CN112529811A - DEM data denoising method for preserving surface structure characteristics of terrain - Google Patents
DEM data denoising method for preserving surface structure characteristics of terrain Download PDFInfo
- Publication number
- CN112529811A CN112529811A CN202011500905.7A CN202011500905A CN112529811A CN 112529811 A CN112529811 A CN 112529811A CN 202011500905 A CN202011500905 A CN 202011500905A CN 112529811 A CN112529811 A CN 112529811A
- Authority
- CN
- China
- Prior art keywords
- original
- triangular
- ith
- dem data
- tin
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 96
- ATJFFYVFTNAWJD-UHFFFAOYSA-N Tin Chemical compound [Sn] ATJFFYVFTNAWJD-UHFFFAOYSA-N 0.000 claims abstract description 48
- 239000013598 vector Substances 0.000 claims abstract description 47
- 238000001914 filtration Methods 0.000 claims abstract description 34
- 238000012916 structural analysis Methods 0.000 claims abstract description 12
- 238000009499 grossing Methods 0.000 claims abstract description 10
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 21
- 230000001788 irregular Effects 0.000 claims description 11
- 230000005484 gravity Effects 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 3
- 239000000284 extract Substances 0.000 abstract description 3
- 238000003696 structure analysis method Methods 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 7
- 230000000694 effects Effects 0.000 description 6
- 238000000605 extraction Methods 0.000 description 5
- 230000000877 morphologic effect Effects 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 238000007781 pre-processing Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002146 bilateral effect Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012732 spatial analysis Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
- G06T17/05—Geographic models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration using local operators
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- Remote Sensing (AREA)
- Computer Graphics (AREA)
- Processing Or Creating Images (AREA)
Abstract
The invention provides a DEM data denoising method for preserving the structural features of a terrain surface, which extracts key elevation points and structural lines of an original terrain curved surface as the terrain structural features by using a structural analysis method and obtains a TIN surface; taking the TIN surface as a guide, and adopting a guide normal vector filter to smooth DEM data of the original terrain curved surface; and continuously taking the TIN surface as a guide, continuously adopting a guide normal vector filter to filter the TIN surface which is obtained in the previous time and is subjected to smoothing treatment, and performing iterative treatment until DEM data meeting the denoising requirement is obtained. According to the method, the reconstructed terrain surface is used as a guide constraint filtering process, the problem that the structure edge is excessively smooth in the traditional filtering process is solved, a structure analysis method is introduced to extract key elevation points and boundary lines, and the structural characteristics of the terrain surface can be effectively preserved while DEM data are denoised.
Description
Technical Field
The invention relates to a DEM data denoising method for preserving the surface structure characteristics of a ground, belonging to the technical field of spatial data processing.
Background
The purpose of processing the spatial data is to process the acquired original geographic data and the data which does not meet the quality requirements of various spatial analysis technologies so as to meet the data quality requirements of various geographic information system applications. The problems of data redundancy, abnormity, interference and the like are solved, conditions are created for effective application of spatial data, and the method is a necessary premise for spatial data analysis. Terrain, an important component of the natural geographic environment, contains many topographical features that are critical for many applications, such as landscape analysis, hydrological analysis, and city design. The natural terrain mainly has two types of characteristics, one is a structural characteristic and mainly represents a structural line and an elevation key point (such as a water system line and a mountain vertex) of the terrain surface. These features are crucial to landscape interpretation and therefore will be preferentially preserved during the digital representation of terrain; the second is morphological characteristics, which mainly reflect the topographic morphology of the earth surface formed by long-term weathering, such as smooth morphology. How to effectively represent these two types of features is critical to the application of terrain analysis. Thus, denoising of surface data also means that both types of features of the terrain surface need to be effectively preserved while surface noise is removed.
With the rapid development of geographic information systems, many methods for representing terrain have been proposed, and among them, DEM (Digital Elevation Model) is widely used due to its good terrain expression capability. The existing terrain denoising technology mainly adopts a filtering method to remove the noise of DEM data, but the traditional filtering method is easy to excessively smooth terrain characteristic points and structural lines (such as mountain peak points and water system lines) on the surface of the DEM.
In remote sensing image processing, filters are typically used to remove small scale features (i.e., noise). A more widely used method is statistical filtering, which reduces the noise contribution of the image in such a way that local variations per unit cell are averaged. However, although the filtering method can adaptively remove part of the noise of the DEM data, it tends to blur the structural features of the terrain by being too smooth. Meanwhile, due to the complex topographic features on the surface of the DEM, although part of high-frequency structural features of the DEM can be kept by part of the filters, no effective measure is provided for extracting and keeping the structural features of the DEM in the filtering process.
In order to retain the topographic structure characteristics, the traditional map comprehensive research field adopts a structure analysis technology in the field of computational geometry, and the method introduces a Triangulated Irregular Network (TIN) to detect key elevation points and topographic structure lines. The method is simple and easy to implement, can effectively extract key points such as peaks and ridges and can construct the relation between the elevation points in the corresponding terrain areas. However, such methods view the topographical surface as a discrete structure, often ignoring smooth morphological features between adjacent topographical elements.
In conclusion, the filtering method can remove the DEM surface noise while effectively maintaining the smooth morphological characteristics of the DEM, but the method is easy to make the structural characteristics of the terrain fuzzy; and the structural analysis method can effectively retain the structural characteristics (such as mountain peaks and water system characteristics) of the terrain. Therefore, how to effectively retain the structural characteristics of the terrain while removing the DEM surface noise is an urgent problem to be solved.
Disclosure of Invention
In order to solve the defects of the prior art, the invention provides the DEM data denoising method for preserving the topographic surface structural features, which can effectively overcome the difficulties of fuzzy topographic structural features, excessive simplification of large-scale DEM data and the like in the DEM data denoising process, and can simultaneously accurately and efficiently preserve important structural features and smooth features of topographic data.
The technical scheme adopted by the invention for solving the technical problem is as follows: the DEM data denoising method for preserving the structural features of the surface of the ground comprises the following steps:
s1, extracting key elevation points and structure lines of the original topographic curved surface by using a structural analysis method as topographic structure features, and obtaining a reconstructed topographic surface, namely a TIN surface, according to the key elevation points and the structure lines;
s2, using the TIN surface as a guide, and adopting a guide normal vector filter to carry out smoothing treatment on DEM data of the original terrain curved surface to obtain a TIN surface after smoothing treatment;
and S3, taking the TIN surface as a guide, continuously adopting a guide normal vector filter to filter the TIN surface obtained in the previous time after smoothing, and repeating the process until DEM data meeting the denoising requirement is obtained.
Step S1 specifically includes the following processes:
s1.1, processing an original terrain curved surface by using a maximum z tolerance method to obtain a triangular net structure frame;
s1.2, analyzing the relation between key elevation points in a near river region by using a D8 flow path method to obtain a water flow line as a structural line;
and S1.3, generating a TIN surface by using the key elevation points and the structural lines as constraint conditions, wherein the TIN surface is used for representing the original terrain curved surface.
Step S1.1 specifically includes the following processes:
s1.1.1, setting a maximum z limit value to obtain key elevation points of DEM data of the original terrain curved surface, and firstly connecting four corner points of a DEM layer to form an initial irregular triangular net;
s1.1.2, finding out the triangle where each key elevation point is located, and solving the interpolation elevation of each triangle interpolation point;
s1.1.3, inserting the key elevation point with the largest interpolation into the irregular triangulation network according to the interpolation elevation and the interpolation of the original elevation point of the interpolation point of the triangle corresponding to the interpolation elevation to update the irregular triangulation network;
s1.1.4, repeating the steps S1.1.2 and S1.1.3 to carry out iterative updating until the elevation difference values of all the key elevation points are smaller than the maximum z limit value or all the key elevation points are inserted, and ending the step, thus obtaining the triangulation network structure framework.
Step S1.2 specifically includes the following processes:
s1.2.1, for a flow line in a near river flow area, supposing that the water flow of a single grid in DEM data of an original terrain curved surface can only flow into 8 adjacent grids, and selecting the direction with the steepest gradient as the water flow direction;
s1.2.2, for each 3 x 3 grid cell group, representing the gradient by using a distance weight fall value, namely calculating the fall between the central grid and the adjacent grid of each grid cell group respectively and dividing the fall by the distance between the central points of the grids; forming a flow direction grating graph by using each gradient in the DEM data;
s1.2.3, counting the number of inflow paths converged into each grid to determine a flow path, and identifying a valley and a watershed therein;
s1.2.4, utilizing Douglas-Peucker algorithm to complete the thinning of the linear elements in the grid graph, and obtaining the water flow line.
Step S2 specifically includes the following processes: extracting the normal of each triangle in the TIN surface and the gravity center of each triangle corresponding to the original terrain curved surface; calculating the space weight and the normal vector deflection angle weight of each adjacent surface normal vector in the TIN surface, normalizing the normal vectors for filtering, and iteratively filtering the noise surface normal vector through the weighted average of the adjacent surface normal vectors; and updating the position of the vertex of the triangle by using a least square rule based on the normal vector of the denoised triangular surface to obtain the smooth TIN surface.
Step S2 specifically includes the following processes:
s2.1, directly constructing according to the original terrain curved surface to obtain an original terrain curved surface, and recording a triangular surface and an adjacent triangular surface in the original terrain curved surface as an ith original triangular surface TiAnd jth original triangular face TjRespectively calculating the ith original triangular surface TiCenter of gravity c ofiAnd jth original triangular face TjCenter of gravity cjThe spatial distance weight s between;
s2.2, respectively finding the ith original triangular surface TiAnd jth original triangular face TjThe corresponding triangles in the TIN surface are respectively marked as the ith reconstruction triangle surface Ti GAnd jth reconstructed triangular surface Tj GRespectively calculating the ith reconstructed triangular surface Ti GCurrent normal ofAnd jth reconstructed triangular surface Tj GCurrent normal ofAn angular deflection weight r in between;
s2.3, according to the jth original triangular surface TjArea A ofjAnd calculating to obtain the jth original triangular surface TjFor ith original triangular surface TiThe weight impact value of (a);
s2.4, repeating the steps S2.1 to S2.3 to obtain the ith original triangular surface TiThe weight influence values of all adjoining triangular faces;
s2.5, setting the ith original triangular surface TiThe weight influence values of all the adjacent triangular surfaces are summed and normalized for filtering to obtain the ith original triangular surface TiAn updated normal;
s2.6, carrying out the process on the ith original triangular surface T by the following formulaiAnd (3) performing iterative updating on the updated normal:
whereinIs the ith original triangular surface T after the T +1 th iterationiThe updated normal line is then used as a reference,is the ith original triangular surface T after the T-th iterationiUpdated normal, KiRepresenting the original triangle surface T for the ithiSet normalization coefficient, N (i) is ith original triangular surface TiA set of contiguous triangular faces of (a);
s2.7, according to the ith original triangular surface TiUpdated normalsUpdating ith original triangular surface TiThe vertex of (1);
and S2.8, repeating the steps from S2.1 to S2.7 to obtain the smoothed TIN surface.
Step S2.7 specifically includes the following processes:
s2.7.1, obtaining the ith original triangular surface T after iterationiNormal to (a) is nfObtaining a normal vector n of the surface according to the property that the normal vector of the surface is orthogonal to the normal vectors of three edges of the regular triangle surfacefAnd ith original triangular surface TiThe relationship equation of the normal vector of each side where the vertex is located:
wherein xo、xpAnd xqRespectively being ith original triangular surface TiThree vertices of (2);
s2.7.2, making the ith original triangular surface T according to least square criterioniThe sum of error values of the products of the three side normal vectors and the updated face normal vector is minimized, and the error formula is as follows:
wherein X represents the ith original triangular surface TiF denotes a set of triangular faces, FqRepresents in which x isqIs a triangular face of apex, n'qIndicating a triangular surface FqThe updated normal vector is obtained by the last iterative calculation of the formula (1),showing that the q-th triangular surface F is formedqA set of boundary lines of (a);
s2.7.3, for the ith original triangular surface T by the following formulaiUpdating the vertex position of (1):
wherein x'oRepresenting a vertex xoUpdatingVertex of the rear, NV(o) denotes the vertex xoAnd λ represents the iteration step.
The invention has the beneficial effects based on the technical scheme that:
the DEM data denoising method for reserving the structural features of the terrain surface extracts the main structural features of the terrain surface by using a structural analysis method, the structural features of the terrain surface are usually positioned at the places (namely key elevation points or boundary lines) where terrain parameters (such as gradient and slope direction) are remarkably changed, after the key elevation points and the boundary lines are extracted, linear elements are thinned, redundant geometric data points are deleted, a simplified curved surface can be updated, and the structural features of the original terrain curved surface can be accurately represented. Secondly, taking a reconstructed terrain surface structure (TIN surface) as a guide constraint filtering process; thirdly, iterative filtering is carried out, the de-noising of the DEM curved surface obtained in the previous iteration is continuously guided by using a reconstructed terrain surface structure (TIN surface), the more the iteration times are, the smoother the result is, and the DEM data can obtain a satisfactory result in the aspect of keeping smooth form and structural characteristics through a certain iteration time; and finally, updating the structure points on the curved surface to obtain a DEM data denoising result. According to the invention, the TIN structure and the water system line obtained by a structural analysis method are used as guidance to constrain the filtering process, so that the excessive smoothness of the DEM surface after denoising can be effectively avoided; in each iteration, the reconstructed TIN structure and the water system line are used for guiding the de-noising of the DEM surface obtained in the previous guided filtering iteration, through a certain number of iterations, the method can obtain a satisfactory de-noising result in the aspect of keeping smooth morphological and structural characteristics, and the effectiveness of the de-noising method provided by the invention is verified by using experiments.
Drawings
FIG. 1 is a flow chart of a DEM data denoising method for preserving topographic surface structure features.
Fig. 2 is a schematic diagram of structural analysis (identification of key elevation points and extraction of structural boundaries), where fig. 2(a) is raw DEM data, fig. 2(b) is key elevation points extracted after preprocessing, fig. 2(c) is structural lines (i.e., water flow lines) extracted after preprocessing, and fig. 2(d) is a reconstructed TIN surface.
Fig. 3 is a simplified schematic diagram of a DEM data denoising method for preserving the structural features of a surface of a ground, where fig. 3(a) is an original DEM surface, fig. 3(b) is an effect after extracting the structural features, fig. 3(c) is an effect after processing by a structural analysis method, and fig. 3(d) is a result of the guided post-processing by the method using the result of the structural analysis method.
Fig. 4 is a DEM data effect diagram under different denoising methods, where fig. 4(a) is original DEM data used in the embodiment, fig. 4(b) is DEM data after noise is added to the data in fig. 4(a), fig. 4(c) is a denoising result diagram of a common filtering denoising method, i.e., a gaussian filtering method, fig. 4(d) is a denoising result diagram of a latest guided filtering denoising method (Yuet al, 2020), and fig. 4(e) is a denoising result diagram of the method provided by the present invention.
Detailed Description
The invention is further illustrated by the following figures and examples.
(1) The main process and the working principle of the invention are as follows:
1. characteristic point identification preprocessing:
structural features of a terrain surface are typically located where terrain parameters (such as slope and direction) vary significantly (i.e., critical elevation points). It is therefore important to achieve extraction of structure points in order to be able to effectively preserve the structural features of the topographical surface. And using a maximum z tolerance algorithm, and dynamically selecting key elevation points by using the original elevation of the grid points and the estimated elevation difference of the triangles containing the points.
Obtaining key elevation points through an original DEM layer and a set maximum Z limit difference, firstly connecting four corner points of the grid DEM to form an initial triangular net, secondly finding a triangle where each elevation point is located, solving an interpolation elevation of an interpolation point of the triangle, easily obtaining a difference value between the interpolation elevation and the original elevation of the point, inserting the elevation point with the maximum difference value into an irregular triangular net to update a triangular net structure frame, and iteratively updating until the elevation difference values of all the elevation points are smaller than the Z limit difference, and then ending the updating to obtain the finally generated triangular net structure frame.
2. Structural boundary extraction preprocessing:
when the structural boundary is extracted, the streamline extraction in the near river region is completed by adopting a D8 algorithm. The algorithm assumes that the water flow in a single grid can only flow into 8 adjacent grids, and selects the direction with the steepest gradient as the water flow direction. In 3 x 3 DEM grids, the slope is replaced by a distance weight drop value, i.e. the drop between the central grid and the adjacent grid is calculated and divided by the distance between the centre points of the grids.
The extraction of the river regional flow line is completed, firstly, the whole DEM image is required to slide through a 3 x 3 grid window, the distance weight difference is calculated to determine the flow direction of each unit, and a flow direction grid diagram is generated; the determination of the flow path is then accomplished by counting the Number of Incoming Paths (NIP) that are merged into each cell. The regions with a greater NIP can be considered as valleys, while the regions with NIP equal to 0 are higher regions, possibly watershed of watershed.
In order to better represent the flow line characteristics, a large number of redundant geometric data points are deleted, and the thinning of the linear elements can be completed through a Douglas-Peucker algorithm, so that the aim of data simplification is fulfilled, and the geometric shape characteristics are reserved. In the implementation process of the Douglas-Peucker algorithm, firstly, the head and tail points of a streamline are connected into a straight line, and the maximum value of the distance between all the intermediate points and the straight line is solved; and comparing the maximum distance value with the rarefaction threshold value, if the maximum distance value is smaller than the rarefaction threshold value, completely eliminating the middle point on the curve, otherwise, dividing the curve into two parts by taking the point as a boundary, and repeating the process on the two parts of the curve until all the points are processed.
3. Smoothing DEM data based on TIN based on an extended joint bilateral filter:
and (3) taking the reconstructed terrain surface obtained in the steps 1 and 2 as a guide constraint filtering process, so that the morphological characteristics of the generated data are more similar to the real conditions. Unlike conventional image processing methods, we set the pixel of the image to the normal, rather than the pixel, during filtering, in view of the convenience of DEM triangulation.
(2) Example (b):
referring to fig. 1 to 3, the data set adopted in this embodiment is DEM data downloaded from the official website of the USGS (united states geological exploration agency), and is composed of 64 × 57 grid cells, and is processed by the DEM data denoising method for preserving surface structural features provided by the present invention, including the following steps:
s1, extracting key elevation points and structure lines of the original topographic curved surface by using a structural analysis method as topographic structure features, and obtaining a reconstructed topographic surface, namely a TIN surface, according to the key elevation points and the structure lines:
s1.1, processing the original terrain curved surface by using a maximum z tolerance method to obtain a triangulation network structure frame:
s1.1.1, setting a maximum z limit value to obtain key elevation points of DEM data of the original terrain curved surface, and firstly connecting four corner points of a DEM layer to form an initial irregular triangular net;
s1.1.2, finding out the triangle where each key elevation point is located, and solving the interpolation elevation of each triangle interpolation point;
s1.1.3, inserting the key elevation point with the largest interpolation into the irregular triangulation network according to the interpolation elevation and the interpolation of the original elevation point of the interpolation point of the triangle corresponding to the interpolation elevation to update the irregular triangulation network;
s1.1.4, repeating the steps S1.1.2 and S1.1.3 to perform iterative updating until the elevation difference values of all the key elevation points are smaller than the maximum z limit value or all the key elevation points are inserted, and ending the iteration updating to obtain a triangulation network structure frame;
s1.2, analyzing the relation between key elevation points in the near river region by using a D8 flow path method to obtain a water flow line as a structural line:
s1.2.1, for a flow line in a near river flow area, supposing that the water flow of a single grid in DEM data of an original terrain curved surface can only flow into 8 adjacent grids, and selecting the direction with the steepest gradient as the water flow direction;
s1.2.2, for each 3 x 3 grid cell group, representing the gradient by using a distance weight fall value, namely calculating the fall between the central grid and the adjacent grid of each grid cell group respectively and dividing the fall by the distance between the central points of the grids; forming a flow direction grating graph by using each gradient in the DEM data;
s1.2.3, counting the number of inflow paths converged into each grid to determine a flow path, and identifying a valley and a watershed therein;
s1.2.4, utilizing a Douglas-Peucker algorithm to finish the thinning of linear elements in the flow direction grid graph to obtain a water flow line;
and S1.3, generating a TIN surface by using the key elevation points and the structural lines as constraint conditions, wherein the TIN surface is used for representing the original terrain curved surface.
And S2, using the TIN surface as a guide, and adopting a guide normal vector filter to smooth the DEM data of the original terrain curved surface to obtain the smooth TIN surface. Specifically, extracting the normal of each triangle in the TIN surface and the gravity center of each triangle corresponding to the original terrain curved surface; calculating the space weight and the normal vector deflection angle weight of each adjacent surface normal vector in the TIN surface, normalizing the normal vectors for filtering, and iteratively filtering the noise surface normal vector through the weighted average of the adjacent surface normal vectors; based on the normal vector of the denoised triangular surface, updating the position of the vertex of the triangle by using a least square criterion to obtain the TIN surface after smoothing treatment:
s2.1, directly constructing according to the original terrain curved surface to obtain an original terrain curved surface, and recording a triangular surface and an adjacent triangular surface in the original terrain curved surface as an ith original triangular surface TiAnd jth original triangular face TjRespectively calculating the ith original triangular surface TiCenter of gravity c ofiAnd jth original triangular face TjCenter of gravity cjThe spatial distance weight s between;
s2.2, respectively finding the ith original triangular surface TiAnd jth original triangular face TjThe corresponding triangles in the TIN surface are respectively marked as the ith reconstruction triangle surface Ti GAnd jth reconstructed triangular surface Tj GRespectively calculating the ith reconstructed triangular surface Ti GCurrent normal ofAnd jth reconstructed triangular surface Tj GCurrent normal ofAn angular deflection weight r in between;
s2.3, according to the jth original triangular surface TjArea A ofjAnd calculating to obtain the jth original triangular surface TjFor ith original triangular surface TiThe weight impact value of (a);
s2.4, repeating the steps S2.1 to S2.3 to obtain the ith original triangular surface TiThe weight influence values of all adjoining triangular faces;
s2.5, setting the ith original triangular surface TiThe weight influence values of all the adjacent triangular surfaces are summed and normalized for filtering to obtain the ith original triangular surface TiAn updated normal;
s2.6, carrying out the process on the ith original triangular surface T by the following formulaiAnd (3) performing iterative updating on the updated normal:
whereinIs the ith original triangular surface T after the T +1 th iterationiThe updated normal line is then used as a reference,is the ith original triangular surface T after the T-th iterationiUpdated normal, KiRepresenting the original triangle surface T for the ithiSet normalization coefficient, N (i) is ith original triangular surface TiA set of contiguous triangular faces of (a);
s2.7, according to the ith original triangular surface TiUpdated normalsUpdating ith original triangular surface TiThe vertex of (a):
s2.7.1, obtaining the ith original triangular surface T after iterationiNormal to (a) is nfObtaining a normal vector n of the surface according to the property that the normal vector of the surface is orthogonal to the normal vectors of three edges of the regular triangle surfacefAnd ith original triangular surface TiThe relationship equation of the normal vector of each side where the vertex is located:
wherein xo、xpAnd xqRespectively being ith original triangular surface TiThree vertices of (2);
s2.7.2, making the ith original triangular surface T according to least square criterioniThe sum of error values of the products of the three side normal vectors and the updated face normal vector is minimized, and the error formula is as follows:
wherein X represents the ith original triangular surface TiF denotes a set of triangular faces, FqRepresents in which x isqIs a triangular face of apex, n'qIndicating a triangular surface FqThe updated normal vector is obtained by the last iterative calculation of the formula (1),showing that the q-th triangular surface F is formedqA set of boundary lines of (a);
s2.7.3, for the ith original triangular surface T by the following formulaiUpdating the vertex position of (1):
wherein x'oRepresenting a vertex xoUpdated vertex, NV(o) denotes the vertex xoThe vertex set of a ring of neighborhoods, and lambda represents the iteration step length; the value of the lambda is small in step length and long in time consumption, the step length is large and unstable, and the lambda can be taken asI.e. vertex xoA ring neighborhood of NV(o) an iterative computation with a set of 6 triangular faces, and a final updated vertex x'oThe calculation formula of the position is as follows:
s2.8, repeating the steps from S2.1 to S2.7 to obtain a smoothed TIN surface;
and S3, taking the TIN surface as a guide, continuously adopting a guide normal vector filter to filter the TIN surface obtained in the previous time after smoothing, and repeating the process until DEM data meeting the denoising requirement is obtained.
(3) And (3) experimental comparison:
in an experiment, a Gaussian filtering method, another latest guided filtering method and the method provided by the invention are respectively realized based on noise DEM data, the drainage characteristic can be kept at the same time of smooth DEM, and the results of the three methods (fig. 4(c), fig. 4(d) and fig. 4(e)) are compared, wherein the terrain surface in the denoising result graph of the Gaussian filtering method in fig. 4(c) has larger roughness, the edge part in the denoising result graph of the latest guided filtering denoising method (Yueal, 2020) in fig. 4(d) is too smooth, and partial abnormal pixel values appear, and the result graph obtained by the guided filtering denoising method provided by the invention in fig. 4(e) has the best comprehensive effect of keeping a structural line and reducing the roughness of the terrain surface.
The following table shows the comparison of the effects of the four denoising methods:
TABLE 1 comparison of the effects of different denoising methods
Referring to table 1, the square mean error (RMSE) and the water grid error matching rate are used to measure the accuracy of the obtained DEM data, and a smaller RMSE value, a smaller matching error and a larger matching accuracy indicate a higher accuracy, thus demonstrating the effectiveness of the method of the present invention.
The DEM data denoising method for preserving the surface structural features provided by the invention takes the original DEM terrain curved surface as a guide constraint filtering process, solves the problem of excessive smoothness of the structure edge in the traditional filtering, performs iterative filtering, guides the smoothness of the DEM curved surface obtained by the previous iteration by using the TIN curved surface reconstructed by a structural analysis method, introduces the structural analysis method to extract key elevation points and boundary lines, and can effectively preserve the surface structural features while denoising the DEM data.
The specific embodiments described herein are merely illustrative of the spirit of the invention. Various modifications or additions may be made to the described embodiments or alternatives may be employed by those skilled in the art without departing from the spirit or ambit of the invention as defined in the appended claims.
Claims (8)
1. A DEM data denoising method for preserving the structural features of a surface of a ground is characterized by comprising the following steps:
s1, extracting key elevation points and structure lines of the original topographic curved surface by using a structural analysis method as topographic structure features, and obtaining a reconstructed topographic surface, namely a TIN surface, according to the key elevation points and the structure lines;
s2, using the TIN surface as a guide, and adopting a guide normal vector filter to carry out smoothing treatment on DEM data of the original terrain curved surface to obtain a TIN surface after smoothing treatment;
and S3, taking the TIN surface as a guide, continuously adopting a guide normal vector filter to filter the TIN surface obtained in the previous time after smoothing, and repeating the process until DEM data meeting the denoising requirement is obtained.
2. The DEM data denoising method for preserving topographic surface texture features as claimed in claim 1, wherein: step S1 specifically includes the following processes:
s1.1, processing an original terrain curved surface by using a maximum z tolerance method to obtain a triangular net structure frame;
s1.2, analyzing the relation between key elevation points in a near river region by using a D8 flow path method to obtain a water flow line as a structural line;
and S1.3, generating a TIN surface by using the key elevation points and the structural lines as constraint conditions, wherein the TIN surface is used for representing the original terrain curved surface.
3. The DEM data denoising method for preserving topographic surface texture features as claimed in claim 2, wherein: step S1.1 specifically includes the following processes:
s1.1.1, setting a maximum z limit value to obtain key elevation points of DEM data of the original terrain curved surface, and firstly connecting four corner points of a DEM layer to form an initial irregular triangular net;
s1.1.2, finding out the triangle where each key elevation point is located, and solving the interpolation elevation of each triangle interpolation point;
s1.1.3, inserting the key elevation point with the largest interpolation into the irregular triangulation network according to the interpolation elevation and the interpolation of the original elevation point of the interpolation point of the triangle corresponding to the interpolation elevation to update the irregular triangulation network;
s1.1.4, repeating the steps S1.1.2 and S1.1.3 to carry out iterative updating until the elevation difference values of all the key elevation points are smaller than the maximum z limit value or all the key elevation points are inserted, and ending the step, thus obtaining the triangulation network structure framework.
4. The DEM data denoising method for preserving topographic surface texture features as claimed in claim 2, wherein: step S1.2 specifically includes the following processes:
s1.2.1, for a flow line in a near river flow area, supposing that the water flow of a single grid in DEM data of an original terrain curved surface can only flow into 8 adjacent grids, and selecting the direction with the steepest gradient as the water flow direction;
s1.2.2, for each 3 x 3 grid cell group, representing the gradient by using a distance weight fall value, namely calculating the fall between the central grid and the adjacent grid of each grid cell group respectively and dividing the fall by the distance between the central points of the grids; forming a flow direction grating graph by using each gradient in the DEM data;
s1.2.3, counting the number of inflow paths converged into each grid to determine a flow path, and identifying a valley and a watershed therein;
s1.2.4, utilizing Douglas-Peucker algorithm to complete the thinning of the linear elements in the grid graph, and obtaining the water flow line.
5. The DEM data denoising method for preserving topographic surface texture features as claimed in claim 1, wherein: step S2 specifically includes the following processes: extracting the normal of each triangle in the TIN surface and the gravity center of each triangle corresponding to the original terrain curved surface; calculating the space weight and the normal vector deflection angle weight of each adjacent surface normal vector in the TIN surface, normalizing the normal vectors for filtering, and iteratively filtering the noise surface normal vector through the weighted average of the adjacent surface normal vectors; and updating the position of the vertex of the triangle by using a least square rule based on the normal vector of the denoised triangular surface to obtain the smooth TIN surface.
6. The DEM data denoising method for preserving topographic surface texture features as claimed in claim 5, wherein: step S2 specifically includes the following processes:
s2.1, directly constructing according to the original terrain curved surface to obtain an original terrain curved surface, and recording a triangular surface and an adjacent triangular surface in the original terrain curved surface as an ith original triangular surface TiAnd jth original triangular face TjRespectively calculating the ith original triangular surface TiCenter of gravity c ofiAnd jth original triangular face TjCenter of gravity cjThe spatial distance weight s between;
s2.2, respectively finding the ith original triangular surface TiAnd jth original triangular face TjThe corresponding triangles in the TIN surface are respectively marked as the ith reconstruction triangle surface Ti GAnd jth reconstructed triangular surface Tj GRespectively calculating the ith reconstructed triangular surface Ti GCurrent normal ofAnd jth reconstructed triangular surface Tj GCurrent normal ofAn angular deflection weight r in between;
s2.3, according to the jth original triangular surface TjArea A ofjAnd calculating to obtain the jth original triangular surface TjFor ith original triangular surface TiThe weight impact value of (a);
s2.4, repeating the steps S2.1 to S2.3 to obtain the ith original triangular surface TiThe weight influence values of all adjoining triangular faces;
s2.5, setting the ith original triangular surface TiThe weight influence values of all the adjacent triangular surfaces are summed and normalized for filtering to obtain the ith original triangular surface TiAn updated normal;
s2.6, carrying out the process on the ith original triangular surface T by the following formulaiAnd (3) performing iterative updating on the updated normal:
whereinIs the ith original triangular surface T after the T +1 th iterationiThe updated normal line is then used as a reference,is the ith original triangular surface T after the T-th iterationiUpdated normal, KiRepresenting the original triangle surface T for the ithiSet normalization coefficient, N (i) is ith original triangular surface TiA set of contiguous triangular faces of (a);
s2.7, according to the ith original triangular surface TiUpdated normalsUpdating ith original triangular surface TiThe vertex of (1);
and S2.8, repeating the steps from S2.1 to S2.7 to obtain the smoothed TIN surface.
7. The DEM data denoising method for preserving topographic surface texture features as claimed in claim 6, wherein: step S2.7 specifically includes the following processes:
s2.7.1, obtaining the ith original triangular surface T after iterationiNormal to (a) is nfObtaining a normal vector n of the surface according to the property that the normal vector of the surface is orthogonal to the normal vectors of three edges of the regular triangle surfacefAnd ith original triangular surface TiThe relationship equation of the normal vector of each side where the vertex is located:
wherein xo、xpAnd xqRespectively being ith original triangular surface TiThree vertices of (2);
s2.7.2, making the ith original triangular surface T according to least square criterioniThe sum of error values of the products of the three side normal vectors and the updated face normal vector is minimized, and the error formula is as follows:
whereinX represents the ith original triangular surface TiF denotes a set of triangular faces, FqRepresents in which x isqIs a triangular face of apex, n'qIndicating a triangular surface FqThe updated normal vector is obtained by the last iterative calculation of the formula (1),showing that the q-th triangular surface F is formedqA set of boundary lines of (a);
s2.7.3, for the ith original triangular surface T by the following formulaiUpdating the vertex position of (1):
wherein x'oRepresenting a vertex xoUpdated vertex, NV(o) denotes the vertex xoAnd λ represents the iteration step.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011500905.7A CN112529811A (en) | 2020-12-17 | 2020-12-17 | DEM data denoising method for preserving surface structure characteristics of terrain |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011500905.7A CN112529811A (en) | 2020-12-17 | 2020-12-17 | DEM data denoising method for preserving surface structure characteristics of terrain |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112529811A true CN112529811A (en) | 2021-03-19 |
Family
ID=75001189
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011500905.7A Pending CN112529811A (en) | 2020-12-17 | 2020-12-17 | DEM data denoising method for preserving surface structure characteristics of terrain |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112529811A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115630181A (en) * | 2022-11-04 | 2023-01-20 | 中科星图数字地球合肥有限公司 | Data thinning method, device and related equipment |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106446910A (en) * | 2016-09-12 | 2017-02-22 | 电子科技大学 | Complex geological curved surface feature extraction and reconstruction method |
CN109410265A (en) * | 2019-01-22 | 2019-03-01 | 江苏省测绘工程院 | A kind of TIN filtering innovatory algorithm based on previous DEM auxiliary |
US20200019856A1 (en) * | 2018-07-11 | 2020-01-16 | Climate Central | System and method for a global digital elevation model |
CN111553292A (en) * | 2020-04-30 | 2020-08-18 | 南京工业大学 | Rock mass structural plane identification and occurrence classification method based on point cloud data |
CN112085750A (en) * | 2020-09-15 | 2020-12-15 | 广东奥普特科技股份有限公司 | Triangular mesh segmentation and denoising method |
-
2020
- 2020-12-17 CN CN202011500905.7A patent/CN112529811A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106446910A (en) * | 2016-09-12 | 2017-02-22 | 电子科技大学 | Complex geological curved surface feature extraction and reconstruction method |
US20200019856A1 (en) * | 2018-07-11 | 2020-01-16 | Climate Central | System and method for a global digital elevation model |
CN109410265A (en) * | 2019-01-22 | 2019-03-01 | 江苏省测绘工程院 | A kind of TIN filtering innovatory algorithm based on previous DEM auxiliary |
CN111553292A (en) * | 2020-04-30 | 2020-08-18 | 南京工业大学 | Rock mass structural plane identification and occurrence classification method based on point cloud data |
CN112085750A (en) * | 2020-09-15 | 2020-12-15 | 广东奥普特科技股份有限公司 | Triangular mesh segmentation and denoising method |
Non-Patent Citations (6)
Title |
---|
WENHAO YU ET.AL.: "An integrated method for DEM simplification with terrain structural features and smooth morphology preserved", 《INTERNATIONAL JOURNAL OF GEOGRAPHICAL INFORMATION SCIENCE》 * |
WENHAO YU ET.AL.: "An integrated method for DEM simplification with terrain structural features and smooth morphology preserved", 《INTERNATIONAL JOURNAL OF GEOGRAPHICAL INFORMATION SCIENCE》, 31 May 2020 (2020-05-31), pages 1 - 26 * |
XIANFANG SUN ET.AL.: "Fast and Effective Feature-Preserving Mesh Denoising", 《IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS》, vol. 13, no. 5, 30 October 2007 (2007-10-30), pages 925 - 938, XP011187636, DOI: 10.1109/TVCG.2007.1065 * |
刘洋等: "一种改进的渐进加密三角网点云滤波算法", 《测绘科学》 * |
刘洋等: "一种改进的渐进加密三角网点云滤波算法", 《测绘科学》, no. 05, 20 May 2020 (2020-05-20) * |
李志林,朱庆,谢潇: "水利水电工程三维可视化设计", 武汉大学出版社, pages: 192 - 79 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115630181A (en) * | 2022-11-04 | 2023-01-20 | 中科星图数字地球合肥有限公司 | Data thinning method, device and related equipment |
CN115630181B (en) * | 2022-11-04 | 2024-03-15 | 中科星图数字地球合肥有限公司 | Data thinning method and device and related equipment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109086773B (en) | Fault plane identification method based on full convolution neural network | |
Zheng et al. | Bilateral normal filtering for mesh denoising | |
CN102592136B (en) | Three-dimensional human face recognition method based on intermediate frequency information in geometry image | |
Lu et al. | Robust mesh denoising via vertex pre-filtering and l1-median normal filtering | |
CN111242864B (en) | Finger vein image restoration method based on Gabor texture constraint | |
Yoon et al. | Surface and normal ensembles for surface reconstruction | |
Zhang et al. | A statistical approach for extraction of feature lines from point clouds | |
Wang | Bilateral recovering of sharp edges on feature-insensitive sampled meshes | |
CN109509163B (en) | FGF-based multi-focus image fusion method and system | |
Dey et al. | Voronoi-based feature curves extraction for sampled singular surfaces | |
CN112085750A (en) | Triangular mesh segmentation and denoising method | |
CN114463644A (en) | River flow remote sensing monitoring method and device | |
CN111507921A (en) | Tunnel point cloud denoising method based on low-rank recovery | |
Moenning et al. | Fast marching farthest point sampling for implicit surfaces and point clouds | |
CN112529811A (en) | DEM data denoising method for preserving surface structure characteristics of terrain | |
CN115019163A (en) | City factor identification method based on multi-source big data | |
CN107274367B (en) | Three-dimensional geometric model denoising method based on structural feature description | |
CN108447038A (en) | A kind of mesh denoising method based on non local full variation operator | |
Sangeetha et al. | A novel exemplar based Image Inpainting algorithm for natural scene image completion with improved patch prioritizing | |
Shi et al. | Low dimensional manifold model with semi-local patches | |
CN112529803B (en) | Feature-preserving three-dimensional Mesh model denoising method | |
Guili et al. | A man-made object detection algorithm based on contour complexity evaluation | |
Tarmissi et al. | Multivariate kernel diffusion for surface denoising | |
CN107356968A (en) | Three-dimensional level set tomography curved surface extraction method based on crease | |
CN106652048B (en) | Three-dimensional model interest point extraction method based on 3D-SUSAN operator |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210319 |