A deformation monitoring method based on DSM data

Abstract

The present invention discloses a deformation monitoring method based on DSM data. The method comprises the following steps: performing aerial survey of the object to be measured, generating orthophotos using photogrammetric modeling software, and extracting at least two phases of DSM data; determining a resampling interval d, and resampling the DSM data; cropping the resampled DSM data, calculating similar points, and obtaining displacement change data in the x, y, and z directions; filtering the obtained displacement change data to remove noise from the data and obtain displacement data; and uniformly performing error correction on the displacement data using correction data to obtain the displacement amount. The present invention utilizes current aerial photogrammetry results, has a low overall cost, quickly obtains large-area data, has a wide monitoring range, and can monitor three-dimensional surface deformation data. While its accuracy is somewhat lower than that of traditional and GNSS measurements, it is higher than that of InSAR measurements, and its error can be controlled to the centimeter and decimeter levels.

Description

Technical Field

[0001] The present invention relates to the field of geotechnical engineering technology, in particular to a deformation monitoring method based on DSM data. Background Art

[0002] Deformation monitoring is widely used in geotechnical engineering and geological disaster prevention. Among the commonly used deformation monitoring methods, one is surface deformation monitoring (appearance). The means currently used in surface deformation monitoring methods mainly include manual total station monitoring, GNSS monitoring, surveying robots, slope radar and other methods. InSAR monitoring is applied to surface displacement monitoring over larger areas. The manual total station monitoring method is the earliest monitoring method and is still the main means of precision monitoring. Its advantages are mature technology, high accuracy and strong adaptability, but it requires the installation of special monitoring points, and manual operation leads to slow monitoring speed. GNSS monitoring technology has high accuracy and can realize wireless automatic collection of deformation data. The initial installation cost is high, but the subsequent cost is low. Monitoring points and monitoring base stations are required. Surveying robots take into account the advantages of manual total station monitoring and automated monitoring, but have the disadvantages of high cost and the need for on-site supervision. Slope radar monitoring technology can obtain timely slope deformation data and is widely used in disaster warnings. However, slope radar scans the slope surface, and the displacement obtained is relative to the radar, not the horizontal and vertical displacement. Slope radar is also expensive. Therefore, a convenient, low-cost monitoring method that can capture the entire deformation is needed. Summary of the Invention

[0003] The object of the present invention is to provide a deformation monitoring method based on DSM data, so as to solve the problem that there is no monitoring method in the prior art that is easy to use, low in cost and capable of obtaining overall deformation.

[0004] The present invention solves the above problems through the following technical solutions:

[0005] A deformation monitoring method based on DSM data, comprising:

[0006] Step S1: perform aerial survey on the object to be measured, generate orthophotos using photogrammetry modeling software, and extract at least two phases of DSM data;

[0007] Step S2: Determine the resampling interval d based on the deformation range of the object to be measured. If the resampling interval d exceeds the preset minimum interval (the minimum distance that meets the deformation measurement requirements) or the original interval of the DSM data, directly resample the DSM data using the smaller value of the minimum interval or the original interval of the DSM data.

[0008] Step S3: Crop the resampled DSM data according to the principle of retaining the maximum amount of data to ensure that the number of rows and columns of the data and the spacing between rows and columns are consistent;

[0009] Step S4, similarity point calculation: Calculate the distance of multiple DSM data points by using the distance formula. The distance formula is as follows:

[0010]

[0011] Based on the first phase of DSM, find the point with the minimum distance (x, y) point by point. min(d(x,y)) , and then perform displacement difference fitting, use Lagrange binomial L2(x) and L2(y) to find the optimal point in the x and y directions, and then perform difference calculation to obtain the horizontal displacement. The vertical displacement is directly subtracted from the corresponding elevation value to obtain the displacement change data in the x, y, and z directions;

[0012] Step S5: filtering the obtained displacement change data to remove noise in the data and obtain displacement data;

[0013] Step S6, error correction: Correct the determined fixed point in the field to obtain correction data of the displacement, and use the correction data to uniformly perform error correction on the displacement data to obtain the displacement amount.

[0014] The process also includes step S7: comparing the DSM data, testing the obtained displacement, determining the credibility of the monitoring data, and producing a displacement cloud map and a vector map in three directions after the data are tested.

[0015] The method for determining the resampling interval d in step S2 is: firstly estimate the error Δ according to the possible deformation of the object to be measured, and estimate the resampling interval d by Δ≈0.2d.

[0016] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0017] (1) The present invention utilizes the current results of aerial photogrammetry, which has a low overall cost, can quickly obtain large-area data, has a wide monitoring range, and can monitor three-dimensional surface deformation data. Its accuracy is slightly lower than that of traditional measurement and GNSS measurement, but higher than that of InSAR measurement, and the error can be controlled at the centimeter and decimeter level.

[0018] (2) The present invention obtains multiple phases of DSM data containing surface deformation information through aerial photogrammetry, extracts these surface deformation data, and uses them to analyze the deformation characteristics of the deformed body. This method has the advantages of being convenient and fast, the monitoring range covers the entire surface of the object to be measured, and the overall cost is low. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a flow chart of the present invention;

[0020] Figure 2 is the x-direction displacement diagram of the present invention;

[0021] Figure 3 is the y-direction displacement diagram of the present invention;

[0022] Figure 4 This is the z (settlement deformation) displacement diagram in the present invention. DETAILED DESCRIPTION

[0023] The present invention will be further described in detail below with reference to the examples, but the embodiments of the present invention are not limited thereto.

[0024] Example:

[0025] The present invention is based on the following theory: through the study of a large number of landslides and slope deformations, it was found that the deformation of slopes (slopes, landslides, etc.) still maintains the overall surface morphology within a local range, and the influence of rotation can be ignored within a small range. Taking this as the basic assumption, on this basis, the displacement increments of the surface at different positions can be obtained through multi-period surface similarity comparisons. Obviously, this assumption has applicable conditions. If the surface undergoes large-scale ground transformation, the similarity point search may fail or obtain wrong conclusions, but this shortcoming can be compensated by means of encrypted monitoring, thereby satisfying the above assumption. Therefore, this assumption is reasonable in theory and feasible in practice.

[0026] Assume that the surface morphology is a continuous function, the first period is F(x,y), the second period is G(x,y),

[0027] Based on the above assumptions, let the coordinates of the calculation point of the first phase DSM data be P0 (x0, y0), the search range is assumed to be a square with a side length of 2b, and let the coordinates of the second phase similar point be P1 (x1, y1).

[0028] Assume that the functions within the square with P0 as the center and side length 2b are:

[0029]

[0030]

[0031] Since point P1 is similar to point P0, point P1 can be moved to point P0 by translation, so that the corresponding calculation areas are consistent. The calculation area D is designed to satisfy x0-b≤x≤x0+b, y0-b≤y≤y0+b.

[0032] Then let the sum of the absolute values ​​of the subtraction of the corresponding terms of the first and second period functions be the distance between the two functions:

[0033] d(x, y) = ∫∫ D |g(x,y)-f(x,y)|dxdy

[0034] Obviously, based on the above assumptions, the set of surface functions that can be obtained conforms to the properties of distance space. According to the properties of distance space, when the functions are very close, their distance approaches 0. Based on this property, it can be assumed that when the distance within the search range of two images is the minimum, the two points can be approximated as similar. Therefore, surface similarity is transformed into the problem of finding the distance within the same search range of two images. After obtaining the similar points, the relative displacement (x1-x0, y1-y0, z1-z0) can be obtained.

[0035] In the actual calculation process, it is impossible to obtain continuous data, and only discrete monitoring points can be obtained. The DSM data spacing used most often is about 5-10 cm. The area of ​​the object to be measured is generally large. If the original spacing is used for calculation, the amount of calculation is very large. In this case, the data should be resampled. Assume that the resampled spacing is d. Where n is half of the number of rows (columns) after the square search box is discretized. The distance formula after discretization is:

[0036]

[0037] In the above formula, d 2 Is a constant, the calculated area is of the same size, so d will be removed in the actual calculation 2 , which can save calculation time. Therefore, the discretized distance formula can be rewritten as:

[0038]

[0039] In this way, we will obtain (2n) in the search range 2 distances, for (2n) 2 The minimum distance is taken as the similar point found.

[0040] Since the data is discrete, there is an error in the similar points found, and the error Δ should be less than half of the spacing d, that is: Obviously, deformation monitoring requires precise measurement, so interpolation estimation is also required for the similar points found. Here, we use polynomial interpolation. After testing, we use the found point as the center, expand one point in each of the top, bottom, left, and right directions, and perform Lagrange quadratic polynomial interpolation in both the horizontal x and y directions. Since the error estimates in the x and y directions are similar, we choose the x direction for error estimation.

[0041] Generally speaking, the more control points used in polynomial interpolation, the more accurate the interpolation is (because there is more control point information). Here, the difference between quadratic interpolation and cubic interpolation is used for error estimation.

[0042] Assume that the key points are x0=-d, x1=0, x2=d, x3=2d, and the distance values ​​of the corresponding points are dx0, dx1, dx2, dx3.

[0043] Lagrange quadratic polynomial interpolation formula:

[0044]

[0045] The corresponding Lagrange cubic polynomial interpolation formula is:

[0046]

[0047] Find the minimum extreme points of L2(x) and L3(x) at [x0, x2] respectively, and then perform difference operation on the extreme points to obtain the estimated error formula:

[0048]

[0049] Because dx1 is the position of the displacement similarity point, the coordinate translation will get dx1 = 0, and then assume the above formula is dx0 = dx2 = dx3, and finally get the approximate error formula:

[0050]

[0051] After obtaining the error estimation formula, the sampling interval can be determined according to the displacement development characteristics of the object to be measured. For example, if the displacement of the object to be measured is 1m and the allowable error is 0.1m, the sampling interval is 0.5m, which can greatly reduce the amount of data required for calculation.

[0052] At present, the minimum spacing of DSM data obtained by aerial photogrammetry is 5 cm, so the minimum deformation error can be controlled at 1 cm. If the deformation of the object to be measured exceeds 1 cm, it can be considered that the deformation is significant. This property can also be used to determine the undeformed area.

[0053] The displacements obtained contain a significant amount of noise, so filtering is a crucial step in data processing. Testing has shown that the deformation data obtained is subject to errors caused by vegetation, atmospheric influences, systematic errors in the measuring equipment, and other factors. These errors appear as impulse noise in the deformation map. Testing has shown that focal statistics from domain analysis can be used in geographic information systems to smooth and reduce noise, with the number of smoothing units typically being 3*3.

[0054] Based on the above theory, combined with the Figure 1 As shown, the present invention proposes a deformation monitoring method based on DSM data, and the implementation steps are as follows:

[0055] K1: Set image control points according to photogrammetry specifications around the object to be measured and its surroundings, conduct aerial surveys, and use relevant photogrammetry modeling software to generate orthophotos and DSM data. The photogrammetry range should extend 50-100 meters into the object, and at least two DSM data sets should be obtained.

[0056] K2: Investigate the deformation range of the object to be measured (or deformation range) and determine the resampling spacing d. If it exceeds the minimum spacing or the spacing of the original DSM data, directly use the minimum spacing and the original DSM spacing and resample the DSM. The size of the sampling spacing d is determined by the error estimation formula above. The required error size should be estimated based on the possible deformation of the object to be measured. The error is generally set to 10-20% of the deformation. The sampling spacing d is estimated using Δ≈0.2d.

[0057] K3: Crop the resampled data of DSM according to the most favorable data volume to ensure the same number of rows and columns and the same spacing.

[0058] K4: Select the initial phase and search for similar points point by point. Here, the distance formula of discrete points is used to calculate the distance:

[0059]

[0060] Then find the point with the minimum distance within the search range (x, y) | min(d(x,y)) After finding the extreme point (i.e. the point with the minimum distance), the displacement difference fitting is performed. The Lagrange binomial L2(x) and L2(y) are used to find the optimal points in the x and y directions, and then the difference calculation is performed to obtain the horizontal displacement. The vertical displacement is directly subtracted from the corresponding elevation value to obtain the displacement changes in the (x, y, z) directions.

[0061] K5: Filter the acquired data to remove noise.

[0062] K6: Calibrate the displacement data, mainly to correct the fixed points determined in the field, obtain the correction data of the displacement, and uniformly correct the errors of the displacement data.

[0063] K7: Verify the displacement obtained by comparing it with the monitoring data to determine the credibility of the monitoring data.

[0064] K8: After passing the inspection, make displacement cloud maps and vector maps in three directions.

[0065] Special parameter settings and handling of special situations:

[0066] (1) Data acquisition range: Considering that deformation requires fixed points for correction during later inspection, it is recommended to expand the area 50-100 meters outside the object to be measured. If there is interference and obstruction due to trees, it should be expanded further until a stable fixed area is found.

[0067] (2) Error correction: When collecting data in the field, the position of the fixed point should be determined, but the effect is often not ideal. There will often be a large area of ​​fixed parts. The mean value within this range can be used for error correction. When the error is small, no correction is required.

[0068] (3) Determination of the search range. It is initially recommended to use twice the estimated deformation as the search range. If it is too large, the amount of calculation will be large and there will be a lot of noise, making it difficult to find similar points. If it is too small, there will be too many failures in finding similar points, making it difficult to process the data later.

[0069] (4) There are many filtering methods. The obtained data are generally output in GeoTiff format. It is recommended to use the focus statistics tool in the geographic information system to perform smoothing filtering and noise reduction.

[0070] The present invention was used to conduct two phases of drone monitoring on a landslide that had sustained deformation. The resampling interval was 0.5m. The obtained x-direction displacement cloud map, y-direction displacement cloud map, and z (settlement) displacement cloud map were shown in Figure 2. Figure 2 、 Figure 3 and Figure 4 shown.

[0071] Figure 2 The x-direction displacement cloud map is displayed. The slope displacement values ​​are basically positive, indicating that the slope as a whole has a tendency to displace eastward. The distribution range is the edge of the slope, but the range is small.

[0072] Figure 3 The y-direction displacement cloud map is displayed. The displacement values ​​are basically positive and concentrated in the upper part of the slope (the soil part), indicating that the soil part of the slope has an overall landslide trend. The deformation range and intensity can be clearly seen through the deformation.

[0073] Figure 4 The z-direction displacement cloud map is shown. The vertical displacement is mainly concentrated in the middle and upper part of the slope (the middle and rear edge of the deformation zone), indicating that the slope as a whole has a tendency to sink. The range of the sinking is consistent with the range of the overall deformation of the slope.

[0074] Therefore, the displacement cloud maps in the x, y, and z directions obtained by this method can comprehensively analyze the scope of slope deformation, the intensity of deformation in each area, the deformation trend and other basic information, which can guide slope rescue and management, and make up for the shortcomings of slope radar, professional monitoring, etc.

[0075] It can be seen that the multi-period DSM data containing surface deformation information obtained by aerial photogrammetry can be extracted to analyze the deformation characteristics of the deformed body. The present invention has the advantages of being convenient and fast, covering the entire surface of the object to be measured, and having low overall cost.

[0076] Although the present invention is described herein with reference to illustrative embodiments of the present invention, the above embodiments are merely preferred embodiments of the present invention, and the embodiments of the present invention are not limited to the above embodiments. It should be understood that those skilled in the art can design many other modifications and implementations, which will fall within the scope and spirit of the principles disclosed in this application.

Claims

1. A deformation monitoring method based on DSM data, characterized in that: include: Step S1: perform aerial survey on the object to be measured, generate orthophotos using photogrammetry modeling software, and extract at least two phases of DSM data; Step S2: Determine the resampling interval d based on the deformation range of the object to be measured. If the resampling interval d exceeds the preset minimum interval or the original interval of the DSM data, directly resample the DSM data using the smaller value of the minimum interval and the original interval of the DSM data. Step S3: Crop the resampled DSM data according to the principle of retaining the maximum amount of data to ensure that the number of rows and columns of the data and the spacing between rows and columns are consistent; Step S4, similarity point calculation: Calculate the distance of multiple DSM data points by using the distance formula. The distance formula is as follows: ； Based on the first phase DSM, find the point with the minimum distance point by point , perform displacement difference fitting, using Lagrange binomial 、 Find the optimal point in the x and y directions, and then perform difference calculation to obtain the horizontal displacement. For the vertical displacement, directly subtract the corresponding elevation value to obtain the displacement change data in the x, y, and z directions. Step S5: filtering the obtained displacement change data to remove noise in the data and obtain displacement data; Step S6, error correction: Correct the determined fixed point in the field to obtain correction data of the displacement, and use the correction data to uniformly perform error correction on the displacement data to obtain the displacement amount.

2. The deformation monitoring method based on DSM data according to claim 1, characterized in that: The process also includes step S7: comparing the DSM data, testing the obtained displacement, determining the credibility of the monitoring data, and producing a displacement cloud map and a vector map in three directions after the data are tested.

3. The deformation monitoring method based on DSM data according to claim 1, characterized in that: The method for determining the resampling interval d in step S2 is: first, estimate the error ∆ based on the possible deformation of the object to be measured, and estimate the resampling interval d by ∆≈0.2d.

Images