JP4331358B2 - Aerial photogrammetry check method using triplet method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to aerial photogrammetry, and more particularly to an aerial photogrammetry check method for improving the reliability of measurement results in aerial photogrammetry.
[0002]
[Prior art]
As countermeasures for sedimentation turbidity problems in dams in mountainous areas, it is necessary to determine changes in secular change in the mountainous areas. If necessary, it is necessary to conduct a landslide survey in the mountainous area. However, for the landslide three-dimensional measurement technique for this purpose, the survey target area is usually located in the back of the mountain and covers a wide area. On-site surveying and measurement were extremely difficult. Therefore, conventionally, a three-dimensional measurement method based on the most recent stereo aerial photograph of the mountain area and a newly taken stereo aerial photograph has been employed. That is, this conventional three-dimensional measurement method is based on the difference calculation between the three-dimensional coordinates of a predetermined point measured from a previously taken stereo aerial photograph and the three-dimensional coordinates of the same point measured from a newly taken stereo aerial photograph. The amount of sediment caused by sediment collapse was calculated.
[0003]
[Problems to be solved by the invention]
However, this conventional three-dimensional measurement method outputs a measurement result by one stereo matching, and cannot be said to be sufficient in terms of reliability of the measurement result. In particular, in a steep terrain, the parallax difference is too large and stereo viewing is possible but cannot be measured, or the left and right aerial photographs have different patterns in steep terrain, and stereo viewing may not be possible. In collapsed terrain, the pattern was flat and white, and stereo vision was often not possible.
[0004]
As this type of measurement technique, there is a so-called AUTO-3D measurement technique as disclosed in Japanese Patent Publication No. 8-16930 by the present applicant. Then, the following problems remain. That is, in the current AUTO-3D measurement method, the image correlation method is used, but in the case of a flat pattern without a stereo pattern, the correlation coefficient is 1.0 everywhere even if it is not the same point (corresponding point), Pass / fail cannot be determined only with the correlation coefficient. In other words, in the current method, even when stereo viewing is not possible due to occlusion, different patterns with a correlation coefficient of 0.8 to 0.9 are regarded as the same pattern and are judged to be good, and therefore sufficiently reliable. It was not possible to obtain sexual measurement results.
[0005]
An object of the present invention is to provide an aerial photogrammetry check method by a triplet method that can solve the problems of the prior art as described above.
[0006]
[Means for Solving the Problems]
According to the present invention, an aerial vehicle for obtaining a three-dimensional coordinate value of a point in the shooting target area based on a series of aerial photographs taken at a predetermined shooting lap rate along a shooting course covering the shooting target area. In the photogrammetry check method, three consecutive aerial photographs are selected from the series of aerial photographs. Among the three aerial photographs, an aerial photograph located on the left is taken as a left photograph and an aerial photograph located on the right. The right photograph, the aerial photograph located between the left photograph and the right photograph is used as the central photograph, AD conversion data of each photograph is created, and the three-dimensional measurement by the left model based on the left photograph and the central photograph is performed. 3D measurement is performed with the right model based on the right photograph and the center photograph, and the corresponding points are compared between the measurement result with the left model and the measurement result with the right model. The three-dimensional measurement by the left model and the three-dimensional measurement by the right model are performed again until the image is obtained, and both ends based on the left photograph and the right photograph are used by using the corresponding point photograph coordinate values where the corresponding points are matched. A three-dimensional coordinate value is obtained by performing three-dimensional measurement using a model.
[0007]
According to one embodiment of the present invention, the comparison of the corresponding points is performed by generating a square mesh on the central photograph and obtaining left and right corresponding points based on the points.
[0008]
According to another embodiment of the present invention, the selection of the three aerial photographs is repeatedly performed after obtaining the three-dimensional coordinate values, and only one photograph is selected from the previous selection. By shifting to the right side, a three-dimensional coordinate value over the entire area to be imaged is obtained.
[0009]
According to one embodiment of the present invention, the predetermined shooting lap rate is 80%.
[0010]
DETAILED DESCRIPTION OF THE INVENTION
Next, based on an accompanying drawing, the present invention is explained in detail about an embodiment and an example of the present invention.
[0011]
FIG. 1 is a flowchart showing an overall procedure of the aerial photogrammetry check method of the present invention, and FIG. 2 is a diagram for explaining a left model, a right model, and a both-end model used in the aerial photogrammetry check method of the present invention. FIG.
[0012]
For example, in the case of applying the aerial photogrammetry check method of the present invention to the implementation of a landslide survey in a mountainous area, first, a series of shooting courses covering a shooting target area including a mountainous area that is the target of the survey are performed. You need to take aerial photos. Before taking such an aerial photograph, you can take a picture at any shooting position and at any shooting interval on any flight course so that all parts are shot over the target area. It is necessary to make a shooting plan in advance about what to do. Therefore, an aerial photography simulation plan is made as shown in step 11 of FIG.
[0013]
This aerial photography simulation plan cannot be projected if an aerial photograph is taken for any shooting course, at any shooting lap rate, and at any course shooting start position for the shooting target area. Finding out if the area can be minimized without actually flying the aircraft. The present applicant discloses an aerial photography plan simulation method for this purpose in the specification and drawings of another patent application. According to this, in the aerial photography plan simulation method for simulating a photography plan for taking an aerial photograph that covers a photography target area, three-dimensional mesh data representing the terrain including at least the photography target area is prepared, Along with parameters representing shooting conditions such as focal length and film size, it is input to a computer, a projection impossible area threshold value is set, a shooting course for the shooting target area is set, and a shooting lap rate is set along the shooting course. Set the course shooting start position, set the shooting position within the course, calculate the numerical value of the non-projectable area at the shooting position within the course, and calculate the numerical value of the non-projectable area and the non-projectable area. The quality of the shooting plan is determined by comparing the threshold value.
[0014]
Next, a series of aerial photographs are taken along a shooting course covering the shooting target area in step 12 in accordance with the shooting plan established as described above in step 11. Usually, the areas subject to landslide investigations include a lot of steep slope topography, and in order to minimize the parts that cannot be measured by occlusion, the shooting lap ratio is set to 80% and the shooting baseline length of stereo photography is made as short as possible. .
[0015]
Photo processing of the aerial photograph thus taken is performed in step 13, aerial triangulation is performed in step 14, and aerial triangulation error is corrected in step 15. In order to enable measurement by a computer, AD conversion data corresponding to each image data of the series of aerial photographs obtained in this way is created in step 16.
[0016]
Next, the three-dimensional coordinate value (X, Y, Z value) of each point in the imaged target area is obtained by the triplet method according to the present invention. First, in step 17, three-dimensional measurement using the left model is performed. In step 18, three-dimensional measurement using the right model is performed. Here, the three-dimensional measurement in the left model and the right model will be described in detail with reference to FIG.
[0017]
FIG. 2 shows three consecutive aerial photographs of a series of aerial photographs taken at a shooting lap rate of 80% along the photographing course in that order. In FIG. 2, the left aerial photograph is the left photograph 1, the right aerial photograph is the right photograph 3, and the aerial photograph between the left photograph 1 and the right photograph 3 is the central photograph 2. In the embodiment described below, the left photograph 1 is the first aerial photograph of the aerial photographs taken along the photographing course, the central photograph 3 is the next aerial photograph, and the right photograph 2 is the next photographed. It is said that it is an aerial photograph.
[0018]
The left model is a pair of stereo aerial photographs composed of a left photograph 1 and a central photograph 2. In step 17, the three-dimensional measurement by the left model is performed using the corresponding AD conversion data, and the so-called AUTO described above. The three-dimensional coordinate value of each point is obtained according to the -3D measurement method. Similarly, the right model is a pair of stereo aerial photographs consisting of the center photograph 2 and the right photograph 3, and in step 18, using the corresponding AD conversion data, according to the so-called AUTO-3D measurement method described above. The three-dimensional coordinate value of each point is obtained.
[0019]
Next, in step 19, the corresponding points are compared. This is done by comparing the three-dimensional coordinate value obtained in step 17 with the three-dimensional coordinate value obtained in step 18. Since it is necessary to check the corresponding points at the shooting lap rate of 80% at the same point, a square mesh is generated as shown by hatching on the center photo 2 of the triplet, and these points are used as the reference. Adopt a method to find the corresponding points. In this way, when the same point is double-measured on the central photograph using three consecutive aerial photographs and converted to ground coordinate values, both are close to the same value. It is determined that the pattern matching of the corresponding points in the dimension measurement is correct, and the measurement result is adopted as reliable data.
[0020]
On the other hand, if the measurement result by the left model and the measurement result by the right model do not match and are considerably different, it is determined that there is a matching point matching error, and the process returns to step 17 and step 18 to Perform measurement with the right model again by selecting the corresponding point. In step 19, the corresponding points are compared again. In this way, such a procedure is repeated until it is determined that matching of corresponding points has been performed correctly.
[0021]
In step 19, after it is determined that the measurement results by the left model and the right model are reliable, the process proceeds to step 20, and the procedure for correcting the three-dimensional coordinate values by both end models is entered. Here, the two-end model is a pair of stereo aerial photographs including a left photograph 1 and a right photograph 3, as shown in FIG.
[0022]
In the present invention, the reason why the measurement result is corrected in step 20 is as follows. In particular, when there are many steep terrains in the measurement target area, matching of matching points is better for the left model with a short shooting base length and a large shooting lap rate (80%) and the right model. Although reliable, the measurement accuracy of the Z value, which indicates the ground height among the three-dimensional coordinate values to be measured, is the same as that of the three-dimensional measurement using the two-end model with a long shooting base length and a small (60%) shooting lap rate. It is because it is thought that becomes high.
[0023]
In step 20, three-dimensional measurement using both end models is performed using corresponding AD conversion data, and the three-dimensional coordinate value of each point is obtained according to the so-called AUTO-3D measurement method described above. Matching of corresponding points in is performed by selecting the points confirmed as reliable in step 19 as corresponding points.
[0024]
Finally, in step 21, the three-dimensional coordinate value of each point is determined as a correct three-dimensional measurement result. Here, this determination is performed by determining that the X value and the Y value of the three-dimensional coordinate values are the steps. The measurement result of the left model or the right model confirmed as having no corresponding point matching error in 19 is adopted, and the measurement result of both end models is adopted in step 20 as the Z value of the three-dimensional coordinate values. It will be done in a way. In terms of measurement accuracy, the measurement results of the two-end model are the most accurate in terms of Z-value measurement, but there may be cases where the measurement values of the left-hand model and right-hand model are the same, but the two-end model cannot be measured by occlusion. However, in this case, the average value of the measurement results of the left model and the right model with a shooting lap rate of 80% is adopted.
[0025]
In order to obtain the three-dimensional coordinate values of all points in the shooting target area, the measurement based on one set of the left photograph, the center photograph, and the right photograph as described above is repeated. In addition, the photo selected as the left photo is shifted to the right by one photo.
[0026]
In the description of the embodiment described above, the shooting lap rate is 80%. However, the present invention is not limited to such a value, and an arbitrary shooting lap rate can be selected. For example, When it is desired to measure an area including a lot of extremely steep and steep terrain, it is conceivable that the photographing lap rate is 80% or more, for example, 90%.
[0027]
【The invention's effect】
The reliability of measurement results by aerial photogrammetry can be increased, and in particular, the reliability of measurement results in areas including steep terrain can be increased. Therefore, the present invention is effective when applied to a landslide investigation in which a region containing a lot of steep terrain is often a subject of investigation.
[Brief description of the drawings]
FIG. 1 is a flowchart illustrating an overall procedure of an aerial photogrammetry check method according to the present invention.
FIG. 2 is a diagram for explaining a left model, a right model, and a both-end model used in the aerial photogrammetry check method of the present invention.
[Explanation of symbols]
1 Left photo 2 Center photo 3 Right photo

Claims (4)

In an aerial photogrammetry check method for obtaining a three-dimensional coordinate value of a point in the shooting target area based on a series of aerial photographs shot at a predetermined shooting lap rate along a shooting course covering the shooting target area, Three consecutive aerial photographs are selected from the series of aerial photographs. Among the three aerial photographs, an aerial photograph located on the left is a left photograph, an aerial photograph located on the right is a right photograph, and the left photograph. The aerial photograph located between the left photograph and the right photograph is used as a central photograph, AD conversion data of each photograph is created, three-dimensional measurement is performed by a left model based on the left photograph and the central photograph, Perform three-dimensional measurement with the right model based on the central photograph, compare the corresponding points between the measurement result with the left model and the measurement result with the right model, and until the corresponding points are matched The three-dimensional measurement by the left-hand model and the three-dimensional measurement by the right-hand model are performed again, and the three-dimensional measurement by the both-end model based on the left photograph and the right photograph is performed using the corresponding point photograph coordinate values in which the corresponding points are matched. An aerial photogrammetry check method characterized by obtaining a three-dimensional coordinate value by performing measurement. 2. The aerial photogrammetry check method according to claim 1, wherein the comparison of the corresponding points is performed by generating a square mesh on the central photograph and obtaining left and right corresponding points based on the square mesh. The selection of the three aerial photographs is repeatedly performed after obtaining the three-dimensional coordinate values, and the selected photograph is shifted to the right by one from the time of the previous selection. The aerial photogrammetry check method according to claim 1 or 2, wherein a three-dimensional coordinate value over the entire target area is obtained. The aerial photogrammetry check method according to claim 1, wherein the predetermined shooting lap ratio is 80%.

Images