Unmanned aerial vehicle intelligent surveying and mapping system based on remote sensing technology

Abstract

The invention relates to the technical field of surveying and mapping geographic information, in particular to an unmanned aerial vehicle intelligent surveying and mapping system based on a remote sensing technology, which comprises a remote sensing image acquisition module, an image correction splicing module, an ortho-rectification module, a ground feature recognition module, an attribute information input module, a map making module and a management database, according to the system, matching pairs are generated by obtaining all surface remote sensing images and extracting descriptors of all the surface remote sensing images, the corresponding relation of the same surface features among different images is determined, a basis is provided for image splicing, and a correction model is trained to convert coordinates of ground control points into image coordinates; the basic attribute table is created according to the basic attributes corresponding to the types of the ground features, the information amount and practicability of the map are improved, the elevation, obtained through interpolation calculation, of each interpolation point in the surveying and mapping area is combined to generate the complete map, the map can present ground feature distribution and can also reflect topographic relief, and the map rich in content is formed.

Description

Technical Field

[0001] The invention relates to the field of surveying and mapping geographic information technology, and in particular to an unmanned aerial vehicle intelligent surveying and mapping system based on remote sensing technology. Background Art

[0002] In today's era of rapid development of digitalization and informatization, the acquisition and processing of geographic spatial information plays a vital role in the development of many fields. From urban planning, land resource management, to agricultural monitoring, disaster emergency response, etc., accurate and timely surveying and mapping data are important bases for scientific decision-making. Traditional surveying and mapping technology is often inefficient when facing large areas and complex terrain areas, and it is difficult to meet the rapidly changing real needs. With the continuous innovation of remote sensing technology, drones have gradually become an emerging force in the field of surveying and mapping with their advantages of flexibility, efficiency and low cost.

[0003] However, existing surveying and mapping technologies have many limitations. For example, the existing Chinese patent with application number 202410198586.0 discloses a surveying and mapping system and method for dynamic remote sensing monitoring. This solution enables multiple different remote sensing images to have reference points by marking the map with data references, thereby achieving more accurate image stitching. At the same time, the reference points can provide a reference for the surveying and mapping data and the position trajectory of the drone, making it convenient for multiple drones to confirm the flight trajectory and the accurate stitching of subsequent remote sensing images; by setting up a base station group, it is used to obtain more accurate spatial data feedback from the drone, and refer to the feedback data with each other to calibrate the remote sensing images obtained by the drone and provide a reference for the surveying and mapping data.

[0004] However, this solution has the following shortcomings: when stitching remote sensing images taken by multiple drones to obtain a complete regional remote sensing image, there may be a problem of geographic coordinate mismatch. Although there are surveying and mapping marks as a reference, in large-scale surveying and mapping areas, due to factors such as image deformation, the geographic coordinates at the image stitching may be inaccurate, affecting the overall accuracy of the map.

[0005] For example, the existing Chinese patent with application number 202211304414.4 discloses a remote sensing camera stereo mapping method and system based on multivariate data fusion. This solution collects the outer contour image information of the survey object from bottom to top, the floor height and specific layout of a certain floor inside the survey object, and performs stereo modeling after data integration and processing. It solves the problem that the measurement data is single, resulting in the results of stereo mapping not being able to display all aspects of the survey object. The modeling results can greatly restore the body of the survey object, thereby improving the surveying and mapping effect.

[0006] However, this solution has the following shortcomings: this solution mainly focuses on the height of the survey object, the measurement of the floor height inside the survey object, the calculation of the area occupied by the survey object, etc., but does not take into account the elevation information of the location when collecting the image information of the outer contour of the survey object, which may affect the accuracy of the final three-dimensional modeling and cannot accurately restore the real spatial relationship between the survey object and the surrounding terrain. Summary of the invention

[0007] In order to overcome the shortcomings of the background technology, an embodiment of the present invention provides an unmanned aerial vehicle intelligent mapping system based on remote sensing technology, which can effectively solve the problems involved in the above-mentioned background technology.

[0008] The purpose of the present invention can be achieved through the following technical solutions: The present invention provides an unmanned aerial vehicle intelligent surveying and mapping system based on remote sensing technology, including: a remote sensing image acquisition module for setting unmanned aerial vehicle parameters, using unmanned aerial vehicles to acquire remote sensing images of various surfaces, and using various surface control points to perform geometric correction on various surface remote sensing images.

[0009] The image correction and stitching module is used to correct the surface remote sensing image through the timestamp correspondence, and extract the descriptors of each surface remote sensing image to generate matching pairs to form a complete remote sensing image of the surveying and mapping area.

[0010] The orthorectification module is used to train the correction model to convert the coordinates of the ground control points into image coordinates, and use the K-fold cross-validation method to adjust the model parameters to within the allowable error range.

[0011] The ground object recognition module is used to build a recognition model based on a convolutional neural network, identify and output the types of ground objects in the surveying area, and delineate the edge contours of each known type of ground object.

[0012] The attribute information entry module is used to create a basic attribute table, associate basic attributes with the objects in the remote sensing images of the surveying and mapping area, and enter the basic attributes corresponding to the types of objects.

[0013] The map making module generates point cloud data containing three-dimensional coordinate information through stereo matching algorithm, obtains the elevation of each interpolation point in the surveying and mapping area through interpolation calculation, forms a continuous surface model, and generates a complete map in combination with the ground object vector data.

[0014] Management database for storing surface remote sensing image data.

[0015] Preferably, the drone parameters include flight altitude and collection time interval.

[0016] Preferably, the remote sensing image acquisition module also includes: the drone takes surface remote sensing images at preset time intervals, the ground control point measurement equipment synchronously records the coordinates of the ground control points at the same time intervals, and marks timestamps on each acquired surface remote sensing image and each ground control point respectively.

[0017] The timestamps of each surface remote sensing image are matched with the timestamps of the ground control points. Through the corresponding relationship between the timestamps, the coordinates of each ground control point are accurately marked on the corresponding surface remote sensing image.

[0018] The corresponding surface remote sensing image is geometrically corrected through the coordinates of the ground control points.

[0019] Preferably, the specific analysis method of the image correction and stitching module is: extract feature points from each surface remote sensing image and generate descriptors, for each surface remote sensing image descriptor, calculate the distance between it and the descriptors of other surface remote sensing images, find the descriptor with the smallest distance from the surface remote sensing image descriptor in the remaining surface remote sensing images, mark this pair of descriptors as a matching pair, obtain each matching pair, set a distance threshold, filter out matching pairs with a distance greater than the threshold, use the filtered matching pairs to estimate the geometric transformation relationship between the surface remote sensing images, and stitch the geometrically corrected surface remote sensing images according to their positions and directions in the surveying and mapping area to form a complete remote sensing image of the surveying and mapping area.

[0020] Preferably, the specific analysis method of the orthorectification module is: obtain each surface remote sensing image with the coordinates of each ground control point imported, select a correction model according to the type of surface corresponding to the remote sensing image of the surveying area, import the coordinates of each ground control point into the correction model, and for any given ground control point coordinate, substitute it into the selected transformation model equation to obtain the corresponding image coordinates.

[0021] Calculate the error between the ground control point coordinates and their corresponding image coordinates. If the error is greater than the set error allowable range, divide the ground control point coordinates and their corresponding image coordinates into K subsets, each subset is used as a validation set in a certain iteration, and the remaining K-1 subsets are used as training sets. The training set is input into the correction model until each subset is used as a validation set once. Train the model and record the performance indicators on the validation set until the model performance of the correction model meets the preset requirements and the training is completed to obtain the correction model after adjusting the parameters.

[0022] The coordinate transformation is performed again using the corrected model after adjusting the parameters, and the transformation error is calculated again until the error is within the set error tolerance range.

[0023] Preferably, the specific analysis method of the land feature recognition module is: constructing a recognition model based on a convolutional neural network, acquiring a large amount of image data and dividing it into a training set and a test set, and inputting the training set into the recognition model for training until the loss function is stable.

[0024] The test set is input into the recognition model until the model parameters of the recognition model meet the preset requirements and the training is completed to obtain a trained recognition model.

[0025] The remote sensing image of the surveying area is input into the trained recognition model for recognition, so as to recognize various objects in the surveying area, and the types of various objects in the surveying area are recognized and output.

[0026] Preferably, the specific analysis method for delineating the edge contours of each known type of land object is: grayscale processing is performed on the remote sensing image of the surveying area, sample areas of each known land object type are selected therefrom, the grayscale values ​​of the pixels in the sample area of ​​each known land object type are calculated, the grayscale value range of the pixels in each sample area is obtained, and the middle value between the minimum and maximum grayscale values ​​is selected as the grayscale value threshold of the sample area of ​​each known land object type.

[0027] According to the gray value threshold of each known type of ground object sample area, the remote sensing image of the surveying and mapping area is binarized, and the pixels in the remote sensing image of the surveying and mapping area are divided into pixels of each known type of ground object and background pixels. The edge contour of each known type of ground object is extracted using the edge detection operator.

[0028] Preferably, the specific analysis method of the attribute information entry module is: converting the acquired types of various features in the surveying and mapping area into pixel values, acquiring the basic attributes of the types of various features, creating a basic attribute table, calculating the spatial distances between the coordinates of various features in the basic attribute table and the coordinates of various ground control points in the remote sensing image of the surveying and mapping area, establishing a mapping relationship between ground control points and features, vectorizing the remote sensing image of the surveying and mapping area, obtaining feature vector boundary data containing feature boundary information, using the vector boundary data as the target layer and the basic attribute table as the connection layer, matching according to the mapping relationship between ground control points and features, and entering the basic attributes corresponding to the types of various features.

[0029] Preferably, the specific operation method of the map making module is: through a stereo matching algorithm, for image pairs with overlapping areas in the remote sensing images of the surveying area, find the same-name ground object points, generate point cloud data containing three-dimensional coordinate information, and use the elevation information of each point extracted therefrom to classify them into ground points and non-ground points.

[0030] The surveying area is divided into grids, and each grid node is used as an interpolation point. A search radius is set, and a number of adjacent ground points are set within the search radius of each interpolation point according to the principle of equal spacing. The distance between each interpolation point and each adjacent ground point is calculated based on the point cloud data, and the weight of each adjacent ground point of each interpolation point is calculated. The elevation of each interpolation point is calculated by weighted average.

[0031] The elevations of each interpolation point are organized in the form of a regular grid to form a continuous surface model.

[0032] Preferably, the map making module also includes: superimposing and fusing the generated continuous surface model with the feature vector data containing the feature boundary information, connecting the basic attribute table with the feature vector data, associating the basic attribute table with the feature vector data through the unique identifier of the feature, and marking the attribute information of the feature to form a complete map.

[0033] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention obtains various surface remote sensing images, extracts the descriptors of various surface remote sensing images to generate matching pairs, determines the correspondence between the same objects in different images, provides a basis for image stitching, forms a complete remote sensing image of the surveying and mapping area, and trains a correction model to convert the coordinates of the ground control points into image coordinates, so that the image has accurate geographic coordinate information.

[0034] 2. The present invention creates a basic attribute table through the basic attributes corresponding to the types of local features, thereby improving the information content and practicality of the map, and generates a complete map by combining the elevations of each interpolation point in the surveying and mapping area obtained by interpolation calculation, so that the map not only presents the distribution of land features, but also reflects the undulations of the terrain, forming a map with rich content. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0036] Figure 1 This is a system module connection diagram of the present invention.

[0037] Figure 2 for Figure 1 Flowchart of the image correction and stitching module.

[0038] Figure 3 for Figure 1 Flowchart of the orthorectification module. DETAILED DESCRIPTION

[0039] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] See also Figure 1 As shown, an unmanned aerial vehicle intelligent mapping system based on remote sensing technology includes a remote sensing image acquisition module, an image correction and stitching module, an orthorectification module, a ground object recognition module, an attribute information entry module, a map making module, and a management database.

[0041] The management data is connected to the orthorectification module, the object recognition module, the attribute information entry module, and the map making module; the image correction and splicing module is connected to the remote sensing image acquisition module and the orthorectification module; and the attribute information entry module is connected to the object recognition module and the map making module.

[0042] The remote sensing image acquisition module is used to set UAV parameters, use the UAV to acquire remote sensing images of various surfaces, and use various surface control points to perform geometric correction on various surface remote sensing images.

[0043] The UAV parameters include flight altitude and collection time interval.

[0044] It should be noted that the specific analysis method of the flight height is: obtaining the focal length of the camera lens , Camera sensor pixel size and target ground resolution , substitute into the formula: Calculate flight altitude .

[0045] In a specific embodiment, the camera sensor pixel size is 5 microns, the lens focal length is 20 mm, and if the target image resolution requires a ground resolution of 5 cm (i.e. 0.05 m), the flight altitude can be obtained by substituting into the formula: rice.

[0046] The image correction and stitching module is used to correct the surface remote sensing image through the timestamp correspondence, and extract the descriptors of each surface remote sensing image to generate matching pairs to form a complete remote sensing image of the surveying and mapping area.

[0047] See also Figure 2As shown, the remote sensing image acquisition module also includes: the drone takes surface remote sensing images at preset time intervals, the ground control point measurement equipment synchronously records the coordinates of the ground control points at the same time intervals, and marks timestamps on each acquired surface remote sensing image and each ground control point respectively; the ground control point coordinates can be accurately associated with the corresponding image, so that the image has accurate geographic coordinate information, improves the geographic positioning accuracy of the image, and meets the needs of high-precision surveying and mapping.

[0048] Match the timestamps of each surface remote sensing image with the timestamps of the ground control points. Through the corresponding relationship between the timestamps, accurately mark the coordinates of each surface control point on the corresponding surface remote sensing image; ensure the accurate association between the ground control points and the image, so that the objects on the image can accurately correspond to the actual geographic coordinates, improve the accuracy of the image's geographic positioning, and meet the needs of high-precision surveying and mapping and geographic information analysis.

[0049] The corresponding surface remote sensing images are geometrically corrected through the coordinates of the ground control points; the geometric accuracy of the corrected images is improved, and the shapes, sizes and positions of the objects are more realistic, which is conducive to subsequent image interpretation, information extraction and analysis.

[0050] The specific analysis method of the image correction and splicing module is as follows: extracting feature points from various surface remote sensing images and generating descriptors; for each surface remote sensing image descriptor, calculating the distance between it and the descriptors of other surface remote sensing images, finding the descriptor with the smallest distance from the surface remote sensing image descriptor in the remaining surface remote sensing images, marking the pair of descriptors as a matching pair, obtaining each matching pair, setting a distance threshold, filtering out matching pairs with a distance greater than the threshold, using the filtered matching pairs to estimate the geometric transformation relationship between various surface remote sensing images, splicing the geometrically corrected surface remote sensing images according to their positions and directions in the surveying and mapping area, and combining them into a complete remote sensing image of the surveying and mapping area; the feature points and descriptors have a certain invariance to the geometric and radiation changes of the images, and can accurately find corresponding points under different viewing angles and lighting conditions, thereby improving the accuracy and reliability of matching.

[0051] It should be noted that the feature points are points with significant features in the surface remote sensing images, such as corner points and edge points. These feature points can remain relatively stable when the image is rotated, scaled, or the brightness changes. They are an important basis for image matching and geometric correction. For example, in the image of high-rise buildings in the city center, the points at the edges and corners of the high-rise buildings are feature points.

[0052] The descriptor is a quantitative representation of the local area features around the feature point, which is used to describe the characteristics of the feature point and facilitate the subsequent calculation of the similarity between the feature points. For example, the SIFT descriptor calculates the gradient direction histogram in the neighborhood of the feature point and generates a 128-dimensional vector to represent the feature point.

[0053] The matching pair is a pair of two feature points from different surface remote sensing images but with the smallest descriptor distance, which are considered to be corresponding points of the same object in different images and are used to establish the association between the images.

[0054] In a specific embodiment, there is a project for mapping a certain urban area, and 100 surface remote sensing images are obtained. The SIFT algorithm is selected to extract feature points from these images. On average, about 500 extreme points per image are used as feature points. Based on the extracted feature points, a gradient direction histogram is calculated in its neighborhood to obtain a 128-dimensional vector as a descriptor of the feature point. Since a floating-point vector such as the SIFT descriptor is used, the Euclidean distance is used to calculate the distance between the descriptor of each surface remote sensing image and all other surface remote sensing image descriptors. In the distance matrix, for each descriptor, the descriptor with the smallest distance from it among the remaining surface remote sensing image descriptors is found, and the feature points corresponding to the two descriptors are marked as a matching pair. By traversing all descriptors, a total of about 20,000 matching pairs are obtained. The distance threshold is set to 0.6, and all 20,000 matching pairs generated are traversed. The distance between each pair of descriptors is checked, and the matching pairs with a distance greater than 0.6 are filtered out to obtain about 15,000 matching pairs after screening.

[0055] The number of iterations is set to 500, and the minimum number of matching pairs is determined to be 4. Four matching pairs are randomly selected from 15,000 matching pairs in each iteration. Using these four matching pairs, the transformation matrix parameters are calculated according to the homography transformation model formula, and a 3×3 homography transformation matrix is ​​obtained by solving the linear equations. , use the remaining 14996 matching pairs to test the transformation matrix. For each remaining matching pair, map its coordinates in one image to the other image through the transformation matrix, calculate the error between the mapped coordinates and the actual matching point coordinates, and set the error threshold to 2 pixels. If the error is within the error threshold, the matching pair is considered to be an inlier. Count the number of inliers in this iteration. If the number of inliers this time is more than the maximum number of inliers recorded previously, update the optimal homography transformation matrix and its number of inliers.

[0056] After 500 iterations, the homography transformation matrix obtained from the iteration with the largest number of inliers is selected. The points on each surface remote sensing image are mapped to specific positions on the remote sensing image of the surveying and mapping area through the homography transformation matrix, thereby establishing a point-to-point correspondence. The geometrically corrected surface remote sensing images are then spliced ​​according to their positions and directions in the surveying and mapping area to form a complete remote sensing image of the surveying and mapping area.

[0057] The orthorectification module is used to train the correction model to convert the coordinates of the ground control points into image coordinates, and use the K-fold cross-validation method to adjust the model parameters to within the allowable error range.

[0058] See also Figure 3 As shown, the specific analysis method of the orthorectification module is: obtain each surface remote sensing image with the coordinates of each ground control point imported, select a correction model according to the type of the surface corresponding to the remote sensing image of the surveying area, import the coordinates of each ground control point into the correction model, and for any given ground control point coordinates, substitute them into the selected conversion model equation to obtain the corresponding image coordinates; accurate coordinate conversion provides a reliable basis for subsequent image stitching, information extraction and other processing, making the processing more accurate and effective.

[0059] The conversion model equation is a specific mathematical expression contained in the correction model. The specific mathematical expression includes the parameters of the model. The coordinates of the ground control point are substituted into the conversion model equation. The mapping rules defined by the equation are used to perform mathematical operations in the form of model equations to obtain the image coordinates corresponding to the ground control point.

[0060] It should be noted that the type of the surface is the category presented by the natural form or material composition of the earth's surface, which is mainly divided based on natural elements such as topography, geological structure, soil type, etc., including mountains, plains, plateaus, and basins.

[0061] Calculate the error between the coordinates of the ground control points and their corresponding image coordinates. If the error is greater than the set error allowable range, divide the coordinates of each ground control point and its corresponding image coordinates into K subsets, each subset is used as a validation set in a certain iteration, and the remaining K-1 subsets are used as training sets. The training set is input into the correction model until each subset is used as a validation set once. Train the model and record the performance indicators on the validation set until the model performance of the correction model meets the preset requirements. The training is completed to obtain the correction model after adjusting the parameters. The cross-validation method is used to divide the subsets for training and verification, which can more effectively evaluate the performance of the model on different data subsets, avoid model overfitting, and improve the stability and reliability of the model.

[0062] It should be noted that the specific analysis method of the error between the ground control point coordinates and the corresponding image coordinates is: the ground control point coordinates and the corresponding image coordinates are marked as , Indicates The number of the ground control point, , through the formula Get the error between the ground control point coordinates and the corresponding image coordinates , Indicates the number of ground control points.

[0063] For plane coordinates, The Euclidean distance formula between , which represents the straight-line distance between two points in the plane space. Here, it is the distance error between the coordinates of a single ground control point and its corresponding image coordinates. In order to measure the overall error, all ground control points need to be considered. First, the square sum of the error of each control point is calculated. , then find the average value, which is divided by the number of control points ,get , in order to make the final error value consistent with the Euclidean distance of a single point in terms of dimension, the square root operation is performed on the average value to obtain the formula .

[0064] Use the correction model with adjusted parameters to re-perform coordinate transformation and recalculate the transformation error until the error is within the set error tolerance range; this can ensure that the final coordinate transformation result has high reliability and provide more accurate data support for subsequent image processing and analysis.

[0065] The allowable error range is set according to the surface type corresponding to the remote sensing image of the surveying area, >2 pixels in plain areas and >3 pixels in mountainous areas.

[0066] The ground object recognition module is used to build a recognition model based on a convolutional neural network, identify and output the types of ground objects in the surveying area, and delineate the edge contours of each known type of ground object.

[0067] The specific analysis method of the ground feature recognition module is: building a recognition model based on a convolutional neural network, acquiring a large amount of image data and dividing it into a training set and a test set, and inputting the training set into the recognition model for training until the loss function is stable; avoiding the tedious process and subjectivity of manually designed features, reducing the impact of human factors on feature extraction, and improving the efficiency and accuracy of feature extraction.

[0068] It should be noted that the convolutional neural network includes an input layer, a convolution layer, a pooling layer, a fully connected layer, and an output layer. The input layer receives image data; the convolution layer performs convolution operations by sliding convolution kernels of different sizes on the image to extract local features in the image; the pooling layer downsamples the feature map to reduce the amount of data and calculation, while retaining the main features to prevent overfitting; the fully connected layer expands the feature map after convolution and pooling processing into a one-dimensional vector, and then performs feature fusion and classification decisions through multiple fully connected layers; the output layer: set according to the specific recognition task.

[0069] It should be noted that the specific operation process until the loss function is stable is: in each training iteration, a batch of image data in the training set is input into the recognition model, the model predicts according to the current parameters to obtain the prediction result, and the loss value between the prediction result and the true label is calculated through the loss function. Based on the calculated loss value, the gradient of the loss function to the model parameters is calculated, and the model parameters are adjusted according to the direction and size of the gradient. As the training progresses, the value of the loss function will continue to change. When the value of the loss function meets the requirements in multiple consecutive training iterations, for example, the difference between the loss values ​​of two adjacent iterations is less than a preset threshold, or the loss value fluctuates within a certain range and no longer decreases significantly, it can be considered that the loss function has reached a stable state.

[0070] The test set is input into the recognition model until the model parameters of the recognition model meet the preset requirements and the training is completed to obtain a trained recognition model; through repeated verification and adjustment, the parameters and structure of the model are continuously optimized to make the model reach the best state in terms of accuracy, stability, etc., thereby improving the effect of ground object recognition.

[0071] The remote sensing image of the surveying area is input into the trained recognition model for recognition, so as to identify the local features in the surveying area and output the types of local features in the surveying area; compared with manual visual interpretation of remote sensing images, using the trained model for ground feature recognition has higher efficiency and speed. A large amount of image data can be processed in a short time, improving work efficiency.

[0072] It should be noted that the land features are various natural and artificial fixed objects on the earth's surface, including lakes, forests, buildings, roads, and farmlands.

[0073] The specific analysis method for delineating the edge contours of each known type of land object is as follows: grayscale processing is performed on the remote sensing image of the surveying area, sample areas of each known land object type are selected therefrom, and the grayscale value of each pixel in the sample area of ​​each known land object type is obtained, thereby obtaining the grayscale value range of the pixels in the sample area of ​​each known land object type, and the middle value between the minimum and maximum grayscale values ​​is selected as the grayscale value threshold of the sample area of ​​each known land object type; by selecting the sample area and calculating the grayscale value range and threshold, the unique grayscale features of different land object types can be extracted, which provides a basis for accurately identifying and distinguishing land objects.

[0074] According to the gray value threshold of each known type of land object sample area, the remote sensing image of the surveying and mapping area is binarized, and the pixels in the remote sensing image of the surveying and mapping area are divided into pixels of each known type of land object and background pixels. The edge detection operator is used to extract the edge contour of each known type of land object. Edge contour extraction can accurately obtain the shape information of the land object, which provides an important basis for the further identification and classification of the land object.

[0075] The attribute information entry module is used to create a basic attribute table, associate basic attributes with the objects in the remote sensing images of the surveying and mapping area, and enter the basic attributes corresponding to the types of objects.

[0076] The specific analysis method of the attribute information entry module is as follows: converting the acquired types of various features in the surveying and mapping area into pixel values, acquiring the basic attributes of the types of various features, creating a basic attribute table, calculating the spatial distances between the coordinates of various features in the basic attribute table and the coordinates of various ground control points in the remote sensing image of the surveying and mapping area, establishing a mapping relationship between ground control points and features, vectorizing the remote sensing image of the surveying and mapping area, obtaining feature vector boundary data containing feature boundary information, using the vector boundary data as the target layer and the basic attribute table as the connection layer, matching according to the mapping relationship between ground control points and features, and entering the basic attributes corresponding to the types of various features; the unified pixel value representation makes the data structure more regular, reduces the complexity of processing and improves the calculation efficiency when performing data processing and analysis, and with the help of the high-precision coordinates of the ground control points, improves the positioning accuracy of the features in the surveying and mapping area, and makes the location information of the features more accurate and reliable.

[0077] It should be noted that the vector boundary data is composed of three basic geometric elements: points, lines and surfaces. Points are used to determine key locations. Lines are composed of a series of continuous points and are used to represent the boundaries of linear features (such as roads and rivers) or surface features. Multiple line elements are closed to form a surface, representing surface features (such as lakes and building areas).

[0078] It should be noted that the coordinates of the ground objects in the basic attribute table and the coordinates of the ground control points of the remote sensing image of the surveying area are recorded as , Indicates The number of the feature, , Indicates The number of the ground control point, The calculation formula for the spatial distance between the coordinates of each feature and the coordinates of each ground control point in the remote sensing image of the surveying area is: .

[0079] The formula for the spatial distance is derived based on the extension of the Pythagorean theorem. In a two-dimensional plane, two points The distance formula is This is a direct application of the Pythagorean theorem. The horizontal and vertical distances between two points are squared and squared to get the straight-line distance. When extended to three-dimensional space, the z-dimension coordinate is added. Suppose there are two points , we can first calculate the distance between the projection points of these two points on the xy plane , and then take the distance in the z direction into consideration. According to the geometric relationship of three-dimensional space and the Pythagorean theorem, we can get the distance formula between the two points in three-dimensional space: .

[0080] It should be noted that, in a specific embodiment, the corresponding pixel value of vegetation-related features is set to 1, the corresponding pixel value of water bodies is set to 2, and the corresponding pixel value of buildings is set to 3. The type of each feature is determined, and then the type of each feature is replaced with the corresponding pixel value according to the set conversion standard.

[0081] Please refer to Table 1 for details, which lists some basic attributes with practical significance.

[0082] Table 1. Basic attribute table

[0083]

[0084] It should be noted that the feature ID is a code used to uniquely identify each feature object, that is, a unique identifier of the feature.

[0085] The map making module generates point cloud data containing three-dimensional coordinate information through stereo matching algorithm, obtains the elevation of each interpolation point in the surveying and mapping area through interpolation calculation, forms a continuous surface model, and generates a complete map in combination with the ground object vector data.

[0086] The specific operation method of the map making module is as follows: through the stereo matching algorithm, for the image pairs with overlapping areas in the remote sensing images of the surveying and mapping area, find the same-name ground object points, generate point cloud data containing three-dimensional coordinate information, and use the elevation information of each point extracted therefrom to classify them as ground points and non-ground points; accurate matching of the same-name ground object points can ensure that the calculated three-dimensional coordinates of the ground object are more accurate, thereby improving the three-dimensional reconstruction accuracy of the entire surveying and mapping area, so that the generated map can more realistically reflect the actual terrain and the spatial position of the ground object.

[0087] It should be noted that the specific analysis method of the stereo matching algorithm is: split the images with overlapping areas in the remote sensing images of the surveying area into two independent images, select a small window centered on a certain pixel point in one of the images, and search for the window with the highest similarity to the window within a set range of the other image, and use the central pixel point of the window with the highest similarity as the same-name ground feature point corresponding to the central pixel point of the window in the first image.

[0088] The specific analysis method of the similarity is: suppose the window selected from the first image is , whose pixel value , Represents the coordinates of the pixels within the window. The window within the search range in the second image is , the pixel value is , substitute into the formula Get the similarity between two windows.

[0089] It should be noted that the similarity calculation formula is based on comparing the pixel value differences of corresponding pixels in the two windows and calculating their difference , in order to avoid the situation where the positive and negative differences cancel each other out and to highlight the impact of larger differences, we square the differences, that is, , summing up the square differences of all pixels in the window, we get .

[0090] The surveying area is divided into grids, and each grid node is used as an interpolation point. A search radius is set, and a number of adjacent ground points are set within the search radius of each interpolation point according to the principle of equal spacing. The distance between each interpolation point and each adjacent ground point is calculated based on the point cloud data, and the weight of each adjacent ground point of each interpolation point is calculated. The elevation of each interpolation point is calculated by weighted average.

[0091] It should be noted that the specific analysis method of the elevation of each interpolation point is: calculate the weight of each adjacent ground point of each interpolation point: ,in Indicates The number of interpolation points, , Indicates The numbers of the adjacent ground points, , Indicates The interpolation point and its The distance between adjacent ground points is calculated by weighted average to obtain the elevation of each interpolation point. ,in Indicates Interpolation point The elevation of a nearby ground point.

[0092] It should be noted that the weight formula of each interpolation point and each adjacent ground point uses the inverse square of the distance as the weight. The larger the square, the The smaller it is, the closer the ground point is to the interpolation point. On the contrary, the closer the distance is, the greater the weight is, and the greater the influence on the elevation of the interpolation point is; in the formula for the elevation of each interpolation point, the numerator is is the elevation of each adjacent ground point The corresponding weight The multiplication and summation reflects the weighted contribution of each adjacent point to the interpolation point elevation. is the sum of the weights of all nearby ground points, which plays a normalization role and ensures that the calculated is a reasonable weighted average. In this way, the interpolation points Elevation The elevation of each nearby ground point and their distance relationship to the interpolation point are comprehensively considered.

[0093] The elevations of each interpolation point are organized in the form of a regular grid to form a continuous surface model.

[0094] The map making module also includes: superimposing and fusing the generated continuous surface model with the feature vector data containing feature boundary information, connecting the basic attribute table with the feature vector data, associating the basic attribute table with the feature vector data through the unique identifier of the feature, marking the feature attribute information to form a complete map; connecting the basic attribute table with the feature vector data to facilitate users to query and analyze the feature attribute information, thereby improving the interactivity and practicality of the map.

[0095] It should be noted that the feature vector data is a data type that represents features in geographic space in vector form, and it accurately describes the location, shape, boundary and area of ​​features through geometric elements such as points, lines and surfaces.

[0096] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations on the present invention. A person skilled in the art may make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention and they are still covered by the protection scope of the present invention.

Claims

1. An unmanned aerial vehicle intelligent mapping system based on remote sensing technology, characterized in that: The system specifically includes the following modules: The remote sensing image acquisition module is used to set the parameters of the drone, use the drone to acquire remote sensing images of various surfaces, and use various surface control points to perform geometric correction on the remote sensing images of various surfaces; The image correction and stitching module is used to correct the surface remote sensing image through the time stamp correspondence, and extract the descriptors of each surface remote sensing image to generate matching pairs to form a complete remote sensing image of the surveying and mapping area; The orthorectification module is used to train the correction model to convert the coordinates of the ground control points into image coordinates, and use the K-fold cross-validation method to adjust the model parameters within the allowable error range; A ground object recognition module, which is used to construct a recognition model based on a convolutional neural network, recognize and output the types of ground objects in the surveying and mapping area, and delineate the edge contours of each known type of ground object; The attribute information entry module is used to create a basic attribute table, associate basic attributes with the objects in the remote sensing images of the surveying area, and enter the basic attributes corresponding to the types of objects in each area; The map making module generates point cloud data containing three-dimensional coordinate information through stereo matching algorithm, obtains the elevation of each interpolation point in the surveying area through interpolation calculation, forms a continuous surface model, and generates a complete map in combination with the ground object vector data; Management database for storing surface remote sensing image data.

2. The UAV intelligent mapping system based on remote sensing technology according to claim 1 is characterized in that: The UAV parameters include flight altitude and collection time interval.

3. The UAV intelligent mapping system based on remote sensing technology according to claim 1 is characterized in that: The remote sensing image acquisition module also includes: The drone takes surface remote sensing images at preset time intervals, and the ground control point measurement equipment records the coordinates of the ground control points at the same time intervals, marking timestamps on each surface remote sensing image and each ground control point respectively; Match the timestamps of each surface remote sensing image with the timestamps of the ground control points, and accurately mark the coordinates of each ground control point on the corresponding surface remote sensing image through the corresponding relationship of the timestamps; The corresponding surface remote sensing image is geometrically corrected through the coordinates of the ground control points.

4. The UAV intelligent mapping system based on remote sensing technology according to claim 1 is characterized in that: The specific analysis method of the image correction and splicing module is as follows: Feature points are extracted from each surface remote sensing image and descriptors are generated. For each surface remote sensing image descriptor, the distance between it and the descriptors of other surface remote sensing images is calculated, and the descriptor with the smallest distance to the surface remote sensing image descriptor in the remaining surface remote sensing images is found. This pair of descriptors is marked as a matching pair, and each matching pair is obtained. A distance threshold is set, and matching pairs with a distance greater than the threshold are filtered out. The filtered matching pairs are used to estimate the geometric transformation relationship between the surface remote sensing images, and the geometrically corrected surface remote sensing images are spliced ​​according to their positions and directions in the surveying and mapping area to form a complete remote sensing image of the surveying and mapping area.

5. The UAV intelligent mapping system based on remote sensing technology according to claim 1 is characterized in that: The specific analysis method of the orthorectification module is as follows: Obtain each surface remote sensing image with the coordinates of each ground control point imported, select a correction model according to the type of the surface corresponding to the remote sensing image of the surveying area, import the coordinates of each ground control point into the correction model, and for any given ground control point coordinate, substitute it into the selected transformation model equation to obtain the corresponding image coordinate; Calculate the error between the coordinates of the ground control points and their corresponding image coordinates. If the error is greater than the set error tolerance range, divide the coordinates of each ground control point and its corresponding image coordinates into K subsets, each subset is used as a validation set in a certain iteration, and the remaining K-1 subsets are used as training sets. The training sets are input into the correction model until each subset is used as a validation set once. The model is trained and the performance indicators on the validation set are recorded until the model performance of the correction model meets the preset requirements and the training is completed to obtain the correction model after adjusting the parameters. The coordinate transformation is performed again using the corrected model after adjusting the parameters, and the transformation error is calculated again until the error is within the set error tolerance range.

6. The UAV intelligent mapping system based on remote sensing technology according to claim 1 is characterized in that: The specific analysis method of the ground object recognition module is: Building a recognition model based on a convolutional neural network, acquiring a large amount of image data and dividing it into a training set and a test set, and inputting the training set into the recognition model for training until the loss function is stable; Inputting the test set into the recognition model until the model parameters of the recognition model meet the preset requirements and the training is completed, so as to obtain a trained recognition model; The remote sensing image of the surveying area is input into the trained recognition model for recognition, so as to recognize various objects in the surveying area, and the types of various objects in the surveying area are recognized and output.

7. The UAV intelligent mapping system based on remote sensing technology according to claim 6 is characterized in that: The specific analysis method for delineating the edge contours of each known type of land feature is: Grayscale processing is performed on the remote sensing images of the surveying and mapping area, and sample areas of each known land object type are selected therefrom. The grayscale values ​​of the pixels in the sample areas of each known land object type are calculated to obtain the grayscale value range of the pixels in each sample area, and the middle value between the minimum and maximum grayscale values ​​is selected as the grayscale value threshold of the sample areas of each known land object type; According to the gray value threshold of each known type of ground object sample area, the remote sensing image of the surveying and mapping area is binarized, and the pixels in the remote sensing image of the surveying and mapping area are divided into pixels of each known type of ground object and background pixels. The edge contour of each known type of ground object is extracted using the edge detection operator.

8. The UAV intelligent mapping system based on remote sensing technology according to claim 1 is characterized in that: The specific analysis method of the attribute information entry module is: The types of various features in the surveying and mapping area are converted into pixel values, the basic attributes of the types of various features are obtained, a basic attribute table is created, the spatial distances between the coordinates of various features in the basic attribute table and the coordinates of various ground control points in the remote sensing image of the surveying and mapping area are calculated, a mapping relationship between ground control points and features is established, the remote sensing image of the surveying and mapping area is vectorized, and feature vector boundary data containing feature boundary information is obtained. The vector boundary data is used as the target layer, and the basic attribute table is used as the connection layer. Matching is performed according to the mapping relationship between ground control points and features, and the basic attributes corresponding to the types of various features are entered.

9. The UAV intelligent mapping system based on remote sensing technology according to claim 1 is characterized in that: The specific operation method of the map making module is as follows: Through the stereo matching algorithm, for the image pairs with overlapping areas in the remote sensing images of the surveying area, find the same-name ground object points, generate point cloud data containing three-dimensional coordinate information, and use the elevation information of each point extracted from it to classify it into ground points and non-ground points; The surveying area is divided into grids, and each node of the grid is used as an interpolation point. The search radius is set. According to the principle of equal spacing, a number of adjacent ground points are set within the search radius of each interpolation point. The distance between each interpolation point and each adjacent ground point is calculated based on the point cloud data. The weight of each adjacent ground point of each interpolation point is calculated, and the elevation of each interpolation point is calculated by weighted average. The elevations of each interpolation point are organized in the form of a regular grid to form a continuous surface model.

10. The UAV intelligent mapping system based on remote sensing technology according to claim 1 is characterized in that: The map making module also includes: The generated continuous surface model is superimposed and fused with the feature vector data containing the feature boundary information, the basic attribute table is connected with the feature vector data, the basic attribute table is associated with the feature vector data through the unique identifier of the feature, the attribute information of the feature is marked, and a complete map is formed.

Images