Method for calculating horizontal displacement vector and vertical displacement vector in estimation of landslide surface shape

Abstract

To provide a method for calculating a horizontal displacement vector and a vertical displacement vector from DEMs at two times without requiring the setting of a rule and a threshold value for collation / determination and the process of error determination and recalculation in the estimation of a landslide surface shape.SOLUTION: A topographic quantity diagram meshed on the basis of altitude information in each period is prepared from a DEM prepared by measuring topographic quantities before and after a landslide variation, an analysis range is determined, and an aggregate of cells in the topographic quantity diagram before the landslide variation is extracted as a template image. The template image is superimposed on the topographic map image after the landslide change, and the difference calculation by the template matching of the SAD method is performed on all the cells in the analysis range, and the positional information is given to the center cell of the template image, and the horizontal displacement vector and the vertical displacement vector are created by setting the difference between the plane coordinates of the center cell of the aggregate of the cells where the value of the calculation is minimum and the center cell of the template image as the horizontal displacement and the altitude difference of the height coordinates as the vertical displacement.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a method for calculating horizontal and vertical displacement vectors when estimating the shape of a landslide slide surface based on displacement measurements of the topography before and after the landslide. [Background technology]

[0002] Landslide surface shapes vary depending on the topography and geology of the landslide site and the cause of the landslide (e.g., excavation work or natural phenomena such as earthquakes or heavy rain). Furthermore, landslide surface shapes often require estimation during disasters. Rapid and accurate assessment of landslide surface shapes and the implementation of appropriate countermeasures tailored to the scale and characteristics of the landslide can ensure the safety of disaster recovery activities and prevent or mitigate the risk of secondary disasters. In other words, a method that ensures speed, ease, and a certain level of accuracy based on the topography, geology, and cause of the landslide is desirable. However, landslides do not always occur in the same location. Given the varying topography, geology, and meteorological conditions at the landslide site, each case requires empirical judgment, which makes automation difficult. Meanwhile, securing human resources remains a future challenge in all industries and business types. Therefore, it is desirable to be able to assess landslide surface shapes with a certain level of accuracy, even without extensive experience in landslide assessment, as mentioned above.

[0003] Patent Document 1 is known as a method for automatically calculating horizontal and vertical displacement vectors, which are basic information for estimating the shape of a landslide surface. Patent Document 2 is a prior art document of Patent Document 1, and its problem is "resolved by meshing based on ground position information, and enabling house movement determination based on the difference in elevation representative values ​​assigned to the grids of laser data 1 acquired at two different times." It is a method for calculating and judging the difference in meshed topographic information between two different times in order to grasp topographic changes over a wide area at once. In contrast, the problem that Patent Document 1 aims to solve is "to provide a method and program for grasping topographic changes based on topographic information from multiple times, which does not require distinctive measurement control points on the topographic surface and can grasp topographic changes over a wide area and evenly across the surface." Therefore, the specification of Patent Document 1, paragraph

[0029] , states that "an image identical to or similar to an image depicted by a pixel set extracted from the pixel group of point cloud data 1 is detected from the window extracted from point cloud data 2 and its surrounding area (H), and a pixel set corresponding to the detected image is extracted (G2). Note that approximation in images can be determined by various methods, such as if the colors or brightness are not the same but fall within a predetermined tolerance range, or if the images of pixels in the pixel set match at a certain rate or more (for example, three out of four pixels match). In other words, this process involves capturing a portion of the ground surface before the change as a surface, as shown in Figure 11(a), extracting multiple surfaces (four surfaces in the figure), and then searching for similar surface combinations from a portion of the ground surface after the change, as shown in Figure 11(b)." This process of selecting pixels that are likely to match and extracting measurement reference points based on judgments under certain conditions imposed on the pixel set can be said to be an important process for calculating the horizontal displacement vector in estimating the landslide surface shape in Patent Document 1.

[0004] Furthermore, in the specification of Patent Document 1, paragraph

[0030] states, "When pixel sets are matched, not all of them necessarily match or are close together. For pixel sets that do not match, information is added to the pixels that make up these pixel sets as an error (does not match), and this is used for error judgment (process M in Figure 1) which will be explained later. Furthermore, pixels that have resulted in an error can be ignored in the analysis, or they can be analyzed after estimating which pixel they will be matched with from the image around them." As such, it can be said that a feature of this method is that accuracy is ensured by making error judgments and re-analyzing.

[0005] However, the method described in Patent Document 1 allows for a variety of methods for matching pixel sets, resulting in as many different judgment results as there are possible. It can be said that this method requires the empirical judgment of a skilled artisan, as mentioned above. Furthermore, if an error occurs in matching pixel sets, first, if the error itself is due to the matching method, re-matching is required. Second, as Patent Document 1's specification

[0034] states, "If the number of error pixels is between the upper and lower thresholds (arrow c), proceed to step Q in Figure 1." Values ​​that serve as the basis for determining the upper and lower thresholds must also be established. Especially for automation, it is difficult to achieve without appropriate matching methods and thresholds, and determining which method and threshold to use from the wide range of options is even more difficult. Furthermore, with the advancement of topographical surveying technology in recent years, there are increasing opportunities for high-resolution survey data. There are concerns that using high-resolution digital elevation models (DEMs) for analysis will further increase the processing load on the error detection and recalculation processes mentioned above. Furthermore, in methods for creating topographical maps using high-resolution data, smoothing of DEM data may be performed. As indicated in Patent Document 3, which states, "

[0029] An image processing method for correcting local brightness or contrast to make both bright and dark areas, or areas with strong and weak contrast, easier to see, smoothing in this technical field is a method for solving problems by providing visual benefits. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Patent No. 4545219 [Patent Document 2] Patent No. 4156320 [Patent Document 3] Patent No. 5198782 Summary of the Invention [Problem to be solved by the invention]

[0007] Therefore, the present invention provides a method for calculating horizontal and vertical displacement vectors from DEMs from two different periods when estimating the shape of a landslide slide surface, without the need to set rules for determining whether images match or thresholds for use in determining matching and / or errors, eliminating the process of searching for a single pixel that is likely to match or the process of recalculating due to errors that occur when matching pixel sets, and further reducing errors in the data after analysis even when using a high-resolution DEM; it is possible to provide a method for calculating horizontal and vertical displacement vectors with a certain level of accuracy guaranteed, without requiring the accumulation of experience by a person skilled in the art. [Means for solving the problem]

[0008] The method of calculating horizontal and vertical displacement vectors when estimating the shape of a landslide surface to solve the above-mentioned problems of the present invention involves acquiring information on plane coordinates and height coordinates for two periods, creating a DEM for each period, creating a topographical quantity map for each period from the DEM for each period, meshing the topographical quantity map for each period based on elevation information, assigning a topographical quantity to each cell of the meshed topography, setting a range for image extraction and difference calculation, extracting a template image from the image before the topography change, the template image being a collection of cells with an odd number of cells (excluding 1) vertically and the same number of cells horizontally as vertically, and with the same number of cells in all directions, starting from a central cell, and overlaying the template image on the plate image after the topography change, and performing difference calculations on the topographical quantities assigned to the overlapping cells using template matching based on the SAD method, and The calculation is performed for all combinations of the collection of cells after the terrain change while shifting the template image by one cell at a time, the difference in planar coordinates between the central cell of the collection of cells after the terrain change for which the calculated value by the difference calculation is the smallest and the central cell of the template image is taken as the horizontal displacement, and the difference in height coordinates is taken as the vertical displacement, the horizontal displacement information and the vertical displacement information are given to the central cell of the template image to create a horizontal displacement vector and a vertical displacement vector, the collection of cells in the same range with the position of the template image shifted by one cell is extracted as a new template image, the horizontal displacement vector and the vertical displacement vector are created by the method, and the extraction of the new template image and the creation of the horizontal displacement vector and the vertical displacement vector by the method are repeated until all combinations of cells within the analysis range before the terrain change are completed. Furthermore, among the above-mentioned features, the difference calculation is performed by setting an upper limit value for movement of the template image starting from a position on the image after the terrain change that is the same position as the position where the template image before the terrain change was extracted, and shifting the template image by one cell at a time, for all combinations within a range specified by the upper limit value for movement of the collection of cells after the terrain change. In addition to the above characteristics, the DEM for each period, which is created by obtaining information on the planar coordinates and height coordinates of the two periods, is high resolution, and the high-resolution DEM for each period is smoothed using an average elevation within a certain range, an image of a topographical map for each period is created from the smoothed DEM for each period, and the topographical map for each period is used for template matching difference calculations. [Effects of the Invention]

[0009] The method of the present invention for calculating horizontal and vertical displacement vectors when estimating the shape of a landslide surface can create horizontal and vertical displacement vectors without the need to set rules for comparing images from two different periods, rules for determining errors, or thresholds used for comparison and determination, thereby significantly reducing the number of situations in which judgments are made based on empirical rules. Furthermore, because the present invention performs differential analysis on all cells within the analysis range, unlike methods that select one pixel with a high probability of matching and then compare the pixel set containing that pixel, there are no cells that do not undergo differential analysis, ensuring greater accuracy. While this may at first glance appear to require more computation, first, the SAD method, which uses addition and subtraction to calculate the difference in template matching, can speed up the calculation compared to methods such as the SSD method, which uses multiplication. Second, it eliminates the need for the "step of determining errors in pixel sets that did not match," the "step of recalculating terrain quantities with error information," and the "step of comparing the recalculated pixel sets and determining errors," thereby preventing the execution program from becoming too complicated when automated and significantly reducing the amount of computation required, thereby improving not only speed but also efficiency. Furthermore, the method of calculating horizontal and vertical displacement vectors for estimating landslide surface shape according to the present invention, which uses a topographical map created from a smoothed DEM, smooths high-resolution DEM data to eliminate the strong shadows scattered in the topographical map and capture the entire topography as a large unit, making it possible to accurately grasp the direction and extent of movement of large-wavelength topography, which is important for estimating landslide surface shape.In addition, the method according to the present invention uses the minimum sum of topographical differences as a judgment criterion, and can eliminate information unnecessary for estimating landslide surface shape, such as boulders and surface topographical changes unrelated to landslides, thereby leading to even higher accuracy. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is a process flow diagram for estimating the shape of a landslide surface according to the present invention. [Figure 2] This is an explanatory diagram of topographical surveys conducted using UAV laser surveying at two different times, before and after the landslide. [Figure 3] This is a topographical map created using airborne laser survey data collected in 2020 (left) and 2023 (right). [Figure 4] Shaded relief maps for 2020 (left) and 2023 (right) created from DEM. [Figure 5] FIG. 1 is a conceptual diagram of template matching using the SAD method. [Figure 6] FIG. 10 is a flowchart illustrating an analysis of template matching using the SAD method. [Figure 7] This is a differential analysis diagram of horizontal displacement in 2020 and 2023. [Figure 8] This is a differential analysis of vertical displacement between 2020 and 2023. [Figure 9] This is a shaded relief map using DEM before smoothing. [Figure 10] This is a shaded relief map smoothed to an average elevation of a 2m radius. [Figure 11]This is a comparison example of the change in topographical quantity between a high-resolution DEM and a DEM after smoothing processing. DETAILED DESCRIPTION OF THE INVENTION

[0011] The present invention will be described with reference to the accompanying drawings, in which: The conditions, drawings, and the like described in the present invention are merely illustrative examples and are not intended to limit the scope of the present invention unless otherwise specified.

[0012] (Flow of the estimation method according to the present invention) An embodiment of a method for calculating horizontal and vertical displacement vectors when estimating the shape of a landslide surface according to the present invention will be described with reference to the process flow diagram of FIG. First, the topography before and after the landslide movement is surveyed to obtain information on the plane coordinates and height coordinates for each period (topography quantity measurement process), and a DEM for each period is created from the plane coordinates and height coordinates (DEM creation process). Second, a topographical quantity map for each period is created based on the topographical map created from the DEM, and the topographical quantity map for each period is meshed based on elevation information, and each cell of the meshed terrain has a pixel value based on the topographical quantity (topographical quantity map creation process). Third, the analysis range is determined based on the topographical map for each of the above periods, and a collection of cells within the analysis range of the topographical map before the landslide movement is extracted as a template image.The collection of cells is configured so that a cell exists in the center and one cell is arranged in each of the four directions, up, down, left, right, and diagonal directions, like a square divided into nine squares (template extraction process). Fourth, the template image is overlaid on the topographical quantity map after the landslide movement, and a template matching difference calculation is performed using the SAD method based on the topographical quantities (pixel values) assigned to the overlapping cells.The template image is then shifted horizontally or vertically by one cell, and the difference calculation using the SAD method is repeated for all cell combinations within the analysis range (template matching process). Fifth, all difference calculations within the analysis range are performed, and the difference in planar coordinates between the central cell of the cell group after the landslide has occurred, which has the smallest calculated value, and the central cell of the template image is defined as the horizontal displacement, and the difference in height coordinates is defined as the vertical displacement. Position information is then provided to the central cell of the template image, and a horizontal displacement vector and a vertical displacement vector are created (displacement vector creation process). Sixth, the template image is shifted horizontally or vertically by one cell from the extracted position, and a collection of cells in the same range as the template image is extracted as a new template image. Horizontal displacement vectors and vertical displacement vectors are created using the method described above. This process is repeated until all cell combinations within the analysis range of the pre-landslide topographical volume map from which new template images can be extracted are completed. Based on the above, the shape of the landslide surface is estimated based on the horizontal displacement vectors and vertical displacement vectors created for all cells within the analysis range (landslide surface shape estimation process).

[0013] (Method for determining displacement measurements of topography) To implement this invention, topographical survey results from at least two periods are required to obtain displacement measurements of the topography (01) before the landslide and the topography (04) after the landslide, as shown in Figure 2. Traditionally, landslide surface shapes have been estimated by core sampling through drilling, borehole testing, and landslide observation using boreholes. However, drilling boreholes in a manner that ensures safety for landslide observation requires advanced technology and costs. However, with the recent spread of topographical measurements using UAV-mounted laser scanners (02), it is now possible to conduct surveys of the topography before and after the landslide without physically inspecting the area. Airborne laser surveying and surveying methods using UAV-mounted laser scanners (02) as shown in Figure 2 are desirable from the perspectives of ensuring safety and speed during observation, as they can calculate displacement vectors from multiple observation points simultaneously. However, there is no limitation to the use of moving stakes, extensometers, surface inclinometers, etc., during observation and surveying to calculate displacement vectors.

[0014] (The process of creating two-time topographic map images from two-time DEMs) A DEM is created from the plane coordinates and height coordinates measured during the two periods before and after the landslide movement in the topographical survey. (Figure 3 is an example of a topographical map created from the DEM.) As shown in Figure 4, a topographical quantity map for each period is created based on the DEMs created for the two periods to be used in analysis. The topographical quantity map used in this example is a shaded relief map created by shining light onto the ground surface from the northwest, so that the northwest side of the uneven ground surface appears white and the southeast side appears black, with pixel values ​​ranging from 0 to 254, but the topographical quantity map used for analysis is not limited to this. The topographical quantity maps created for the two periods are then meshed based on elevation information, so that the smallest unit for which a pixel value is assigned is one cell.

[0015] (Template image extraction process) A template image is a pre-landslide topographic map composed of a collection of cells cut out from a pre-specified analysis area. As shown in template image a06 in Figure 5, a template image is a square composed of an odd number of cells in both the vertical and horizontal directions. In this example, a template image of nine cells is composed of a central cell a07, with one cell each located above, below, left, right, and diagonally around it. As will be described in detail later, this invention performs differential analysis based on the template image, assigns positional information to the central cell of the template image, and then creates horizontal and vertical displacement vectors. Therefore, it is desirable to extract the number of cells containing the central cell. Furthermore, Figure 4 shows a pixel (cell) size of 20 cm x 20 cm, and the template image cutout size is 20 m (100 pixels) x 20 m (100 pixels). However, if the number of pixels is not an odd number, it is recommended to adjust the size by adding or subtracting one pixel, for example, to 101 pixels x 101 pixels. The size of the template image is preferably such that a certain degree of contrast is included within the image, but the number of cells (pixels) included in one template image is not limited to this.

[0016] (Template matching process flow) As shown in Figures 5 and 6, the specific template matching process involves first extracting a template image a06, a collection of cells, from the pre-landslide topographical map (Figure 6(A)). Then, extract image b08 from the post-landslide topographical map (Figure 6(B)) at the same location and in the same area as template image a06. Then, overlay template image a06 (Figure 6(C)) on image b08 at the same location and in the same area. Next, template matching using the SAD method is performed based on the pixel values ​​assigned to template image a06 and image b08 (Figure 6(D)). Formula 09 for template matching using the SAD method is the formula shown in Figure 5. It calculates the pixel value errors between cells in the overlapping positions of template image a06 and image b08, and then calculates the sum of the absolute values ​​of these errors. If template image a06 and image b08 are identical, the sum of the absolute values ​​of the errors is zero. The smaller this value, the better the match is determined. First, the sum of the absolute values ​​of the errors for each cell in the superimposed template image a06 and image b08 is calculated (the "=270" at the bottom of the formula in Figure 5 is the sum of the absolute values ​​of the pixel errors for each cell, 10). The position of template image a06 on the post-landslide topographical map is shifted one cell at a time from the position of image b08, and the difference calculation using template matching is similarly performed to calculate the sum of the absolute values ​​of the errors for each cell, 10. This is repeated until the entire analysis range on the post-landslide topographical map has been completed. Then, once all combinations of images for the analysis range on the post-landslide topographical map have been completed, the position of the group of cells (image c) with the smallest sum of the absolute values ​​of the errors for each cell, 10, is determined. Next, the group of nine cells is shifted one cell, enclosed by a dotted line, and the sum of the absolute values ​​of the errors for each cell in template image a is calculated, and the minimum value within the analysis range is calculated. This process is repeated in this manner. If multiple minimum values ​​after calculation overlap, when calculating the horizontal displacement (dx, dy) described later in the specification (0017), it is advisable to output the average value of the difference between the planar coordinates of the multiple images where the minimum values ​​after calculation overlap as the horizontal displacement (dx, dy).In addition, if the same minimum value occurs frequently, the cutout size of the template image is small, which may result in low contrast and an inability to correctly grasp the terrain undulations and their movement.To solve this problem, it is recommended to enlarge the cutout size to an extent that several terrain undulations can be recognized within one template image and then perform the calculation again.

[0017] (Relationship between the value calculated by template matching and the displacement vector) Next, the difference between the planar coordinates of template image a and image c at the position of the minimum value determined in the above step (Fig. 6(e)) is defined as the horizontal displacement (dx, dy). Furthermore, the difference between the height coordinate after the landslide, given to the center cell of image c (Fig. 6(f)), and the height coordinate given to the center cell of template image a, is defined as the vertical displacement (dz). Horizontal and vertical displacement information (dx, dy, dz) is given to the center cell of template image a (Fig. 6(g)). After the horizontal and vertical displacement information for template image a (Fig. 6(h)) has been given, a new template image a' is extracted by shifting it by one cell from the position of template image a. The same steps as those performed for template image a are repeated, and the horizontal and vertical displacement information (dx', dy', dz') is given to the center cell of the new template image a'. This process is repeated until all cell groups that can be extracted within the analysis range on the pre-landslide topographical map are combined. Once horizontal and vertical displacement information is provided for all cells within the analysis range, horizontal and vertical displacement vectors for all cells are created based on that displacement information, and by combining these displacement vectors with a topographic map, the horizontal displacement differential analysis diagram shown in Figure 7 and the vertical displacement differential analysis diagram shown in Figure 8 can be created.

[0018] (About differential analysis using template matching) Template matching is a method of searching for the location on a post-landslide topographical map that is most similar to a template image extracted from a pre-landslide topographical map, and one of the methods used to search for this similarity is the Sum of Absolute Difference (SAD) method mentioned above. For example, another method, the SSD method, calculates the sum of the squared values ​​of the errors in the pixel values ​​of each cell and searches for the minimum value, just like the SAD method, and because the errors in each cell are squared, it has the characteristic that an error in a single pixel value can have a large impact on the sum. The purpose of creating horizontal and vertical displacement vectors in this invention is to estimate the shape of a landslide surface. The term "landslide" itself can be interpreted as the downward movement of rock, soil, or a mixture of these down a slope, and as a phenomenon in which a certain amount of the terrain on a slope shifts (or collapses). In estimating the shape of this landslide surface, it is important to understand topographical changes over a wide area, rather than just a small area. Therefore, it is desirable to be less susceptible to errors arising from local topography, such as the presence of small rocks. The SAD method is a method that is less susceptible to such errors. In addition, the SAD method is faster in calculations than the SSD method, which uses multiplication, making it a desirable method in terms of speed and ease.

[0019] (About setting the upper limit of template image movement) Landslide surface shape estimation typically involves not only UAV laser surveying but also on-site measurements, and it is sometimes possible to estimate an upper limit on the amount of displacement before performing differential calculations. In such cases, based on the estimated upper limit of displacement, it is advisable to set an upper limit on the number of cells to which the template image can be moved, starting from the same position on the post-terrain change image as the position where the template image before the terrain change was extracted, and then perform differential calculations for all combinations of post-terrain change cell clusters within a limited range. Setting the upper limit of movement can also be determined by visual estimation from the topographical volume image with experience. Adding this step to the template matching step (Figure 6(d)) significantly reduces the amount of calculation compared to performing differential calculations for all cell cluster combinations, enabling more rapid differential calculations.

[0020] (Regarding the estimation method of another embodiment of the present invention) Next, another embodiment of the method for generating horizontal and vertical displacement vectors when estimating the shape of a landslide surface according to the present invention will be described with reference to the flow chart of FIG. In this embodiment, after the first DEM creation process shown in Figure 1, a process of smoothing the DEM created between the first and second DEMs is added, and a topographical map is created based on the smoothed DEM data, and horizontal displacement vectors and vertical displacement vectors are created in the same way as in the previous embodiment.

[0021] (About the smoothing process) In recent topographical surveys, the ground resolution of laser surveys using general aircraft or UAVs is on the order of several tens of centimeters, but there are significant advantages to obtaining survey results as high-resolution data, such as drawing detailed three-dimensional maps that show subtle differences in shape and elevation, or using them to detect small obstacles such as falling rocks and boulders and to prevent accidents. For this reason, it can be said that the current mainstream is to use high-resolution data in estimating landslide surface shapes as well. Figure 9 shows a topographical map created using a DEM with high-resolution data (average elevation within a 0.2-meter radius), while Figure 10 shows a topographical map (shaded relief map) created by smoothing the high-resolution DEM (average elevation within a 2-meter radius). Comparing the two maps, the shaded relief map created using the DEM with high-resolution data in Figure 9 clearly shows the subtle topographical irregularities through the intensity of color. However, landslides are phenomena in which a certain area shifts en masse. To understand landslide movements caused by deep underground slides, it is important to understand the direction and extent of movement of relatively large-wavelength topography, such as ridges and valleys. For example, changes in surface irregularities due to the tilt of trees or boulders caused by landslides frequently occur in landslide phenomena, so focusing on subtle changes in surface irregularities does not necessarily guarantee accuracy. Figure 11 shows the topographical changes before and after a landslide when small changes in surface roughness occur. The left image shows the topography before the landslide, and the center image shows the topography after the landslide. The top row shows the high-resolution image, while the bottom row shows the smoothed image. The right image shows a comparison of the topography before and after the landslide. First, the right image, which shows the left and center images superimposed, shows a comparison at high resolution. This allows for the capture of small changes in topography as topographical quantities, thereby capturing many small differences. However, because template matching calculates the sum of the absolute values ​​of these differences, there is a risk that many small, incidental changes in topography may be interpreted as large differences. On the other hand, smoothing reduces the occurrence of small differences compared to high resolution. In other words, estimating the landslide surface shape using a topographical quantity map created by smoothing high-resolution DEM data can further reduce the influence of errors caused by local topographical changes during the template matching process, thereby improving the accuracy of the landslide surface shape estimation. Therefore, as shown in Figure 10, a more desirable method for estimating the shape of a landslide surface is to create a shaded relief map after smoothing the high-resolution DEM data, eliminate the small, scattered strong shadows, and capture the entire terrain as a large unit.

[0022] (Differences from conventional smoothing processing techniques) As mentioned above, there are inventions that smooth DEM data in methods for creating topographical maps, etc., but Patent Document 3 describes the smoothing process in the method for creating topographical maps as "an image processing method that corrects local brightness or contrast to make both bright and dark areas, or areas with strong and weak contrast, easier to see," and aims to solve the problem by providing a visual benefit. Therefore, the present invention, which obtains numerical values ​​that serve as basic data for understanding topographical changes by smoothing DEM data and then creates horizontal and vertical displacement vectors from the smoothed numerical values, has a completely different purpose and is recognized as novel and an invention that is recognized as having an inventive step based on the new idea that "high resolution does not equal high accuracy when estimating landslide surface shapes."

[0023] (Conclusion) From the above, the method of calculating horizontal and vertical displacement vectors when estimating the landslide surface shape according to the present invention is particularly quick and easy among the methods for estimating the landslide surface and measuring displacement based on displacement measurements of the topography before and after the landslide, and because it meets a certain level of accuracy, it is useful for ensuring safety in disaster recovery activities and quickly avoiding and mitigating the risk of secondary disasters. Furthermore, it is a method that can be used to understand the landslide surface shape with a certain level of accuracy, even if you do not have extensive experience in landslide judgments. [Explanation of symbols]

[0024] 01 Topography before the landslide 02 UAV-mounted laser scanner 03 Crack 04 Topography after landslide 05 Sliding cliff 06 Template image a 07 Center cell 08 Image B 09 Calculation formula using the SAD method 10. Sum of absolute values ​​of pixel value errors in each cell 11 Topography before landslide (high resolution) 12 Topography before landslide (after smoothing) 13 Topography after landslide (high resolution) 14 Topography after landslide (after smoothing) 15 Comparison of topography before and after landslide (high resolution) 16 Comparison of topographical features before and after landslide (after smoothing)

Claims

1. Obtain information on the plane coordinates and height coordinates for two periods, create DEMs for each period, and create topographic maps for each period from the DEMs for each period. The topographical quantity map for each time period is meshed based on elevation information, a topographical quantity is assigned to each cell of the meshed topography, a range for image extraction and difference calculation is set, and a template image is extracted from the image before the topographical change; The template image is a collection of cells with an odd number of cells (excluding 1) vertically, an equal number of cells horizontally, and an equal number of cells vertically and diagonally from a central cell, and the template image is superimposed on the plate image after the topography change, and the topography amounts given to the overlapping cells are calculated by template matching using the SAD method; The difference calculation is performed for all combinations of the cell aggregates after the topography change while shifting the template image by one cell at a time; the difference in planar coordinates between the central cell of the cluster of cells after the terrain change, which has the smallest calculated value by the difference calculation, and the central cell of the template image, is taken as the horizontal displacement, and the difference in height coordinates is taken as the vertical displacement, and the horizontal displacement and vertical displacement information is given to the central cell of the template image to create a horizontal displacement vector and a vertical displacement vector; A collection of cells in the same range as the template image is extracted by shifting the position of the template image by one cell, and horizontal displacement vectors and vertical displacement vectors are created by the method described above. The extraction of the new template image and the creation of the horizontal displacement vector and the vertical displacement vector by the method are repeated until all the combinations of cells within the analysis range before the topography change are completed. A method for calculating horizontal and vertical displacement vectors when estimating the shape of a landslide surface.

2. The difference calculation is performed by setting an upper limit value for movement of the template image starting from a position on the image after the terrain change that is the same as the position where the template image before the terrain change was extracted, and shifting the template image by one cell at a time for all combinations within a range specified by the upper limit value for movement of the collection of cells after the terrain change.

2. A method for calculating horizontal displacement vectors and vertical displacement vectors when estimating a landslide surface shape according to claim 1.

3. The DEMs for each period, created by obtaining information on the plane coordinates and height coordinates for the two periods, have high resolution, The high-resolution DEM for each period is smoothed at a certain range of average elevation, A topographic map of each period was created from the smoothed DEM for each period. The topographical map at each time is used for difference calculation of template matching.

3. A method for calculating horizontal and vertical displacement vectors when estimating the shape of a landslide surface according to claim 1 or 2.

Images