A joint terrain modeling method using satellite laser altimetry data and stereo image pairs

Through satellite laser height measurement data and stereoscopic image joint terrain modeling, a variety of screening conditions and filtering technologies are used to solve the accuracy problems caused by the difficulty of obtaining control points in remote areas, and the generation of high-precision digital elevation models is realized, and the land resources are achieved. Surveys provide better basic geographic information.

CN118710838BActive Publication Date: 2024-11-19NANJING UNIV OF INFORMATION SCI & TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411174654.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-26
Publication Date
2024-11-19
Estimated Expiration
2044-08-26

AI Technical Summary

Technical Problem

In remote areas and dense vegetation areas, it is difficult to obtain large-area uniformly distributed ground control points, resulting in limited accuracy and reliability of satellite stereo pairs and laser altitude measurement data, making it difficult to build a high-precision digital elevation model.

Method used

Joint topographic modeling methods through satellite laser height measurement data and stereoscopic images, including obtaining stereoscopic images to perform radiation correction and matching of the same name point, filtering laser point data using multiple filtering conditions and constraint criteria, combining direct linear transformation models and asymmetry Triangular mesh filtering technology generates high-precision digital elevation model.

Benefits of technology

It effectively improves the accuracy of the digital elevation model, can provide higher-precision basic geographic information products for applications such as land and resource investigation and disaster monitoring, and solves the accuracy problems caused by difficulties in obtaining control points.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118710838B_ABST
    Figure CN118710838B_ABST
Patent Text Reader

Abstract

The invention discloses a satellite laser altimetry data and stereo image pair joint terrain modeling method, which relates to the technical field. Firstly, the stereo image pair of a study area is preprocessed by matching the same-name points and free network adjustment to obtain a relatively oriented stereo image; secondly, the satellite laser altimetry data of the study area is obtained, and the h5 file is parsed into a CSV track file of attribute data related to the elevation data of the satellite laser altimetry point; thirdly, the obtained laser altimetry data is screened by using a plurality of constraint criteria to obtain laser points with high position accuracy, and then the laser points are used as control points to puncture points on the stereo image pair and are used as constraint conditions to solve error equations; finally, a DSM is generated by the front intersection of the stereo image pair, and a digital elevation model DEM is generated after filtering, thereby effectively improving the accuracy of the DEM and providing basic geographic information products with higher accuracy for applications such as land and resources surveys.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of remote sensing and land resources surveying and mapping, and in particular to a terrain modeling method combining satellite laser altimetry data and stereo image pairs. Background Art

[0002] Digital Elevation Model (DEM) represents ground elevation through a set of ordered numerical arrays. It is a discrete mathematical expression of the topography of the earth's surface and an important basic data for geographic information system data processing and spatial analysis. At present, DEM data is widely used in surveying and mapping engineering construction, precipitation runoff analysis, and flood inundation range analysis. Therefore, constructing high-precision DEM is an important task to provide basic geographic information products for land and resources surveys and disaster monitoring.

[0003] Stereo pair photogrammetry is an important technology for terrain modeling using the principle of stereo vision. By acquiring two pairs of images with overlapping areas, a three-dimensional terrain model can be reconstructed. However, accurate terrain modeling requires the assistance of control points, which are used to calibrate the images and improve the accuracy of the model. Stereo pairs have a wide range of applications, covering everything from satellite images to aerial photography. For example, satellite stereo pairs can be generated by images taken by remote sensing satellites at different times, while aerial stereo pairs are constructed by continuous images taken by aircraft or drones during flight. Both satellite and aerial stereo pairs can provide detailed terrain information and are widely used in map making, environmental monitoring, urban planning and other fields. Field measurement is the traditional way to obtain ground control points, but this method of manual measurement on the spot requires a lot of manpower and material resources. In remote areas and areas with dense vegetation, it is often difficult to obtain large areas of evenly distributed measured control points.

[0004] As an active remote sensing technology, satellite laser altimetry has been updated from the wavelength of centimeters and millimeters to the current nanometer level, and the measurement accuracy has gradually improved. With the development of satellite laser altimetry technology, laser altimetry data has greatly enriched the earth's stereoscopic observation information due to its wide coverage, high accuracy, and open source, and has provided an important guarantee for improving the accuracy of image geometric positioning. For example, ICESat-2 (Ice, Cloud and Land Elevation Satellite-2) is an advanced satellite laser altimetry mission. Through its ATLAS (Advanced Terrain Laser Altimeter System) sensor, ICESat-2 can measure ice sheet thickness, ocean height, forest structure, and terrain changes with extremely high accuracy. However, due to atmospheric interference, surface reflection characteristics, and errors in the instrument itself, this type of data may contain noise and inaccurate measurement results. In order to ensure the reliability and accuracy of the data, it must be carefully screened and corrected. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides a method for joint terrain modeling using satellite laser altimetry data and stereo image pairs, comprising the following steps:

[0006] S1, obtain the stereo image pair remote sensing image of the study area, and perform radiation correction preprocessing on it; then perform homonymous point matching and free network adjustment calculation on the stereo image pair data to obtain the stereo image of the study area after relative orientation; then select the filter to perform dense point cloud matching to generate point cloud, and obtain the digital elevation model without control points by performing progressive triangulation filtering on the point cloud;

[0007] S2, obtain the satellite laser altimetry data of the study area and preprocess it, parse each h5 file of the data into a CSV track file, retain the strong light track file, integrate multiple CSV files, and obtain a CSV file containing attribute data related to the satellite laser altimetry point elevation data;

[0008] S3, the laser altimetry point elevation data and its attributes obtained in step S2 include 9 attributes related to the satellite laser altimetry point elevation quality, and 10 screening conditions from a to j are designed using the 9 attributes related to the satellite laser altimetry point elevation quality respectively;

[0009] S4. Formulate eight different levels of constraint criteria based on the ten screening conditions designed in step (3), and number the eight constraint criteria from level one to level eight, with the complexity of the constraint criteria increasing in sequence from level one to level eight;

[0010] S5. Use the first-level to eighth-level criteria to filter the original laser point data in sequence. The higher the level of constraint criteria, the fewer laser points will be filtered out. According to the size of the experimental sample area or the task requirements, the optimal constraint criteria is selected as the final standard to obtain the filtered laser point data.

[0011] S6, puncturing the control points obtained in step S5 on the stereo image pair, and bringing the control point data into the direct linear transformation model compensation equation as a constraint condition to solve the error equation; then updating the stereo image imaging geometric model to obtain the spatial coordinates of the connection points after the accuracy is improved;

[0012] S7. The coordinates of the encrypted points are obtained by the front intersection of the stereo image pair, and then the ground points are obtained by using progressive triangulation filtering, and finally a digital elevation model is generated.

[0013] The technical solution further defined in the present invention is:

[0014] Furthermore, in step S1, a scale-invariant feature transformation algorithm is used to match the points of the same name. The specific process includes detection of scale space extreme values, positioning key points and determining feature directions, feature point description, and feature point matching. The equation is shown below:

[0015]

[0016]

[0017]

[0018]

[0019] in, Represents the scale space of the image; Represents the coordinates of the original image; represents a scalable two-dimensional Gaussian function; represents the scale space factor; represents the Difference of Gaussian pyramid; Indicates scale; and They respectively represent the amplitude and argument of each pixel in the neighborhood window where the feature point is located.

[0020] As described above, in a satellite laser altimetry data and stereo image pair joint terrain modeling method, in step S1, a bundle adjustment method is used to perform free network adjustment calculation, as shown in the following formula:

[0021]

[0022] in, Represents the optimal parameter vector after optimization solution; Represents the parameter vector that needs to be optimized, including the camera's internal orientation elements, external orientation elements, and the three-dimensional coordinates of the object point; Indicates the number of observed image points; is a function, which refers to The reprojection error of each observed image point.

[0023] As described above, in the method for joint terrain modeling of satellite laser altimetry data and stereo image pairs, in step S1, the specific method of radiation correction is: using the sensor calibration coefficient to complete the radiation calibration, using the radiation transfer model method to perform atmospheric correction on the stereo image pairs to eliminate noise.

[0024] In the method for joint terrain modeling of satellite laser altimetry data and stereo image pairs as described above, in step S2, the attributes related to the satellite laser altimetry point elevation data are set to 16 types, namely, the reference digital elevation model elevation value, the laser point fitting elevation in each 100m statistical unit, the laser point fitting elevation in each 20m statistical unit, the laser point interpolation land elevation, the maximum elevation of the laser point in each 100m statistical unit, the minimum elevation of the laser point in each 100m statistical unit, the median elevation of the laser point in each 100m statistical unit, the average elevation of the laser point in each 100m statistical unit, the number of laser points divided in each 100m statistical unit, the quality mark, the slope mark, the cloud cover mark, the city mark, the quantity mark, the day and night mark, and the standard deviation of the laser point elevation.

[0025] As described above, in a method for joint terrain modeling of satellite laser altimetry data and stereo image pairs, in step S3, the nine attributes related to the quality of the satellite laser altimetry point elevation are the reference digital elevation model elevation value, the laser point interpolated land elevation, the average elevation of the laser point in each 100m statistical unit, the quality mark, the slope mark, the cloud cover mark, the city mark, the quantity mark, and the day and night mark.

[0026] As described above, in the satellite laser altimetry data and stereo image pair joint terrain modeling method, in step S3, the 10 screening conditions are:

[0027] a.|Reference digital elevation model elevation value - laser point interpolation land elevation|≤3 , =25m;

[0028] b.|Average elevation of laser points within each 100m statistical unit - interpolated land elevation of laser points|≤3 , =25m;

[0029] c.Quality flag == 0;

[0030] d. Slope mark < 0.01;

[0031] e. Cloud cover mark <3;

[0032] f.city tag == 0;

[0033] g. Quantity mark ≥ 20;

[0034] h.day and night mark = 1;

[0035] i.|Reference digital elevation model elevation value - laser point interpolation land elevation|≤3 , =1m;

[0036] j.|Average elevation of laser points in each 100m statistical unit - interpolated land elevation of laser points|≤3 , =1m.

[0037] As described above, in the satellite laser altimetry data and stereo image pair joint terrain modeling method, in step S4, the eight constraint criteria are:

[0038] First-level criteria: satisfying screening conditions a and b at the same time;

[0039] Secondary criteria: meet screening conditions a, b, and c at the same time;

[0040] Level 3 criteria: meet screening conditions a, b, c, and d at the same time;

[0041] Level 4 criteria: meet screening conditions a, b, c, d, and e at the same time;

[0042] Level 5 criteria: meet screening conditions a, b, c, d, e, and f at the same time;

[0043] Six-level criteria: meet screening conditions a, b, c, d, e, f, g at the same time;

[0044] Seventh level criteria: meet the screening conditions a, b, c, d, e, f, g, i at the same time;

[0045] Level 8 criteria: satisfy the screening conditions a, b, c, d, e, f, g, i, and j at the same time.

[0046] In the aforementioned method for joint terrain modeling of satellite laser altimetry data and stereo image pairs, in step S5, the original laser point data are filtered in sequence using the first-level to eighth-level criteria. When the laser point density after filtering is in the range of 1 to 10 per square kilometer, the constraint criterion at this time is considered to be a candidate criterion. When there are multiple candidate criteria, the constraint criterion containing the most filtering conditions is selected as the optimal criterion.

[0047] As described above, in the satellite laser altimetry data and stereo image pair joint terrain modeling method, in step S6, the direct linear transformation model error equation is:

[0048]

[0049]

[0050] in, Represents the spatial coordinates of an object point; represents the image plane coordinates of the image point; Represents the undetermined parameters of the direct linear transformation method. The 11 parameters are used to represent the relationship between the object coordinate system and the image coordinate system.

[0051] The beneficial effects of the present invention are:

[0052] In the present invention, firstly, the stereo image pair of the study area is matched with the same name points and pre-processed by free network adjustment to obtain the stereo image after relative orientation; secondly, the satellite laser altimetry data of the study area is obtained, and the h5 file is parsed into a CSV track file of attribute data related to the elevation data of the satellite laser altimetry point; thirdly, the obtained laser altimetry data is screened using a variety of constraint criteria to obtain laser points with high position accuracy, and then the points are used as control points to puncture the stereo image pair and the error equation is solved as constraint conditions. Finally, the DSM is generated by the front intersection of the stereo image pair, and the digital elevation model DEM is generated after filtering, thereby effectively improving the accuracy of the digital elevation model DEM, and providing more accurate basic geographic information products for applications such as land and resources surveys. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a schematic diagram of the overall process of the present invention;

[0054] Figure 2 This is a schematic diagram of a remote sensing image of a Gaofen-7 satellite stereo image pair of a sample area in an embodiment of the present invention;

[0055] Figure 3 A schematic diagram of a digital elevation model DEM without control points generated for a sample area in an embodiment of the present invention;

[0056] Figure 4 This is a laser point distribution map obtained by screening the sample area using different constraint criteria in the embodiment of the present invention;

[0057] Figure 5 A distribution diagram of ICESat-2 control points selected for the sample area in the embodiment of the present invention on a stereo image pair;

[0058] Figure 6 This is a schematic diagram of the digital elevation model DEM generated by adding ICESat-2 control points to the sample area in an embodiment of the present invention. DETAILED DESCRIPTION

[0059] This embodiment provides a satellite laser altimetry data and stereo image pair joint terrain modeling method, such as Figure 1 As shown, the following steps are included:

[0060] S1. For example Figure 2The GF-7 satellite stereo remote sensing image of the sample area shown in the figure is obtained, and the GF-7 satellite stereo remote sensing image of the study area is obtained, and the radiation correction preprocessing is performed on the stereo image pair data; then the stereo image pair data is matched with the same name points and the free network adjustment is calculated to obtain the stereo image of the study area after the image pair is relatively oriented; then the filter is selected to perform dense point cloud matching to generate a point cloud, and the digital elevation model (Digital Elevation Model, DEM) without control points is obtained by performing progressive triangulation filtering on the point cloud. The controlless digital elevation model DEM is shown in Figure 3 shown.

[0061] In step S1, the specific method of radiation correction is: using sensor calibration coefficients to complete radiation calibration, using the radiation transfer model method to perform atmospheric correction on the stereo image pair to eliminate noise.

[0062] The scale-invariant feature transform (SIFT) algorithm is used to match the same-name points. The specific process includes the detection of scale space extreme values, positioning key points and determining feature directions, feature point description, and feature point matching. The equation is shown below:

[0063]

[0064]

[0065]

[0066]

[0067] in, Represents the scale space of the image; Represents the coordinates of the original image; represents a scalable two-dimensional Gaussian function; represents the scale space factor, also called the fuzzy coefficient, The larger the value, the blurrier the image; represents the Difference of Gaussian pyramid; Indicates scale; and They respectively represent the amplitude and argument of each pixel in the neighborhood window where the feature point is located.

[0068] The free network adjustment calculation is performed using the bundle adjustment method (BA), as shown in the following formula:

[0069]

[0070] in, Represents the optimal parameter vector after optimization solution; Represents the parameter vector that needs to be optimized, including the camera's internal orientation elements, external orientation elements, and the three-dimensional coordinates of the object point; Indicates the number of observed image points; is a function, which refers to The reprojection error of the observed image point is usually the difference between the observed image point coordinates and the current estimated parameters. The difference between the calculated image point coordinates.

[0071] S2. Download the ICESat-2 laser altimetry data of the study area from the NASA EARTH DATA platform in the United States and preprocess it. Parse each h5 file of the data into 6 CSV track files, retain 3 strong light track files, and integrate multiple CSV files to obtain a CSV file containing 16 types of attribute data related to the satellite laser altimetry point elevation data.

[0072] The 16 attributes related to the satellite laser altimetry point elevation data are the reference digital elevation model elevation value, the laser point fitting elevation in each 100m statistical unit, the laser point fitting elevation in each 20m statistical unit, the laser point interpolated land elevation, the maximum elevation of the laser point in each 100m statistical unit, the minimum elevation of the laser point in each 100m statistical unit, the median elevation of the laser point in each 100m statistical unit, the average elevation of the laser point in each 100m statistical unit, the number of laser points divided in each 100m statistical unit, the quality mark, the slope mark, the cloud cover mark, the city mark, the quantity mark, the day and night mark, and the standard deviation of the laser point elevation.

[0073] S3. According to the laser altimetry point elevation data and its 16 attributes obtained in step S2, with reference to the data dictionary of the ATL08 product of the ICESat-2 satellite, and based on the characteristics of ICESat-2 data accuracy related to factors such as terrain slope, surface cover and signal-to-noise ratio, 9 attributes related to the quality of satellite laser altimetry point elevation are selected, as shown in Table 1 below. The 9 attributes related to the quality of satellite laser altimetry point elevation are respectively the reference digital elevation model elevation value, the laser point interpolated land elevation, the average elevation of the laser point in each 100m statistical unit, the quality mark, the slope mark, the cloud cover mark, the city mark, the quantity mark and the day and night mark.

[0074] Table 1 ICESat-2 laser point attribute data

[0075]

[0076] These 9 attributes related to the quality of satellite laser altimetry points are used to design 10 screening conditions from a to j. The 10 screening conditions are:

[0077] a.|Reference digital elevation model elevation value - laser point interpolation land elevation|≤3 , =25m;

[0078] b.|Average elevation of laser points within each 100m statistical unit - interpolated land elevation of laser points|≤3 , =25m;

[0079] c.Quality flag == 0;

[0080] d. Slope mark < 0.01;

[0081] e. Cloud cover mark <3;

[0082] f.city tag == 0;

[0083] g. Quantity mark ≥ 20;

[0084] h.day and night mark = 1;

[0085] i.|Reference digital elevation model elevation value - laser point interpolation land elevation|≤3 , =1m;

[0086] j.|Average elevation of laser points in each 100m statistical unit - interpolated land elevation of laser points|≤3 , =1m.

[0087] S4. According to the 10 screening conditions designed in step (3), 8 different levels of constraint criteria are formulated. The 8 constraint criteria are numbered from level 1 to level 8. The complexity of the constraint criteria increases from level 1 to level 8. The 8 constraint criteria are:

[0088] First-level criteria: satisfying screening conditions a and b at the same time;

[0089] Secondary criteria: meet screening conditions a, b, and c at the same time;

[0090] Level 3 criteria: meet screening conditions a, b, c, and d at the same time;

[0091] Level 4 criteria: meet screening conditions a, b, c, d, and e at the same time;

[0092] Level 5 criteria: meet screening conditions a, b, c, d, e, and f at the same time;

[0093] Six-level criteria: meet screening conditions a, b, c, d, e, f, g at the same time;

[0094] Seventh level criteria: meet the screening conditions a, b, c, d, e, f, g, i at the same time;

[0095] Level 8 criteria: satisfy the screening conditions a, b, c, d, e, f, g, i, and j at the same time.

[0096] The first-level criterion can provide a reference for preliminary screening and eliminate laser points with serious elevation errors caused by clouds, aerosols, terrain, etc.; after adding quality marks to the second-level criterion, points that are seriously higher than the reference value threshold can be eliminated, greatly improving the accuracy of laser points; after adding slope marks to the third-level criterion, the elevation accuracy of laser points can be better than 0.1 m; after adding cloud cover marks to the fourth-level criterion, only data with cloud cover less than or equal to 20% are retained; due to the large number of buildings in urban areas, there are more noise points than in non-urban areas, so the fifth-level criterion can be selected when studying terrain problems; the sixth-level criterion adds quantity marks, which can screen out laser points with more than 20 laser points in a 100m statistical unit, and obtain laser points with stable signal strength and belonging to the ground; the seventh-level criterion adds day and night marks to only retain night data, which will lose 50% of the data, but can improve data accuracy. If the data volume is sufficient, the seventh-level criterion can be selected; the eighth-level criterion sets the absolute elevation accuracy of the reference digital elevation model DEM to 1m, which can further retain higher-precision laser points. According to the topography of the experimental sample area, the amount of data, and the experimental requirements, the quality and quantity of the laser points were screened out according to different criteria for comparison, and appropriate screening conditions were selected.

[0097] In this embodiment, in the laser altimetry point screening results obtained by using 8 constraint criteria, there are a total of 15279 ICESat-2 original laser points, 7831 points are obtained after screening according to the first-level criteria; 826 points are obtained after screening according to the second-level criteria; 561 points are obtained after screening according to the third-level criteria; 539 points are obtained after screening according to the fourth-level criteria; 514 points are obtained after screening according to the fifth-level criteria; 301 points are obtained after screening according to the sixth-level criteria; 91 points are obtained after screening according to the seventh-level criteria; and 88 points are obtained after screening according to the eighth-level criteria.

[0098] S5. Use the first-level to eighth-level criteria to filter the original laser point data in sequence. The higher the level of constraint criteria, the fewer laser points will be filtered out. According to the size of the experimental sample area or the task requirements, select the optimal constraint criterion as the final standard to obtain the filtered laser point data.

[0099] The specific method for selecting the optimal constraint criterion is: use the first-level to eighth-level criteria in sequence to filter the original laser point data. When the density of laser points after filtering is in the range of 1 to 10 per square kilometer, the constraint criterion at this time is considered to be a candidate criterion. When there are multiple candidate criteria, the constraint criterion containing the most filtering conditions is selected as the optimal criterion.

[0100] After the first-level criterion gross error screening, the retention rate was about 50%; after the second-level criterion added quality marks, the laser point retention rate dropped sharply, but the quality of the laser points was greatly improved; after adding subsequent criteria, the laser point retention rate slowly decreased, and the quality of the laser points gradually improved; after the seventh-level criterion added day and night marks, the laser points were reduced by nearly 70%, which did not meet the number of control points required by the sample area of ​​this embodiment. Therefore, combined with the terrain and mission requirements of the sample area of ​​the embodiment, criterion 6 was selected as the constraint criterion for ICESat-2 laser points, and finally 301 high-quality laser points were obtained; the point distribution obtained by screening with different constraint criteria is as follows Figure 4 shown.

[0101] S6, 30 of the 301 laser points obtained in step S5 are uniformly selected as control points, and the remaining 271 points are used as check points. The distribution diagram of the selected ICESat-2 control points on the stereo image pair is shown in FIG. Figure 5 As shown, it is brought into the direct linear transformation (DLT) compensation equation as a constraint condition to solve the error equation; then the stereo imaging geometric model is updated to obtain the spatial coordinates of the connection points with improved accuracy.

[0102] The error equation of the direct linear transformation model is:

[0103]

[0104]

[0105] in, Represents the spatial coordinates of an object point; represents the image plane coordinates of the image point; Represents the undetermined parameters of the direct linear transformation method. The 11 parameters are used to represent the relationship between the object coordinate system and the image coordinate system.

[0106] S7. Obtain the coordinates of the encrypted points by the intersection of the stereo image pairs in front, select the type of digital terrain model (DTM) and the filter for the connection points, and perform dense point cloud matching to generate a digital surface model (DSM). The final digital elevation model (DEM) is obtained by performing progressive triangulation filtering on the DSM. The generated digital elevation model DEM is as follows: Figure 6 shown.

[0107] For the final terrain modeling results, the root mean square error (RMSE) is used for quantitative evaluation, and the formula is:

[0108]

[0109] Where H represents the elevation value of the checkpoint on the generated final digital elevation model DEM; h represents the elevation value of the ICESat-2 satellite laser altimetry data of the checkpoint; and n represents the number of checkpoints.

[0110] The quantitative evaluation results are shown in Table 2 below. Without control points, the vertical accuracy of the digital elevation model DEM generated by GF-7 is 7.86m; after adding ICESat-2 control points, the accuracy is improved to 5.31m, which means that the present invention has high improvement and practicality.

[0111] Table 2 Accuracy evaluation of terrain modeling results

[0112]

[0113] The method of this embodiment first performs homonymous point matching and free network adjustment preprocessing on the stereo image pair of the study area to obtain a relatively oriented stereo image; secondly, obtain the satellite laser altimetry data of the study area, and parse the h5 file into a CSV track file of attribute data related to the satellite laser altimetry point elevation data; thirdly, the obtained laser altimetry data is screened using a variety of constraint criteria to obtain laser points with high position accuracy, and then the points are used as control points to puncture the stereo image pair and used as constraint conditions to solve the error equation. Finally, a DSM is generated by the front intersection of the stereo image pair, and a digital elevation model DEM is generated after filtering, thereby effectively improving the accuracy of the digital elevation model DEM, and providing more accurate basic geographic information products for applications such as land and resources surveys.

[0114] In addition to the above embodiments, the present invention may also have other implementation modes. Any technical solution formed by equivalent replacement or equivalent transformation falls within the protection scope required by the present invention.

Claims

1. A method for joint terrain modeling using satellite laser altimetry data and stereo image pairs, characterized in that: The following steps are involved: S1, obtain the stereo image pair remote sensing image of the study area, and perform radiation correction preprocessing on it; then perform homonymous point matching and free network adjustment calculation on the stereo image pair data to obtain the stereo image of the study area after relative orientation; then select the filter to perform dense point cloud matching to generate point cloud, and obtain the digital elevation model without control points by performing progressive triangulation filtering on the point cloud; S2, obtain the satellite laser altimetry data of the study area and preprocess it, parse each h5 file of the data into a CSV track file, retain the strong light track file, integrate multiple CSV files, and obtain a CSV file containing attribute data related to the satellite laser altimetry point elevation data; S3, the laser altimetry point elevation data and its attributes obtained in step S2 include 9 attributes related to the satellite laser altimetry point elevation quality, and 10 screening conditions from a to j are designed using the 9 attributes related to the satellite laser altimetry point elevation quality respectively; S4. Formulate eight different levels of constraint criteria based on the ten screening conditions designed in step (3), and number the eight constraint criteria from level one to level eight, with the complexity of the constraint criteria increasing in sequence from level one to level eight; S5. Use the first-level to eighth-level criteria to filter the original laser point data in sequence. The higher the level of constraint criteria, the fewer laser points will be filtered out. According to the size of the experimental sample area or the task requirements, the optimal constraint criteria is selected as the final standard to obtain the filtered laser point data. S6, puncturing the control points obtained in step S5 on the stereo image pair, and bringing the control point data into the direct linear transformation model compensation equation as a constraint condition to solve the error equation; then updating the stereo image imaging geometric model to obtain the spatial coordinates of the connection points after the accuracy is improved; S7, obtaining the coordinates of the encrypted points by the front intersection of the stereo image pair, then using progressive triangulation filtering to obtain the ground points, and finally generating a digital elevation model; In step S1, the scale-invariant feature transformation algorithm is used to match the same-name points. The specific process includes the detection of scale space extreme values, positioning key points and determining feature directions, feature point description, and feature point matching. The equation is shown in the following formula: in, Represents the scale space of the image; Represents the coordinates of the original image; represents a scalable two-dimensional Gaussian function; represents the scale space factor; represents the Difference of Gaussian pyramid; Indicates scale; and They respectively represent the amplitude and argument of each pixel in the neighborhood window where the feature point is located.

2. The method for joint terrain modeling using satellite laser altimetry data and stereo image pairs according to claim 1, characterized in that: In step S1, the free network adjustment calculation is performed using the bundle adjustment method, as shown in the following formula: in, Represents the optimal parameter vector after optimization solution; Represents the parameter vector that needs to be optimized, including the camera's internal orientation elements, external orientation elements, and the three-dimensional coordinates of the object point; Indicates the number of observed image points; is a function, which refers to The reprojection error of each observed image point.

3. The method for joint terrain modeling using satellite laser altimetry data and stereo image pairs according to claim 1, characterized in that: In step S1, the specific method of radiation correction is: using sensor calibration coefficients to complete radiation calibration, using the radiation transfer model method to perform atmospheric correction on the stereo image pair to eliminate noise.

4. The method for joint terrain modeling using satellite laser altimetry data and stereo image pairs according to claim 1, characterized in that: In the step S2, the attributes related to the satellite laser altimetry point elevation data are set to 16 types, namely, the reference digital elevation model elevation value, the laser point fitting elevation in each 100m statistical unit, the laser point fitting elevation in each 20m statistical unit, the laser point interpolation land elevation, the maximum elevation of the laser point in each 100m statistical unit, the minimum elevation of the laser point in each 100m statistical unit, the median of the laser point elevation in each 100m statistical unit, the average elevation of the laser point in each 100m statistical unit, the number of laser points divided in each 100m statistical unit, the quality mark, the slope mark, the cloud cover mark, the city mark, the quantity mark, the day and night mark, and the standard deviation of the laser point elevation.

5. The method for joint terrain modeling using satellite laser altimetry data and stereo image pairs according to claim 1, characterized in that: In step S3, the nine attributes related to the quality of the satellite laser altimetry point elevation are the reference digital elevation model elevation value, the laser point interpolated land elevation, the average elevation of the laser point in each 100m statistical unit, the quality mark, the slope mark, the cloud cover mark, the city mark, the quantity mark and the day and night mark.

6. The method for joint terrain modeling using satellite laser altimetry data and stereo image pairs according to claim 5, characterized in that: In step S3, the 10 screening conditions are: a.|Reference digital elevation model elevation value - laser point interpolation land elevation|≤3 , =25m; b.|Average elevation of laser points within each 100m statistical unit - interpolated land elevation of laser points|≤3 , =25m; c.Quality flag == 0; d. Slope mark < 0.01; e. Cloud cover mark <3; f.city mark == 0; g. Quantity mark ≥ 20; h.day and night mark = 1; i.|Reference digital elevation model elevation value - laser point interpolation land elevation|≤3 , =1m; j.|Average elevation of laser points in each 100m statistical unit - interpolated land elevation of laser points|≤3 , =1m.

7. The method for joint terrain modeling using satellite laser altimetry data and stereo image pairs according to claim 1, characterized in that: In step S4, the eight constraint criteria are: First-level criteria: satisfying screening conditions a and b at the same time; Secondary criteria: meet screening conditions a, b, and c at the same time; Level 3 criteria: meet screening conditions a, b, c, and d at the same time; Level 4 criteria: meet screening conditions a, b, c, d, and e at the same time; Level 5 criteria: meet screening conditions a, b, c, d, e, and f at the same time; Six-level criteria: meet screening conditions a, b, c, d, e, f, g at the same time; Seventh level criteria: meet the screening conditions a, b, c, d, e, f, g, i at the same time; Level 8 criteria: satisfy the screening conditions a, b, c, d, e, f, g, i, and j at the same time.

8. The method for joint terrain modeling using satellite laser altimetry data and stereo image pairs according to claim 1, characterized in that: In step S5, the original laser point data are filtered using the first-level to eighth-level criteria in sequence. When the laser point density after filtering is in the range of 1 to 10 per square kilometer, the constraint criterion at this time is considered to be a candidate criterion. When there are multiple candidate criteria, the constraint criterion containing the most filtering conditions is selected as the optimal criterion.

9. The method for joint terrain modeling using satellite laser altimetry data and stereo image pairs according to claim 1, characterized in that: In step S6, the direct linear transformation model error equation is: in, Represents the spatial coordinates of an object point; represents the image plane coordinates of the image point; Represents the undetermined parameters of the direct linear transformation method. The 11 parameters are used to represent the relationship between the object coordinate system and the image coordinate system.