Method and system for constructing water-land integrated digital elevation model
Through multi-source data fusion processing and feature point splicing, the problems of inconsistent coordinates and data discontinuity in the construction of lake water-land integrated digital elevation model are solved, and high-precision three-dimensional visual expression and interactive analysis of lake water-land integrated integrated digital elevation model are realized, which improves data acquisition and analytical efficiency.
Patent Information
- Application Number
- CN202410979337.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-11-15
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing integrated digital elevation model construction method of lakes is inefficient, the coordinates on and underwater are not uniform, and the data accuracy is inconsistent, making it difficult to realize intuitive three-dimensional visual expression and interactive analysis, especially in areas such as steep cliffs and mud along the coast, and lacks variety. data visualization method.
The multi-source data fusion method is adopted, including coordinate changes and geometric registration pre-processing of MODIS terrain data, LiDAR terrain data and multi-beam sounding data, to construct a lake land error prediction model, and generate water error values and regression analysis through weighted fusion and regression analysis. Feature weight diagram, determine feature points for splicing, and realize the integrated scene construction of lake water and land.
It realizes three-dimensional visual expression of lakes integrated water and land, improves data accuracy and splicing efficiency, helps researchers to conduct lake research and data analysis more conveniently, shortens the time period for on-site terrain measurement, and ensures data integrity and high accuracy.
Smart Images

Figure CN118968289A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of digital elevation models, and more particularly to a method and system for constructing a land-water integrated digital elevation model. Background Art
[0002] As the digital reform of lakes is further promoted, smart lakes will be a new task for building high-quality development of lakes, and new requirements will be put forward for lake management decision-making and display methods. As an important new infrastructure, real-scene 3D urgently needs to integrate the underwater terrain of the nested water space to serve the goals of urban space planning, water space governance and control. At present, the construction of the integrated water and land digital elevation model of lakes is mainly divided into two independent projects, one on the water and the other underwater. The water part is mostly manually run using real-time dynamic RTK technology or using 3D laser scanning technology, aerial / sky photogrammetry technology, and drone remote sensing technology to obtain data; while underwater projects generally use single-beam echo sounders, multi-beam echo sounders, etc. to collect water depth data. This method has low operating efficiency, inconsistent coordinate benchmarks on the water and underwater, and inconsistent data accuracy. People cannot reach difficult areas such as cliffs, steep slopes and muddy areas along the coast, resulting in discontinuous water and underwater data and incomplete terrain data.
[0003] At present, there are many models of integrated water and land construction. For example, Zhang Xiaoqing et al. converted underwater terrain model data into interpolated point cloud data, and then used Context Capture and other software to perform integrated fusion modeling of water and underwater; Xu Rui, Zhou Yang et al. integrated terrain, tilt model and other data to develop a three-dimensional digital reservoir-land geographic information platform; Yu Dajie et al. combined the waterway data collected by rapid surveying and mapping methods to realize spatial analysis and waterway simulation functions based on three-dimensional surveying and mapping data. However, these three-dimensional models of integrated water and land systems lack a variety of data visualization methods, making it difficult to intuitively and vividly show the spatial and temporal distribution of data, and the interactive analysis of the basin is insufficient, and the water body information data is not fully mined.
[0004] Therefore, it is very necessary to construct a digital elevation model of lakes and land. This model integrates multi-source data, constructs an integrated scene of the lake's underwater topography and land, and realizes a three-dimensional visual expression of water and land. It can obtain lake data information more intuitively and quickly, help researchers conduct lake research more conveniently, assist in field measurement and inspection data, analyze lake information data, and explore the characteristics of lake underwater data information from many aspects. Summary of the invention
[0005] In view of this, the present invention provides a method and system for constructing a land and water integrated digital elevation model to solve the problems in the background technology.
[0006] In order to achieve the above object, the present invention adopts the following technical solution:
[0007] In one aspect, the present invention discloses a method for constructing a land-water integrated digital elevation model, comprising:
[0008] Collect multi-source data of the lake area and surrounding land: MODIS terrain data, LiDAR terrain data, and multi-beam bathymetric data;
[0009] Based on the basic surveying and mapping data, coordinate transformation and geometric registration preprocessing are performed on MODIS terrain data, LiDAR terrain data and multi-beam bathymetric data;
[0010] Taking MODIS terrain data and LiDAR terrain data as training samples, respectively constructing lake and land error prediction models to obtain land deviation values and feature weight values; using the land deviation values and feature weight values to perform weighted fusion on the MODIS terrain data and the LiDAR terrain data to obtain lake and land terrain data;
[0011] MODIS terrain data and multi-beam bathymetry data, LiDAR terrain data and multi-beam bathymetry data are superimposed and compared, the overlapping area between the two sets of data is determined, and the overlapping parts in the data are respectively clipped to obtain the overlapping area of MODIS terrain data and the overlapping area of LiDAR terrain data; a regression equation is generated for the overlapping area of MODIS terrain data and the overlapping area of LiDAR terrain data, and then two sets of lake bottom terrain data of the entire lake area are calculated for difference calculation to obtain a water area error value and a water area feature weight map; lake water area terrain data is obtained through the water area error value and the water area feature weight map;
[0012] Determine a stitching area according to the lake land terrain data and the lake water terrain data, and mark a first feature point and a second feature point in the stitching area respectively; in the stitching area, select a point whose distance from the first feature point and the second feature point meets a preset requirement as a reference point; correct the lake land terrain data in a manner of correcting the first feature point to the reference point to determine a lake land terrain image, and correct the lake water terrain data in a manner of correcting the second feature point to the reference point to determine a lake water terrain image; stitch the lake land terrain image and the lake water terrain image.
[0013] Preferably, in the above-mentioned method for constructing a land-water integrated digital elevation model, the preprocessing step comprises:
[0014] Coordinate transformation between terrain data: Taking the coordinate system of basic surveying and mapping data as the spatial reference, corresponding coordinate transformation is performed on MODIS terrain data, LiDAR terrain data, and multi-beam bathymetric data to achieve coordinate unification of multi-source data;
[0015] Data geometric registration: Using basic surveying and mapping data as the reference image, a certain number of ground features with the same name are selected as control points in the corresponding image area, and the MODIS terrain data, LiDAR terrain data, and multi-beam bathymetric data are geometrically registered using the polynomial correction method.
[0016] Preferably, in the above-mentioned method for constructing a land-water integrated digital elevation model, constructing a lake-land error prediction model comprises:
[0017] The LiDAR terrain data uses a first error prediction model to obtain a first elevation deviation value and a first feature weight map; the MODIS terrain data uses a second error prediction model to obtain a second elevation deviation value and a second feature weight map;
[0018] The first elevation deviation value is used to correct the LiDAR terrain data to obtain a radar image to be fused; the second elevation deviation value is used to correct the MODIS terrain data to obtain a remote sensing image to be fused;
[0019] The radar image to be fused and the remote sensing image to be fused are weightedly fused according to the first feature weight map and the second feature weight map to obtain fused elevation data.
[0020] Preferably, in the above-mentioned land-water integrated digital elevation model construction method, the first error prediction model and the second error prediction model training steps include:
[0021] Collect MODIS terrain data, LiDAR terrain data and their corresponding measured error samples as training data;
[0022] Construct a deep neural network model, take the elevation data as input, and predict the corresponding error value, wherein the deep neural network model includes multiple hidden layers and uses a negatively correlated mapping function G to learn the complex relationship between data and error;
[0023] Use training data to train the model and optimize network parameters to minimize the difference between the predicted error and the measured error;
[0024] The trained model can be used to predict elevation deviation values or feature weight maps.
[0025] Preferably, in the above-mentioned method for constructing an integrated land and water digital elevation model, the calculation weights of the first feature weight map and the second feature weight map are calculated using the inverse proportion of the squared elevation error or from the normalized residual.
[0026] Preferably, in the above-mentioned method for constructing a land-water integrated digital elevation model, the water area error value and the water area feature weight value are specifically obtained in the following steps:
[0027] According to the obtained overlapping area, binary regression analysis is performed with the coordinate values (x, y) as independent variables to determine the binary polynomial regression equations between the two types of data and evaluate the accuracy of the regression equations;
[0028] Based on the determined bivariate polynomial regression equation, two sets of lake bottom topography are calculated for the entire lake area at the spatial resolution of the multibeam bathymetric data;
[0029] The generated lake bottom topography is subtracted from the basic surveying and mapping data to obtain the water area error value and water area characteristic weight map.
[0030] Preferably, in the above-mentioned method for constructing a land-water integrated digital elevation model, the lake land terrain image and the lake water terrain image are spliced, and the method further includes:
[0031] Determine that a specified pixel point in the lake land terrain image is a feature point A, and image features of the feature point A and image features of pixels near the feature point A meet a preset difference condition;
[0032] Determine whether the image feature of feature point B and the image feature of pixel points near feature point B meet a preset difference condition, wherein feature point B is a pixel point corresponding to feature A in the lake water area topography image;
[0033] If so, it is determined that the feature point A is the first feature point and the feature point B is the second feature point.
[0034] On the other hand, the present invention discloses a system for constructing a land-water integrated digital elevation model, which applies the above method and includes:
[0035] The acquisition module collects multi-source data of the lake area and surrounding land: MODIS terrain data, LiDAR terrain data, and multi-beam bathymetric data;
[0036] The preprocessing module uses the basic surveying and mapping data as a benchmark to perform coordinate transformation and geometric registration preprocessing on MODIS terrain data, LiDAR terrain data, and multi-beam bathymetric data;
[0037] The land terrain data acquisition module uses MODIS terrain data and LiDAR terrain data as training samples to construct lake and land error prediction models respectively, and obtain land deviation values and feature weight values; and uses the land deviation values and feature weight values to perform weighted fusion on the MODIS terrain data and the LiDAR terrain data to obtain lake and land terrain data;
[0038] The water area terrain data acquisition module is used to superimpose and compare MODIS terrain data with multi-beam bathymetry data, LiDAR terrain data with multi-beam bathymetry data, determine the overlapping area between the two sets of data, and respectively clip the overlapping parts in the data to obtain the MODIS terrain data overlapping area and the LiDAR terrain data overlapping area; generate a regression equation for the MODIS terrain data overlapping area and the LiDAR terrain data overlapping area, and then calculate the difference between the two sets of lake bottom terrain data of the entire lake area to obtain the water area error value and the water area feature weight map; obtain the lake water area terrain data through the water area error value and the water area feature weight map;
[0039] A stitching module determines a stitching area according to the lake land terrain data and the lake water terrain data, and marks a first feature point and a second feature point in the stitching area respectively; in the stitching area, selects a point whose distance from the first feature point and the second feature point meets a preset requirement as a reference point; corrects the lake land terrain data in a manner of correcting the first feature point to the reference point to determine a lake land terrain image, and corrects the lake water terrain data in a manner of correcting the second feature point to the reference point to determine a lake water terrain image; and stitches the lake land terrain image and the lake water terrain image.
[0040] Preferably, in the above-mentioned land and water integrated digital elevation model construction system, the land terrain data acquisition module includes:
[0041] A first prediction unit, wherein the LiDAR terrain data uses a first error prediction model to obtain a first elevation deviation value and a first feature weight map;
[0042] The second prediction unit uses the second error prediction model to obtain the second elevation deviation value and the second feature weight map of the MODIS terrain data;
[0043] a first correction unit, wherein the first elevation deviation value corrects the LiDAR terrain data to obtain a radar image to be fused;
[0044] A second correction unit, wherein the second elevation deviation value corrects the MODIS terrain data to obtain a remote sensing image to be fused;
[0045] A fusion unit performs weighted fusion on the radar image to be fused and the remote sensing image to be fused according to the first feature weight map and the second feature weight map to obtain fused elevation data.
[0046] Preferably, in the above-mentioned integrated land and water digital elevation model construction system, the water area terrain data acquisition module includes:
[0047] The regression unit, based on the obtained overlapping area, performs binary regression analysis with the coordinate values (x, y) as independent variables, determines the binary polynomial regression equations between the two types of data and evaluates the accuracy of the regression equations;
[0048] A calculation unit calculates two sets of lake bottom topography for the entire lake area at the spatial resolution of the multibeam bathymetric data according to the determined bivariate polynomial regression equation;
[0049] The difference unit calculates the difference between the generated lake bottom topography and the basic surveying and mapping data to obtain the water area error value and water area characteristic weight map.
[0050] Through the above technical solutions, it can be known that compared with the prior art, the present invention discloses a method and system for constructing a water-land integrated digital elevation model, which integrates and processes multi-source data, constructs an integrated scene of the lake underwater terrain and land, and realizes a three-dimensional visual expression of water-land integration, which can obtain lake data information more intuitively and quickly, help researchers to conduct lake research more conveniently, assist in field measurement and inspection data, analyze lake information data, and explore the characteristics of lake bottom data information from many aspects. By utilizing multi-source data such as basic surveying and mapping data, optical remote sensing image data, LiDAR data, and multi-beam bathymetric data, the advantages of each type of data are fully utilized, and their shortcomings are avoided. Correlation analysis is performed on the data, and lake land terrain data and lake water terrain data are obtained respectively, and corrections are made with reference points to achieve splicing, and smoothing processing is considered at the seams between images, so as to achieve the acquisition of the complete terrain of the lake, that is, the time period for field measurement of lake terrain is effectively shortened, and the integrity and high precision of lake terrain data are guaranteed, and the production efficiency of lake bottom terrain data with a large area is optimized. In addition, two trained error prediction models are used to predict the errors of the two types of data to be fused respectively, and the two types of data to be fused are corrected according to the two predicted deviation values to make them more accurate. The two types of elevation data to be fused are weightedly fused according to the two weight maps generated in the error prediction process, thereby making up for the differences between different elevation data and improving the accuracy of the fused elevation data. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.
[0052] Figure 1 A flow chart of the method provided by the present invention;
[0053] Figure 2This is a schematic diagram of the splicing provided by the present invention. DETAILED DESCRIPTION
[0054] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0055] The embodiment of the present invention discloses a method for constructing a land-water integrated digital elevation model, such as Figure 1 As shown, including:
[0056] Collect multi-source data of the lake area and surrounding land: MODIS terrain data, LiDAR terrain data, and multi-beam bathymetric data;
[0057] The MODIS image used in the present invention has a temporal resolution of 1-2 days and a scanning width of 2330 km. It has a high temporal resolution and eliminates images that cannot be used due to cloud coverage of the target area or ghosting and distortion of target objects caused by shooting angles.
[0058] The data used in this step are multi-view MODIS 500M resolution remote sensing images within a seasonal variation cycle, daily measured water level data of multiple hydrological stations in the lake area, annual precipitation data in the basin, and historical lake bottom topography data of three cross sections of the lake.
[0059] Due to the different ways of acquiring multivariate data, there are some differences in the coordinate system settings between the data, and the deformation and offset caused by the limitation of data accuracy. In order to facilitate the subsequent calculation and analysis between multi-source data, the data is preprocessed, including: coordinate transformation and geometric registration between data;
[0060] Based on the basic surveying and mapping data, coordinate transformation and geometric registration preprocessing are performed on MODIS terrain data, LiDAR terrain data and multi-beam bathymetric data;
[0061] Taking MODIS terrain data and LiDAR terrain data as training samples, respectively constructing lake and land error prediction models to obtain land deviation values and feature weight values; using the land deviation values and feature weight values to perform weighted fusion on the MODIS terrain data and the LiDAR terrain data to obtain lake and land terrain data;
[0062] MODIS terrain data and multi-beam bathymetry data, LiDAR terrain data and multi-beam bathymetry data are superimposed and compared, the overlapping area between the two sets of data is determined, and the overlapping parts in the data are respectively clipped to obtain the overlapping area of MODIS terrain data and the overlapping area of LiDAR terrain data; a regression equation is generated for the overlapping area of MODIS terrain data and the overlapping area of LiDAR terrain data, and then two sets of lake bottom terrain data of the entire lake area are calculated for difference calculation to obtain a water area error value and a water area feature weight map; lake water area terrain data is obtained through the water area error value and the water area feature weight map;
[0063] In the prior art, the common method for image stitching is to use digital image processing methods to first perform histogram stretching on adjacent images, and then perform histogram matching to make the grayscale distribution of the overlapping area of the two images roughly the same. However, the relevant technology only solves the problem of local image deformation, but cannot solve the problem of overall correction of the overlapping area.
[0064] The reason is that if the difference in radiometry between adjacent images is not processed for image mosaicking (image stitching), artificial false boundaries will often be generated at the stitching line. This false boundary will make image interpretation difficult and misleading, and also affect the overall effect of the image. Therefore, the radiometry of the image to be fused must be corrected before fusion.
[0065] like Figure 2 As shown, a stitching area is determined according to the lake land terrain data and the lake water terrain data, and a first feature point and a second feature point are respectively marked in the stitching area; in the stitching area, a point whose distance from the first feature point and the second feature point meets the preset requirements is selected as a reference point; the lake land terrain data is corrected in a manner of correcting the first feature point to the reference point to determine a lake land terrain image, and the lake water terrain data is corrected in a manner of correcting the second feature point to the reference point to determine a lake water terrain image; the lake land terrain image and the lake water terrain image are stitched.
[0066] Further, selecting a point whose distance from the first feature point and the second feature point meets a preset requirement as a reference point includes: selecting a point on the perpendicular bisector of a line connecting the first feature point and the second feature point as a reference point.
[0067] Furthermore, there are multiple first feature points and multiple second feature points, and the difference between the specified parameter value of each first feature point and the specified parameter value of the corresponding second feature point is less than a preset threshold. That is, each pair of first feature points and second feature points should have corresponding features, such as similar brightness, similar contrast, etc., that is, when the difference between the specified parameters of the two corresponding points with the same name should be less than a certain range, it can be proved that the two feature points are a pair of corresponding feature points.
[0068] In order to further optimize the above technical solution, the pretreatment step includes:
[0069] Coordinate transformation between terrain data: Taking the coordinate system of basic surveying and mapping data as the spatial reference, corresponding coordinate transformation is performed on MODIS terrain data, LiDAR terrain data, and multi-beam bathymetric data to achieve coordinate unification of multi-source data;
[0070] Data geometric registration: Using basic surveying and mapping data as the reference image, a certain number of ground features with the same name are selected as control points in the corresponding image area, and the MODIS terrain data, LiDAR terrain data, and multi-beam bathymetric data are geometrically registered using the polynomial correction method.
[0071] In order to further optimize the above technical solutions, the lake land error prediction model is constructed including:
[0072] The LiDAR terrain data uses a first error prediction model to obtain a first elevation deviation value and a first feature weight map;
[0073] The MODIS terrain data uses a second error prediction model to obtain a second elevation deviation value and a second feature weight map;
[0074] The first elevation deviation value is used to correct the LiDAR terrain data to obtain a radar image to be fused; the second elevation deviation value is used to correct the MODIS terrain data to obtain a remote sensing image to be fused;
[0075] The radar image to be fused and the remote sensing image to be fused are weightedly fused according to the first feature weight map and the second feature weight map to obtain fused elevation data.
[0076] Specifically, the first elevation deviation value and the LiDAR terrain data are added to obtain the radar image to be fused;
[0077] The second elevation deviation value and the MODIS terrain data are added to obtain the remote sensing image to be fused.
[0078] Furthermore, weighted fusion refers to the fusion of elevation data X = W 雷达 ⊙X 雷达 +W遥感 ⊙X 遥感;
[0079] Where X is the fused elevation data, X 雷达 is the radar image with fusion; W 雷达 is the radar image weight value; X 遥感 is a fused remote sensing image; W 遥感 is the weight value of the remote sensing image;
[0080] In order to further optimize the above technical solution, the first error prediction model and the second error prediction model training steps include:
[0081] Collect MODIS terrain data, LiDAR terrain data and their corresponding measured error samples as training data;
[0082] Construct a deep neural network model, take the elevation data as input, and predict the corresponding error value, wherein the deep neural network model includes multiple hidden layers and uses a negatively correlated mapping function G to learn the complex relationship between data and error;
[0083] Use training data to train the model and optimize network parameters to minimize the difference between the predicted error and the measured error;
[0084] The trained model can be used to predict elevation deviation values or feature weight maps.
[0085] In order to further optimize the above technical solution, the calculation weights of the first feature weight map and the second feature weight map adopt the inverse proportion of the squared elevation error or calculate the weights from the normalized residual.
[0086] In order to further optimize the above technical solution, the specific steps of water area error value and water area characteristic weight value are as follows:
[0087] According to the obtained overlapping area, binary regression analysis is performed with the coordinate values (x, y) as independent variables to determine the binary polynomial regression equations between the two types of data and evaluate the accuracy of the regression equations;
[0088] Based on the determined bivariate polynomial regression equation, two sets of lake bottom topography are calculated for the entire lake area at the spatial resolution of the multibeam bathymetric data;
[0089] The generated lake bottom topography is subtracted from the basic surveying and mapping data to obtain the water area error value and water area characteristic weight map.
[0090] In order to further optimize the above technical solution, the lake land terrain image and the lake water terrain image are spliced, and the following is also included:
[0091] Determine that a specified pixel point in the lake land terrain image is a feature point A, and image features of the feature point A and image features of pixels near the feature point A meet a preset difference condition;
[0092] Determine whether the image feature of feature point B and the image feature of pixel points near feature point B meet a preset difference condition, wherein feature point B is a pixel point corresponding to feature A in the lake water area topography image;
[0093] If so, it is determined that the feature point A is the first feature point and the feature point B is the second feature point.
[0094] Before determining the designated pixel point in the lake land terrain image as the feature point A, the method further includes: using a color balance method to adjust the radiometric difference between the lake land terrain image and the lake water terrain image.
[0095] The lake land terrain image and the lake water terrain image are adjusted by using a histogram adjustment method so that the cumulative histogram of the designated features of the lake land terrain image is consistent with the cumulative histogram of the lake water terrain image.
[0096] Another embodiment of the present invention discloses a system for constructing a land-water integrated digital elevation model, which applies the above method and includes:
[0097] The acquisition module collects multi-source data of the lake area and surrounding land: MODIS terrain data, LiDAR terrain data, and multi-beam bathymetric data;
[0098] The preprocessing module uses the basic surveying and mapping data as a benchmark to perform coordinate transformation and geometric registration preprocessing on MODIS terrain data, LiDAR terrain data, and multi-beam bathymetric data;
[0099] The land terrain data acquisition module uses MODIS terrain data and LiDAR terrain data as training samples to construct lake and land error prediction models respectively, and obtain land deviation values and feature weight values; and uses the land deviation values and feature weight values to perform weighted fusion on the MODIS terrain data and the LiDAR terrain data to obtain lake and land terrain data;
[0100] The water area terrain data acquisition module is used to superimpose and compare MODIS terrain data with multi-beam bathymetry data, LiDAR terrain data with multi-beam bathymetry data, determine the overlapping area between the two sets of data, and respectively clip the overlapping parts in the data to obtain the MODIS terrain data overlapping area and the LiDAR terrain data overlapping area; generate a regression equation for the MODIS terrain data overlapping area and the LiDAR terrain data overlapping area, and then calculate the difference between the two sets of lake bottom terrain data of the entire lake area to obtain the water area error value and the water area feature weight map; obtain the lake water area terrain data through the water area error value and the water area feature weight map;
[0101] A stitching module determines a stitching area according to the lake land terrain data and the lake water terrain data, and marks a first feature point and a second feature point in the stitching area respectively; in the stitching area, selects a point whose distance from the first feature point and the second feature point meets a preset requirement as a reference point; corrects the lake land terrain data in a manner of correcting the first feature point to the reference point to determine a lake land terrain image, and corrects the lake water terrain data in a manner of correcting the second feature point to the reference point to determine a lake water terrain image; and stitches the lake land terrain image and the lake water terrain image.
[0102] In order to further optimize the above technical solution, the land terrain data acquisition module includes:
[0103] A first prediction unit, wherein the LiDAR terrain data uses a first error prediction model to obtain a first elevation deviation value and a first feature weight map;
[0104] The second prediction unit uses the second error prediction model to obtain the second elevation deviation value and the second feature weight map of the MODIS terrain data;
[0105] a first correction unit, wherein the first elevation deviation value corrects the LiDAR terrain data to obtain a radar image to be fused;
[0106] A second correction unit, wherein the second elevation deviation value corrects the MODIS terrain data to obtain a remote sensing image to be fused;
[0107] A fusion unit performs weighted fusion on the radar image to be fused and the remote sensing image to be fused according to the first feature weight map and the second feature weight map to obtain fused elevation data.
[0108] In order to further optimize the above technical solution, the water terrain data acquisition module includes:
[0109] The regression unit, based on the obtained overlapping area, performs binary regression analysis with the coordinate values (x, y) as independent variables, determines the binary polynomial regression equations between the two types of data and evaluates the accuracy of the regression equations;
[0110] A calculation unit calculates two sets of lake bottom topography for the entire lake area at the spatial resolution of the multibeam bathymetric data according to the determined bivariate polynomial regression equation;
[0111] The difference unit calculates the difference between the generated lake bottom topography and the basic surveying and mapping data to obtain the water area error value and water area characteristic weight map.
[0112] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.
[0113] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for constructing a land-water integrated digital elevation model, characterized in that: include: Collect multi-source data of the lake area and surrounding land: MODIS terrain data, LiDAR terrain data, and multi-beam bathymetric data; Based on the basic surveying and mapping data, coordinate transformation and geometric registration preprocessing are performed on MODIS terrain data, LiDAR terrain data and multi-beam bathymetric data; Taking MODIS terrain data and LiDAR terrain data as training samples, respectively constructing lake and land error prediction models to obtain land deviation values and feature weight values; using the land deviation values and feature weight values to perform weighted fusion on the MODIS terrain data and the LiDAR terrain data to obtain lake and land terrain data; MODIS terrain data and multi-beam bathymetry data, LiDAR terrain data and multi-beam bathymetry data are superimposed and compared to determine the overlapping area between the two sets of data, and the overlapping parts of the data are clipped to obtain the overlapping area of MODIS terrain data and the overlapping area of LiDAR terrain data; Generate a regression equation for the overlapping area of the MODIS terrain data and the overlapping area of the LiDAR terrain data, and then calculate the difference between the two sets of lake bottom terrain data of the entire lake area to obtain a water area error value and a water area feature weight map; Obtain lake water area topography data through the water area error value and the water area characteristic weight map; Determine a splicing area according to the lake land terrain data and the lake water terrain data, and mark a first feature point and a second feature point in the splicing area respectively; In the stitching area, a point whose distance from the first feature point and the second feature point meets a preset requirement is selected as a reference point; The lake land terrain data is corrected in a manner of correcting the first characteristic point to the reference point to determine the lake land terrain image, and the lake water area terrain data is corrected in a manner of correcting the second characteristic point to the reference point to determine the lake water area terrain image; The lake land terrain image and the lake water terrain image are stitched together.
2. The method for constructing a land-water integrated digital elevation model according to claim 1, characterized in that: The pre-processing step comprises: Coordinate transformation between terrain data: Taking the coordinate system of basic surveying and mapping data as the spatial reference, corresponding coordinate transformation is performed on MODIS terrain data, LiDAR terrain data, and multi-beam bathymetric data to achieve coordinate unification of multi-source data; Data geometric registration: Using basic surveying and mapping data as the reference image, a certain number of ground features with the same name are selected as control points in the corresponding image area, and the MODIS terrain data, LiDAR terrain data, and multi-beam bathymetric data are geometrically registered using the polynomial correction method.
3. The method for constructing a land-water integrated digital elevation model according to claim 1, characterized in that: The construction of lake land error prediction model includes: The LiDAR terrain data uses a first error prediction model to obtain a first elevation deviation value and a first feature weight map; the MODIS terrain data uses a second error prediction model to obtain a second elevation deviation value and a second feature weight map; The first elevation deviation value is used to correct the LiDAR terrain data to obtain a radar image to be fused; the second elevation deviation value is used to correct the MODIS terrain data to obtain a remote sensing image to be fused; The radar image to be fused and the remote sensing image to be fused are weightedly fused according to the first feature weight map and the second feature weight map to obtain fused elevation data.
4. The method for constructing a land-water integrated digital elevation model according to claim 3, characterized in that: The first error prediction model and the second error prediction model training steps include: Collect MODIS terrain data, LiDAR terrain data and their corresponding measured error samples as training data; Construct a deep neural network model, take the elevation data as input, and predict the corresponding error value, wherein the deep neural network model includes multiple hidden layers and uses a negatively correlated mapping function G to learn the complex relationship between data and error; Use training data to train the model and optimize network parameters to minimize the difference between the predicted error and the measured error; The trained model can be used to predict elevation deviation values or feature weight maps.
5. The method for constructing a land-water integrated digital elevation model according to claim 3, characterized in that: The calculation weights of the first feature weight map and the second feature weight map are calculated using the inverse proportion of the squared elevation error or from the normalized residual.
6. The method for constructing a land-water integrated digital elevation model according to claim 1, characterized in that: Specific steps for water area error value and water area characteristic weight value: According to the obtained overlapping area, binary regression analysis is performed with the coordinate values (x, y) as independent variables to determine the binary polynomial regression equations between the two types of data and evaluate the accuracy of the regression equations; Based on the determined bivariate polynomial regression equation, two sets of lake bottom topography are calculated for the entire lake area at the spatial resolution of the multibeam bathymetric data; The generated lake bottom topography is subtracted from the basic surveying and mapping data to obtain the water area error value and water area characteristic weight map.
7. The method for constructing a land-water integrated digital elevation model according to claim 1, characterized in that: The lake land terrain image and the lake water terrain image are stitched together, and also include: Determine that a specified pixel point in the lake land terrain image is a feature point A, and image features of the feature point A and image features of pixels near the feature point A meet a preset difference condition; Determine whether the image feature of feature point B and the image feature of pixel points near feature point B meet a preset difference condition, wherein feature point B is a pixel point corresponding to feature A in the lake water area topography image; If so, it is determined that the feature point A is the first feature point and the feature point B is the second feature point.
8. A system for constructing a land-water integrated digital elevation model, using the land-water integrated digital elevation model construction method according to any one of claims 1 to 7, characterized in that: include: The acquisition module collects multi-source data of the lake area and surrounding land: MODIS terrain data, LiDAR terrain data, and multi-beam bathymetric data; The preprocessing module uses the basic surveying and mapping data as a benchmark to perform coordinate transformation and geometric registration preprocessing on MODIS terrain data, LiDAR terrain data, and multi-beam bathymetric data; The land terrain data acquisition module uses MODIS terrain data and LiDAR terrain data as training samples to construct lake and land error prediction models respectively, and obtain land deviation values and feature weight values; and uses the land deviation values and feature weight values to perform weighted fusion on the MODIS terrain data and the LiDAR terrain data to obtain lake and land terrain data; The water area terrain data acquisition module overlays and compares MODIS terrain data with multi-beam bathymetry data, and LiDAR terrain data with multi-beam bathymetry data, determines the overlapping area between the two sets of data, and cuts the overlapping parts of the data respectively to obtain the overlapping area of MODIS terrain data and the overlapping area of LiDAR terrain data; Generate a regression equation for the overlapping area of the MODIS terrain data and the overlapping area of the LiDAR terrain data, and then calculate the difference between the two sets of lake bottom terrain data of the entire lake area to obtain a water area error value and a water area feature weight map; Obtain lake water area topography data through the water area error value and the water area characteristic weight map; A splicing module, which determines a splicing area according to the lake land terrain data and the lake water terrain data, and marks a first feature point and a second feature point in the splicing area respectively; In the stitching area, a point whose distance from the first feature point and the second feature point meets a preset requirement is selected as a reference point; The lake land terrain data is corrected in a manner of correcting the first characteristic point to the reference point to determine the lake land terrain image, and the lake water area terrain data is corrected in a manner of correcting the second characteristic point to the reference point to determine the lake water area terrain image; The lake land terrain image and the lake water terrain image are stitched together.
9. The land-water integrated digital elevation model construction system according to claim 1, characterized in that: The land terrain data acquisition module includes: A first prediction unit, wherein the LiDAR terrain data uses a first error prediction model to obtain a first elevation deviation value and a first feature weight map; The second prediction unit uses the second error prediction model to obtain the second elevation deviation value and the second feature weight map of the MODIS terrain data; a first correction unit, wherein the first elevation deviation value corrects the LiDAR terrain data to obtain a radar image to be fused; A second correction unit, wherein the second elevation deviation value corrects the MODIS terrain data to obtain a remote sensing image to be fused; A fusion unit performs weighted fusion on the radar image to be fused and the remote sensing image to be fused according to the first feature weight map and the second feature weight map to obtain fused elevation data.
10. The method for constructing a land-water integrated digital elevation model according to claim 1, characterized in that: The water terrain data acquisition module includes: The regression unit, based on the obtained overlapping area, performs binary regression analysis with the coordinate values (x, y) as independent variables, determines the binary polynomial regression equations between the two types of data and evaluates the accuracy of the regression equations; A calculation unit calculates two sets of lake bottom topography for the entire lake area at the spatial resolution of the multibeam bathymetric data according to the determined bivariate polynomial regression equation; The difference unit calculates the difference between the generated lake bottom topography and the basic surveying and mapping data to obtain the water area error value and water area characteristic weight map.
Citation Information
Patent Citations
Image splicing method
CN104331872A
A seasonal lake bottom topography generation method based on multi-source data fusion
CN109446485A
Multi-source elevation data fusion method and device, computer equipment and medium
CN116244662A
Construction of topographic map through integrating land and sea at coastal area
KR101019555B1
Cited By
Slope surface positioning method for geological surveying and mapping
CN120428292A
Lake water surface area regulation and control scheme generation method and device
CN120542313A
