two ortho images to be processed by the approximation calculating unit before calculating the approximation by the approximation calculating unit. The image joining apparatus according to any one of 17. An image joining program for joining two ortho image data obtained by photographing from different directions to a computer and overlapping at least a part thereof,
An initial search range setting function for setting an initial search range in an overlapping portion of the two ortho image data and storing the initial search range in a storage unit of the computer;
An approximation calculation function for calculating the approximation degree of the two ortho image data in the initial search range and storing it in the storage means;
Pre-processing for extracting pixels in the approximate value range where the approximate value obtained by the approximate value calculation function is equal to or greater than a predetermined approximate value, and grouping the adjacent pixels and extracting the junction line search target pixels Function and
In accordance with the junction line search target pixel, a junction line search function that sequentially searches for a label close to a predetermined direction from a predetermined label of the junction line search target pixel and sets a junction line to the predetermined label and each searched label. When,
The preprocessing function is executed in a smaller approximation value range for the jump section of the junction line, and the junction line search target pixel extracted in this manner is sequentially started from a predetermined label of the junction line search target pixel. Repeat function that allows you to select a label that is close to a given direction and set a joining line for each selected label,
An image joining program for realizing a joining function for joining the two ortho image data according to the joining line and generating composite image data.   The image joining program according to claim 19, wherein the preprocessing function has a function of causing the computer to thin the joining line search target pixel.   21. The image according to claim 19, further comprising a function of causing the computer to set a joint line connecting the jump section when the jump section of the joint line is within a predetermined distance. Joining program.   The image joining program according to any one of claims 19 to 21, wherein the predetermined label is a label having the largest area among the labels of the joining line search target pixels.   The predetermined label is a label having the longest length along a joint line determination reference line set at a predetermined distance from the two ortho images among the labels of the joint line search target pixels. The image joining program according to any one of claims 19 to 21.   A function for searching for a label that is sequentially close to a predetermined direction in one direction along a bonding line determination reference line set at a predetermined distance from the two ortho images from the predetermined label; The image joining program according to claim 22 or 23, wherein the computer realizes a function of searching for a label close to a direction opposite to one direction along the joining line determination reference line.   20. The predetermined label is any one of a label including an entry point set in the initial search range and a label of the junction line search target pixel closest to the entry point. 21. The image joining program according to any one of 21.   The image joining program according to any one of claims 19 to 25, wherein the repetitive function is executed every time the jump section is detected.   26. The repetition function is executed after completing the setting of the joint line to each label of the joint line search target pixel for the same approximation value range. Image joining program.   Further, a building that refers to the ground elevation data and surface elevation data stored in the storage means and excludes the building from the two ortho images to be processed by the approximation calculation function before the approximation calculation function The image joining program according to any one of claims 19 to 27, wherein the computer is caused to perform an exclusion function.

Images