Ground surface microtopography extraction method and device based on vegetation coverage conditions

Abstract

The invention discloses an earth surface microtopography extraction method and device based on vegetation coverage conditions, and the method comprises the steps: uniformly distributing control points at two sides of an observation region through the control point function of Agisoft Metashape software, obtaining the coordinates of each control point, setting the origin of the coordinates, carrying out the photographing to obtain an image data set, and carrying out the extraction of the earth surface microtopography; agisoft Lens open source software is used to automatically correct distorted photos of the image data set to obtain a corrected image data set and export a correction coefficient, photo alignment is carried out to generate sparse point clouds, camera alignment optimization is carried out according to the sparse point clouds and the control point coordinates to obtain dense point clouds, and the dense point clouds are subjected to image alignment optimization. And performing high-precision classification on the dense point clouds according to the HSV value to obtain classified point clouds, selecting the classified point clouds of which the point cloud distribution category is the ground, and performing interpolation to generate a digital elevation model, thereby realizing direct acquisition and processing of the slope surface microtopography DEM under the vegetation coverage condition, reducing the influence of vegetation on surface microtopography extraction, and improving the extraction efficiency of the slope surface microtopography. And the detection of the microtopography change process under the vegetation coverage condition is enhanced.

Description

Technical Field

[0001] The present application relates to the technical field of topographic measurement, and in particular to a method and device for extracting surface microtopography based on vegetation coverage conditions. Background Art

[0002] Microtopography, as a product of soil erosion and human activities, changes dynamically in the form of elevation fluctuations through tiny undulations, bumps, potholes and other terrain forms in a small area during the erosion process. This is of great significance to the surface hydrological cycle and the evolution of soil properties. However, due to the randomness of microtopography distribution and the uncertainty and complexity of the erosion process, especially after considering the vegetation coverage factor, the presence of plants not only increases the complexity of the surface structure, but also makes direct observation or measurement more difficult due to plant shading. Currently, commonly used technical means mainly include laser scanning technology and close-range photogrammetry to monitor changes in surface microtopography. Although these methods can meet the needs to a certain extent, they also have obvious limitations and challenges.

[0003] Although laser scanning technology can provide relatively accurate spatial location information, it is expensive and bulky and difficult to carry. More importantly, when faced with dense vegetation coverage, this technology is difficult to effectively penetrate the vegetation layer to accurately obtain the surface details of the obscured area, resulting in a large number of data collection blind spots. Close-range photogrammetry technology relies on capturing photos of the target object from different angles for three-dimensional reconstruction. However, this method is affected by factors such as large fluctuations in shooting distance (different distances), the inability to ensure uniform sampling density, and lens distortion problems, which often lead to low accuracy in the final modeling results, especially in small-scale micro-topography analysis that requires fine characterization.

[0004] In summary, the existing technical solutions have obvious shortcomings in dealing with surface microtopography extraction under vegetation coverage conditions, and it is urgent to develop more efficient and accurate new methods to overcome the above difficulties. Summary of the invention

[0005] The purpose of this application is to overcome the existing technical defects and provide a method and device for extracting surface microtopography based on vegetation coverage conditions. Vegetation point clouds are reconstructed through cocoon-style surround photography and software modeling, and high-precision point cloud models and DEMs of the surface under vegetation coverage conditions are obtained. Intelligent separation software is used to perform high-precision classification of slope point clouds, thereby realizing the monitoring and quantification of surface microtopography under vegetation coverage conditions.

[0006] The purpose of this application is achieved through the following technical solutions:

[0007] In a first aspect, the present application proposes a method for extracting surface micro-topography based on vegetation coverage conditions, the method comprising:

[0008] Use the control point function of Agisoft Metashape software to evenly distribute control points on both sides of the observation area, obtain the coordinates of each control point and set the coordinate origin;

[0009] Based on the coordinates of the origin and control points, a camera is used to take photos along the vegetation-covered slope using the cocoon photogrammetry method to obtain an image dataset;

[0010] Use Agisoft Lens open source software to automatically correct the distorted photos of the image dataset to obtain the corrected image dataset and derive the correction coefficients;

[0011] The corrected image data set and correction coefficients are imported into the software for camera calibration and then photo alignment is performed to generate a sparse point cloud;

[0012] Camera alignment optimization is performed based on the sparse point cloud and control point coordinates to obtain a dense point cloud;

[0013] Intelligent separation software is used to perform high-precision classification of dense point clouds according to HSV values ​​to obtain classified point clouds. The intelligent separation software is a script written based on Python.

[0014] Select the classified point cloud with the point cloud assignment category of ground to interpolate and generate a digital elevation model.

[0015] In a possible implementation, the step of performing camera alignment optimization according to the sparse point cloud and the control point coordinates to obtain a dense point cloud includes:

[0016] Detect and mark the coordinates of the control points, and then import the coordinates of the control points into the sparse point cloud to obtain a sparse point cloud with spatial coordinates;

[0017] After the sparse point cloud with spatial coordinates is adjusted to the center of the control point by software target, the camera alignment is optimized to obtain the sparse point cloud with optimized alignment;

[0018] The optimized aligned sparse point cloud is subjected to multi-view stereo matching to obtain a dense point cloud.

[0019] In a possible implementation, the method further includes:

[0020] If the observation area is outdoors, the control point columns are used to obtain the control point coordinates;

[0021] If the observation area is indoors, the control points will be pasted on the vertical surfaces of the soil trough side walls, and the spacing between the control points will be ensured to be the same, and the coordinates of the control points will be obtained.

[0022] In a possible implementation manner, the correction parameters used during automatic correction include radial distortion parameters and tangential distortion parameters.

[0023] In one possible implementation, the classified point cloud includes ground, high-level vegetation, and buildings.

[0024] In a possible implementation manner, the overlap between two adjacent images in the image data set is greater than 80%.

[0025] In a second aspect, the present application also proposes a surface micro-topography extraction device based on vegetation coverage conditions, the device comprising:

[0026] The acquisition module is used to evenly distribute control points on both sides of the observation area using the control point function of Agisoft Metashape software, obtain the coordinates of each control point and set the coordinate origin;

[0027] A photographing module is used to obtain an image data set by taking photos along a vegetation-covered slope using a camera device and a cocoon photogrammetry method based on the coordinate origin and the control point coordinates;

[0028] A correction module is used to automatically correct the distorted photos of the image data set using the Agisoft Lens open source software to obtain a corrected image data set and derive the correction coefficient;

[0029] The photo alignment module is used to use the corrected image data set and correction coefficients to import the software for camera calibration and then perform photo alignment to generate a sparse point cloud;

[0030] The camera alignment optimization module is used to optimize the camera alignment according to the sparse point cloud and the control point coordinates to obtain a dense point cloud;

[0031] The classification module is used to use intelligent separation software to perform high-precision classification of dense point clouds according to HSV values ​​to obtain classified point clouds. The intelligent separation software is a script written based on Python;

[0032] The interpolation module is used to select the classified point cloud whose point cloud assignment category is ground to interpolate and generate a digital elevation model.

[0033] In a possible implementation, the camera alignment optimization module is further used to:

[0034] Detect and mark the coordinates of the control points, and then import the coordinates of the control points into the sparse point cloud to obtain a sparse point cloud with spatial coordinates;

[0035] After the sparse point cloud with spatial coordinates is adjusted to the center of the control point by software target, the camera alignment is optimized to obtain the sparse point cloud with optimized alignment;

[0036] The optimized aligned sparse point cloud is subjected to multi-view stereo matching to obtain a dense point cloud.

[0037] In a possible implementation, the device further includes:

[0038] An outdoor coordinate acquisition module is used to acquire the coordinates of control points using control point columns if the observation area is outdoors;

[0039] The indoor coordinate acquisition module is used to paste the control points on the vertical surfaces of the soil trough side walls if the observation area is indoors, and ensure that the distances between the control points are the same to obtain the control point coordinates.

[0040] In one possible implementation, the classified point cloud includes ground, high-level vegetation, and buildings.

[0041] In a possible implementation manner, the overlap between two adjacent images in the image data set is greater than 80%.

[0042] The above-mentioned main scheme of the present application and its further options can be freely combined to form multiple schemes, all of which are schemes that can be adopted and claimed for protection in the present application; and in the present application, (non-conflicting options) options and other options can also be freely combined. After understanding the scheme of the present application, those skilled in the art can understand that there are multiple combinations based on the prior art and common knowledge, all of which are technical schemes to be protected by the present application, and they are not exhaustively listed here.

[0043] The present application discloses a method and device for extracting surface micro-topography based on vegetation coverage conditions. The method comprises the following steps: firstly, the control point function of Agisoft Metashape software is used to evenly distribute control points on both sides of an observation area, coordinates of each control point are obtained and the coordinate origin is set, and a camera is used to take photos along a slope covered with vegetation to obtain an image data set. Then, the distorted photos of the image data set are automatically corrected using Agisoft Lens open source software to obtain a corrected image data set, and a correction coefficient is derived. Photo alignment is performed to generate a sparse point cloud. Camera alignment optimization is performed according to the sparse point cloud and the coordinates of the control points to obtain a dense point cloud. The dense point cloud is classified according to HSV values ​​with high precision to obtain a classified point cloud. The classified point cloud with a point cloud assignment category of ground is selected for interpolation to generate a digital elevation model, so as to realize direct acquisition and processing of a DEM of surface micro-topography of a slope under vegetation coverage conditions, reduce the influence of vegetation on the extraction of surface micro-topography, and strengthen the detection of the micro-topography change process under vegetation coverage conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0045] Figure 1 A schematic flow chart of a method for extracting surface micro-topography based on vegetation coverage conditions proposed in an embodiment of the present application is shown.

[0046] Figure 2 The diagram shows the layout of the observation area and control points proposed in the embodiment of the present application.

[0047] Figure 3 The dense point cloud-based corn plant removal process and result diagram proposed in the embodiment of the present application are shown.

[0048] Figure 4 A comparison chart of the surface microtopography extraction method proposed in the embodiment of the present application and the conventional DEM is shown. DETAILED DESCRIPTION

[0049] The following describes the embodiments of the present application through specific examples, and those skilled in the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification. The present application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0050] Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of this application.

[0051] The problems and defects in the existing technology are as follows: the commonly used laser scanning technology is expensive and not easy to carry; after the vegetation coverage condition is superimposed, the spatial relationship becomes more complicated, and the vegetation blocks the information acquisition, and it is impossible to accurately solve the surface information under the vegetation coverage condition, and there are many blind spots for data collection; the shooting distance of close-range photogrammetry is sometimes far and sometimes close, and it is impossible to always maintain consistent shooting uniformity, which affects the modeling accuracy. In addition, lens distortion also has a significant impact on the quality of the modeling results. Therefore, the existing methods are difficult to apply to the extraction of surface microtopography under vegetation coverage conditions.

[0052] Therefore, in order to solve the above-mentioned technical problems, the embodiment of the present application proposes a method and device for extracting surface micro-topography based on vegetation coverage conditions, which takes into account the influence of vegetation planting and growth, avoids the vegetation from hindering the acquisition of surface information in space, reconstructs the vegetation point cloud through cocoon-style surround photography and software modeling, obtains the high-precision point cloud model and DEM of the surface under vegetation coverage conditions, and uses intelligent separation software to perform high-precision classification of slope point clouds, thereby realizing the monitoring and quantification of surface micro-topography under vegetation coverage conditions, which is described in detail below.

[0053] Please refer to Figure 1 , Figure 1 A schematic flow chart of a method for extracting surface micro-topography based on vegetation coverage conditions proposed in an embodiment of the present application is shown. The method is applied under vegetation coverage conditions and includes the following steps:

[0054] Step S1: Use the control point function of Agisoft Metashape software to evenly distribute control points on both sides of the observation area, obtain the coordinates of each control point and set the coordinate origin.

[0055] If the observation area is outdoors, the control point columns are used to obtain the control point coordinates;

[0056] If the observation area is indoors, the control points will be pasted on the vertical surfaces of the soil trough side walls, and the spacing between the control points will be ensured to be the same, and the coordinates of the control points will be obtained.

[0057] First, place the control points. If you use control point columns in an outdoor environment, divide these control point columns evenly on both sides of the observation area and ensure that the distance between them is equal. Select a control point in the lower left corner as the coordinate origin. If you are in an indoor environment, paste the control points on the facade of the soil trough side wall. Also, ensure that the control points are spaced at the same distance and cover the entire observation area. Select the first control point in the lower left corner as the coordinate origin.

[0058] In Agisoft Metashape, you can manually specify a control point as the coordinate origin to define the coordinate system of the entire project, then use a camera to take pictures of the observation area from different angles to ensure that all control points are clearly captured, and import the photos into the Agisoft Metashape software. In the software, through the "Detect Control Points" function, the centers of the control points marked in the picture are automatically detected one by one, and their known actual coordinate values ​​are entered. The software will automatically match the feature points in the image with the positions of the actual control points, and use the tools provided by the software to optimize the control points to improve the accuracy of the overall model, check the error value of each control point, and adjust the position of the control point or retake some photos if necessary.

[0059] Different from conventional self-made cross-shaped or circular control points, the control points in the embodiment of the present application are preset options of the Agisoft Metashape software and can be automatically identified. The marking type can select any one of the 12-, 14-, 16- or 20-bit coding patterns.

[0060] The control point columns in outdoor environments should be made of rigid materials that are not easily deformed to ensure stability in all weather conditions. The control point columns should be placed on both sides of the observation area and ensure that the spacing between them is the same. The position of each control point should be accurately recorded to facilitate precise coordinate input in the software.

[0061] If the control points in the indoor environment are pasted on the facade, the control points are pasted on the facade of the side wall of the soil trough. This method can better capture the complete control point image during oblique photography and improve recognition accuracy. If horizontal pasting is used, although the control points can also be pasted horizontally on the ground, this method may cause some areas of the control points in some images to be blocked, affecting the recognition effect. In addition, the distance between the control points should be based on the distance between the centers of the two control point targets to ensure that all control points are evenly distributed and have consistent spacing.

[0062] Step S2: Based on the coordinate origin and the coordinates of the control points, a camera is used to take photos along the vegetation-covered slope using a cocoon photogrammetry method to obtain an image data set.

[0063] The overlap between two adjacent images in the image data set is greater than 80%.

[0064] The video equipment includes but is not limited to cameras and mobile phones. Any machine that can obtain HSV value images of real features can be used. Cocoon-style photogrammetry is used to obtain a high-precision three-dimensional model. A cocoon body is generated along the normal direction of the outer contour of the shooting target at a specified shooting distance, perfectly mimicking the shooting target, and shooting is performed at equal distances within the cocoon body to ensure uniform coverage of the entire shooting target and to ensure that the overlap of two adjacent images is greater than 80%.

[0065] In addition, it is necessary to ensure that the distance from the target is consistent each time the shooting is performed, that the shooting points are evenly distributed, that there is no missed area, that the overlap between two adjacent images is greater than 80%, and that the shooting range completely covers the target area to be modeled. Before starting photography, it is necessary to locate the image based on the known coordinate origin and control point coordinates. In the final image data set, the overlap between two adjacent images is greater than 80% to ensure the accuracy of subsequent processing.

[0066] Step S3: Use Agisoft Lens open source software to automatically correct the distorted photos of the image data set to obtain a corrected image data set, and derive the correction coefficient.

[0067] Agisoft Lens open source software automatically calibrates photo distortion to reduce the impact of lens distortion on the quality of modeling results, exports correction coefficients as .xml format files, and then intuitively deselects redundant or blurred images in advance to ensure the accuracy of point cloud reconstruction, and creates multidimensional blocks for batch processing to speed up the efficiency of the method. The captured image dataset is imported into Agisoft Lens software, and the image is automatically calibrated using Agisoft Lens software to eliminate lens distortion, and the correction parameters are exported as .xml format files.

[0068] The correction parameters used in automatic correction include radial distortion parameters and tangential distortion parameters.

[0069] Lens distortion has a significant impact on the quality of the modeling results. Any change in the straight line projected between the object and the sensor can be attributed to radial distortion and causes misalignment of image properties. The parameters of radial distortion are K1, K2, and K3, and the parameters of tangential distortion are P1 and P2.

[0070] Step S4: Use the corrected image data set and correction coefficients to import the software to perform camera calibration and then perform photo alignment to generate a sparse point cloud.

[0071] Deselect redundant or blurred images to ensure that only high-quality images are used for subsequent processing, create multidimensional blocks for batch processing to improve work efficiency, export the corrected image dataset, import the corrected image dataset and correction coefficients into the image processing software (Agisoft Metashape), align the photos in the image processing software, and generate a sparse point cloud.

[0072] Step S5: Perform camera alignment optimization based on the sparse point cloud and the control point coordinates to obtain a dense point cloud.

[0073] The sparse point cloud is a preliminary 3D point set generated by image alignment, which contains feature points in the image. The control point coordinates are known spatial coordinate points. The sparse point cloud is combined with the control point coordinates, and the camera position and posture are automatically adjusted by software to match the known positions of the control points as closely as possible. This optimization can improve the accuracy of the camera parameters and thus the accuracy of the overall model. After the camera alignment optimization, the multi-view stereo matching algorithm is used to further refine the points in the sparse point cloud to generate a denser and more detailed 3D point set.

[0074] Step S5 includes:

[0075] Detect and mark the coordinates of the control points, and then import the coordinates of the control points into the sparse point cloud to obtain a sparse point cloud with spatial coordinates;

[0076] After the sparse point cloud with spatial coordinates is adjusted to the center of the control point by software target, the camera alignment is optimized to obtain the sparse point cloud with optimized alignment;

[0077] The optimized aligned sparse point cloud is subjected to multi-view stereo matching to obtain a dense point cloud.

[0078] Image processing software (Agisoft Metashape) was used to detect control points, optimize alignment, and generate dense point clouds. The image processing software was used to automatically detect and mark the control points, import the detected control point coordinates into the sparse point cloud, give the sparse point cloud spatial information, and use the software program to automatically calibrate the target and adjust it to the center of the control point. Finally, based on data analysis and mathematical algorithms rather than personal interpretation, the camera alignment was automatically optimized to improve the accuracy and reliability of the model. The sparse point cloud after optimized alignment was converted into a dense point cloud through multi-view stereo matching.

[0079] Step S6: Using intelligent separation software to perform high-precision classification of the dense point cloud according to the HSV value to obtain a classified point cloud.

[0080] The intelligent separation software is a script written in Python. HSV is a color model, where H represents hue, S represents saturation, and V represents value. Different types of ground objects, such as vegetation and ground, can be distinguished by HSV values. Using the intelligent separation software written in Python, the dense point cloud is classified with high precision according to the HSV value, and the point cloud is divided into two categories: "higher vegetation" and "ground". From the classified point cloud, the point cloud with the category of "ground" is selected, and the selected "ground" point cloud is interpolated using image processing software to generate a digital elevation model (DEM).

[0081] Step S7: Select the classified point cloud whose point cloud assignment category is ground for interpolation to generate a digital elevation model.

[0082] The classified point cloud includes ground, advanced vegetation, and buildings.

[0083] The intelligent separation software is a script written in Python. After extracting the image pixel data of the photo based on the HSV discrimination method, it analyzes the HSV value characteristics of the green vegetation point cloud and non-vegetation, and automatically identifies and filters them according to the threshold and assigns them as "higher vegetation" or "ground". Only the point cloud classified as "ground" is selected to generate a digital elevation model (DEM). The point cloud classified as "ground" is selected from the classified point cloud, and finally the interpolation algorithm of the image processing software is used to generate a digital elevation model (DEM).

[0084] In a possible implementation, the specific application of indoor and outdoor crops and farming measures in the embodiments of the present application is further described, and an indoor corn covering experiment is conducted in the soil erosion laboratory of Sichuan Agricultural University in Wenjiang District, Chengdu City, Sichuan Province, a field rapeseed experiment is conducted in the Wenjiang Huihe Experimental Base of Sichuan Agricultural University in Chengdu City, Sichuan Province, and a flat, horizontal ridge and along ridge field corn experiment is conducted in the teaching and research park of Sichuan Agricultural University in Dujiangyan City, Sichuan Province. The only difference between indoor and field experiments is the layout of control points.

[0085] In the field experiment, the control point columns are placed on both sides of the observation area, the distance between the control point columns is measured and kept without any displacement during the photography process, and the control point columns are kept horizontal. The bottom of the left control point column is the coordinate origin, i.e. (0, 0, 0). A steel ruler is used to measure the distance between the centers of the top and bottom control points and record them to obtain the coordinates of the four vertex control points. The remaining coordinates can be obtained by calculation. In the indoor experiment, the control points are pasted on both sides of the soil trough (soil trough specifications are 2m*1m*0.5m), and the control points are pasted accurately according to the settings based on the distance between the control point centers (a total of 12 control points, evenly distributed on both sides of the soil trough every 40cm) to ensure the accuracy of their coordinates. Figure 2 The diagram shows the layout of the observation area and control points proposed in the embodiment of the present application.

[0086] After determining the control points, use the camera to wrap around the slope in a cocoon-like manner to collect images. It is best to do this on a cloudy day or in an environment without direct sunlight. Light will affect the image quality and thus affect the reconstruction process of the object. Insufficient light can easily lead to more image noise, and too much light will cause shadows and thus incorrect edge detection. Due to the occlusion of corn in space, the camera position is constantly changed during the shooting process, but the shooting distance is always kept fixed, and the surface image is continuously obtained in multiple views in the normal direction of the object. The image content overlap is at least 80% to ensure the correct recognition of continuous photography.

[0087] Then the camera was calibrated, and Agisoft Lens software was used to reduce the significant impact of lens distortion on the quality of modeling results. Therefore, the corresponding sensor was used to obtain the chessboard image specified by the detection, calculate the distortion parameters, and export the correction coefficients in .xml format. The correction coefficients were K1: 0.00674, K2: -0.00346 and K3: 0.00076; the tangential distortion parameters were P1: 0.00139 and P2: -0.00083. Redundant or blurred images were intuitively deselected in advance to speed up the workflow.

[0088] Import the images and correction coefficients into Metashape software, align the photos to find the same features in different images through feature matching, use SIFT algorithm to extract the points with obvious features in the images, use Gaussian kernel functions at different scales to convolve with the images, generate Gaussian pyramids, build Gaussian filters, and then perform multiple downsampling processes on the images that need feature extraction to determine the corresponding spatial relationship between feature points between images. Solve the three-dimensional coordinates based on the triangulation method, and use the iterative bundle adjustment method to optimize the slope image, and finally construct a sparse point cloud through an incremental matching mode.

[0089] Detect the mark, the software automatically identifies the location of the control point, imports each control point separately, and places the software target at the center of the control point image.

[0090] After ensuring the accuracy of the control points, the sparse point cloud is readjusted by optimizing the camera alignment method to ensure that it has accurate spatial information and point cloud correlation.

[0091] Using sparse point clouds as seed points, point clouds are created through one-sided diffusion. By integrating multiple images with high overlap, point cloud filtering, point cloud registration, and error patch filtering are performed using parallax information and stereo matching algorithms to improve the quality of point cloud data. In order to further improve the accuracy and robustness of the model, multiple filtering and diffusion measures are combined to accurately obtain dense and accurate dense point cloud results.

[0092] According to the different light and dark colors presented by the HSV values ​​of the dense point cloud, a script written in Python is used to extract the image pixel data of the photo based on the HSV discrimination method, and the HSV value characteristics of the green vegetation point cloud and non-vegetation are analyzed. According to the threshold, the dense point cloud is automatically identified and screened and assigned to the point cloud category for classification processing. Corn and rapeseed vegetation are "advanced vegetation" type point clouds, and slopes are "ground" type point clouds. Figure 3 The corn plant removal process and result diagram based on dense point cloud proposed in the embodiment of the present application are shown. Compared with the conventional method, the present method uses intelligent separation software to classify the slope point cloud with high precision, and then directly obtains and processes the slope surface micro-topography DEM under vegetation coverage conditions by selecting the point cloud category for generating DEM. It can be efficiently, stably and accurately implemented under different vegetation and terrain conditions. The present invention realizes the separate extraction of micro-topography under superimposed vegetation conditions, creatively promoting the progress of research in the field of micro-topography. Figure 4 A comparison chart of the surface microtopography extraction method proposed in the embodiment of the present application and the conventional DEM is shown.

[0093] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0094] First, high-precision classification of dense point clouds through HSV values ​​can effectively distinguish vegetation from the ground, thereby solving the problem of vegetation occlusion, accurately extracting the micro-topographic features of the surface, and correctly constructing the surface morphology even when covered by vegetation.

[0095] Second, the intelligent separation software and image processing software written in Python are simple to operate and low-cost, suitable for more application scenarios, especially those with limited budgets.

[0096] Third, it has good directionality and high degree of freedom, can be flexibly adjusted according to actual needs, and can perform precise measurements in various complex environments, improving the flexibility and adaptability of applications.

[0097] Fourth, based on existing image processing technology and script automation processing, the operation is simple and easy to learn, which lowers the operation threshold and enables more researchers and technicians to use this method for micro-topography research.

[0098] Fifth, through high-precision classification and interpolation processing, the generated DEM has higher resolution and accuracy, which can more accurately detect and analyze the micro-topography change process and improve the reliability of the research results.

[0099] Sixth, the independent extraction of microtopography under superimposed vegetation conditions was achieved, which provided new tools and methods for the research in the field of microtopography and promoted scientific progress in related fields.

[0100] A possible implementation of a surface micro-topography extraction device based on vegetation coverage conditions is given below, which is used to execute the various execution steps and corresponding technical effects of the surface micro-topography extraction method shown in the above embodiments and possible implementations. The device includes:

[0101] The acquisition module is used to evenly distribute control points on both sides of the observation area using the control point function of Agisoft Metashape software, obtain the coordinates of each control point and set the coordinate origin;

[0102] A photographing module is used to obtain an image data set by taking photos along a vegetation-covered slope using a camera device and a cocoon photogrammetry method based on the coordinate origin and the control point coordinates;

[0103] A correction module is used to automatically correct the distorted photos of the image data set using the Agisoft Lens open source software to obtain a corrected image data set and derive the correction coefficient;

[0104] The photo alignment module is used to use the corrected image data set and correction coefficients to import the software for camera calibration and then perform photo alignment to generate a sparse point cloud;

[0105] The camera alignment optimization module is used to optimize the camera alignment according to the sparse point cloud and the control point coordinates to obtain a dense point cloud;

[0106] The classification module is used to use intelligent separation software to perform high-precision classification of dense point clouds according to HSV values ​​to obtain classified point clouds. The intelligent separation software is a script written based on Python;

[0107] The interpolation module is used to select the classified point cloud whose point cloud assignment category is ground to interpolate and generate a digital elevation model.

[0108] In a possible implementation, the camera alignment optimization module is further used to:

[0109] Detect and mark the coordinates of the control points, and then import the coordinates of the control points into the sparse point cloud to obtain a sparse point cloud with spatial coordinates;

[0110] After the sparse point cloud with spatial coordinates is adjusted to the center of the control point by software target, the camera alignment is optimized to obtain the sparse point cloud with optimized alignment;

[0111] The optimized aligned sparse point cloud is subjected to multi-view stereo matching to obtain a dense point cloud.

[0112] In a possible implementation, the device further includes:

[0113] An outdoor coordinate acquisition module is used to acquire the coordinates of control points using control point columns if the observation area is outdoors;

[0114] The indoor coordinate acquisition module is used to paste the control points on the vertical surfaces of the soil trough side walls if the observation area is indoors, and ensure that the distances between the control points are the same to obtain the control point coordinates.

[0115] In one possible implementation, the classified point cloud includes ground, high-level vegetation, and buildings.

[0116] In a possible implementation manner, the overlap between two adjacent images in the image data set is greater than 80%.

[0117] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for extracting surface micro-topography based on vegetation coverage conditions, characterized in that: The method comprises: Use the control point function of Agisoft Metashape software to evenly distribute control points on both sides of the observation area, obtain the coordinates of each control point and set the coordinate origin; Based on the coordinates of the origin and control points, a camera is used to take photos along the vegetation-covered slope using the cocoon photogrammetry method to obtain an image dataset; Use Agisoft Lens open source software to automatically correct the distorted photos of the image dataset to obtain the corrected image dataset and derive the correction coefficients; The corrected image data set and correction coefficients are imported into the software for camera calibration and then photo alignment is performed to generate a sparse point cloud; Camera alignment optimization is performed based on the sparse point cloud and control point coordinates to obtain a dense point cloud; Intelligent separation software is used to perform high-precision classification of dense point clouds according to HSV values ​​to obtain classified point clouds. The intelligent separation software is a script written based on Python. Select the classified point cloud with the point cloud assignment category of ground to interpolate and generate a digital elevation model.

2. The method for extracting surface micro-topography according to claim 1, characterized in that: The steps of optimizing camera alignment to obtain a dense point cloud based on the sparse point cloud and control point coordinates include: Detect and mark the coordinates of the control points, and then import the coordinates of the control points into the sparse point cloud to obtain a sparse point cloud with spatial coordinates; After the sparse point cloud with spatial coordinates is adjusted to the center of the control point by software target, the camera alignment is optimized to obtain the sparse point cloud with optimized alignment; The optimized aligned sparse point cloud is subjected to multi-view stereo matching to obtain a dense point cloud.

3. The method for extracting surface micro-topography according to claim 1, characterized in that: The method further comprises: If the observation area is outdoors, the control point columns are used to obtain the control point coordinates; If the observation area is indoors, the control points will be pasted on the vertical surfaces of the side walls of the soil trough, and the spacing between the control points will be ensured to be the same, and the coordinates of the control points will be obtained.

4. The method for extracting surface micro-topography according to claim 1, characterized in that: The correction parameters used in automatic correction include radial distortion parameters and tangential distortion parameters.

5. The method for extracting surface micro-topography according to claim 1, characterized in that: The classified point cloud includes ground, advanced vegetation, and buildings.

6. The method for extracting surface micro-topography according to claim 1, characterized in that: The overlap between two adjacent images in the image data set is greater than 80%.

7. A surface micro-topography extraction device based on vegetation coverage conditions, characterized in that: The device comprises: The acquisition module is used to evenly distribute control points on both sides of the observation area using the control point function of Agisoft Metashape software, obtain the coordinates of each control point and set the coordinate origin; A photographing module is used to obtain an image data set by taking photos along a vegetation-covered slope using a camera device and a cocoon photogrammetry method based on the coordinate origin and the control point coordinates; A correction module is used to automatically correct the distorted photos of the image data set using the Agisoft Lens open source software to obtain a corrected image data set and derive the correction coefficient; The photo alignment module is used to use the corrected image data set and correction coefficients to import the software for camera calibration and then perform photo alignment to generate a sparse point cloud; The camera alignment optimization module is used to optimize the camera alignment according to the sparse point cloud and the control point coordinates to obtain a dense point cloud; The classification module is used to use intelligent separation software to perform high-precision classification of dense point clouds according to HSV values ​​to obtain classified point clouds. The intelligent separation software is a script written based on Python; The interpolation module is used to select the classified point cloud whose point cloud assignment category is ground to interpolate and generate a digital elevation model.

8. The surface micro-topography extraction device according to claim 7, characterized in that: The camera alignment optimization module is also used for: Detect and mark the coordinates of the control points, and then import the coordinates of the control points into the sparse point cloud to obtain a sparse point cloud with spatial coordinates; After the sparse point cloud with spatial coordinates is adjusted to the center of the control point by software target, the camera alignment is optimized to obtain the sparse point cloud with optimized alignment; The optimized aligned sparse point cloud is subjected to multi-view stereo matching to obtain a dense point cloud.

9. The surface micro-topography extraction device according to claim 7, characterized in that: The device also includes: An outdoor coordinate acquisition module is used to acquire the coordinates of control points using control point columns if the observation area is outdoors; The indoor coordinate acquisition module is used to paste the control points on the vertical surfaces of the soil trough side walls if the observation area is indoors, and ensure that the distances between the control points are the same to obtain the control point coordinates.

10. The surface micro-topography extraction device according to claim 7, characterized in that: The classified point cloud includes ground, advanced vegetation, and buildings.

Images