High-resolution digital elevation model generation processing method

Abstract

The invention relates to the technical field of digital elevation model generation, and provides a high-resolution digital elevation model generation processing method. Comprising the following steps: acquiring a low-resolution digital elevation model of a to-be-generated region of a high-resolution digital elevation model, and a high-resolution digital elevation model and a low-resolution digital elevation model of a region with the same address type as the to-be-generated region, and constructing a training data set; establishing a feature value loss function and taking the feature value loss function as a loss function, training the convolutional neural network by using the training data set, obtaining a feature map extraction neural network, obtaining a feature map of the to-be-generated region by using the feature map extraction neural network, and dividing feature data blocks; and determining a feature difference function, taking the feature difference function as an objective function of a least square method, obtaining reconstructed high-resolution digital elevation data of the to-be-generated region by using the least square method, and obtaining a reconstructed digital elevation model of the to-be-generated region. The invention aims to generate a more accurate high-resolution digital elevation model.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of digital elevation model generation, in particular to a high-resolution digital elevation model generation processing method. BACKGROUND

[0002] A high-resolution digital elevation model DEM is a spatial data model based on geographic coordinates, which stores elevation information in the form of regular or irregular grids, and is used to represent the topographic surface morphology of the earth's surface or a specific region. Compared with a low-resolution digital elevation model, a high-resolution digital elevation model can capture more subtle topographic features such as ridges, valleys and riverbeds, but the equipment cost, time cost and data acquisition cost of a high-resolution digital elevation model are relatively high.

[0003] In order to obtain a high-resolution digital elevation model at a lower cost, a neural network can be used to reconstruct a low-resolution digital elevation model to obtain a corresponding high-resolution digital elevation model. However, the prior art does not consider the influence of the topographic features in the digital elevation model on the reconstructed high-resolution digital elevation model in the process of training the neural network, which often leads to poor reconstruction results. SUMMARY

[0004] The application provides a high-resolution digital elevation model generation processing method to solve the problem of poor reconstruction results caused by not considering the influence of digital topographic features on the reconstruction process in the process of reconstructing a high-resolution digital elevation model from a low-resolution digital elevation model. The technical solution adopted is as follows:

[0005] One embodiment of the application provides a high-resolution digital elevation model generation processing method, which comprises the following steps:

[0006] Collecting a low-resolution digital elevation model of a to-be-generated area of a high-resolution digital elevation model, a high-resolution digital elevation model of a preset number of regions of the same type as the to-be-generated area and a low-resolution digital elevation model, obtaining a gradient map, a curvature map and a valley line elevation map from all the high-resolution digital elevation models and low-resolution digital elevation models of the same region, dividing data blocks, and constructing a training data set;

[0007] Establishing a feature value loss function according to the training data set, using the feature value loss function as a loss function, training a convolutional neural network using the training data set, obtaining a feature map extraction neural network, using the feature map extraction neural network to obtain the feature maps of the gradient map, the curvature map and the valley line elevation map of the low-resolution digital elevation model of the to-be-generated area, and dividing all the feature maps of the to-be-generated area into equal-sized feature data blocks, wherein each feature data block corresponds to a data block;

[0008] According to the feature data block contained in the feature map and the elevation value of the pixel point in the data block corresponding to the feature data block, the type of the feature data block is determined, and the gradient weight, the curvature weight and the elevation value weight of the feature data block are respectively valued, according to the difference between the low-resolution digital elevation model of the region to be generated and the high-resolution digital elevation model generated according to the low-resolution digital elevation model, and the gradient weight, the curvature weight and the elevation value weight of the feature data block, the feature difference function is determined, the feature difference function is taken as the objective function of the least square method, and the reconstruction high-resolution digital elevation data of the region to be generated is obtained by using the least square method, and the digital elevation model is generated according to the reconstruction high-resolution digital elevation data, and the reconstruction digital elevation model of the region to be generated is obtained.

[0009] Further, the gradient map, the curvature map and the valley line elevation map are obtained by:

[0010] The gradient map, the curvature map and the elevation map of each high-resolution digital elevation model and low-resolution digital elevation model corresponding to the same region are obtained respectively;

[0011] The valley line in each high-resolution digital elevation model and low-resolution digital elevation model corresponding to the same region is obtained respectively, the pixel points not contained in the valley line in the high-resolution digital elevation model and the pixel points contained in the valley line are identified in the elevation map, the elevation value of the pixel points not contained in the valley line in the elevation map is valued as 0, the elevation value of the pixel points contained in the valley line in the high-resolution digital elevation model and the elevation map is retained, and the valley line elevation map of the high-resolution digital elevation model is obtained, the elevation value of the pixel points not contained in the valley line in the elevation map is valued as 0, the elevation value of the pixel points contained in the valley line in the low-resolution digital elevation model and the low-resolution digital elevation model is retained, and the valley line elevation map of the low-resolution digital elevation model is obtained.

[0012] Further, the data block and the training data set are obtained by:

[0013] The gradient map, the curvature map and the valley line elevation map of all high-resolution digital elevation models and low-resolution digital elevation models corresponding to the same region are divided into data blocks of a predetermined size;

[0014] The data set composed of all data blocks is recorded as a training data set.

[0015] Further, the feature value loss function is specifically:

[0016]

[0017] In the formula, L represents the feature value loss function, n represents the number of pixel points contained in the gradient map of the high-resolution digital elevation model, G i represents the gradient value of the i-th pixel point in the gradient map of the high-resolution digital elevation model. G i represents the gradient value of the i th pixel point in the gradient map of the low-resolution digital elevation model in the corresponding pixel point in the corresponding feature map; G max C represents the maximum value of the gradient values of all pixel points in the gradient map of the high-resolution digital elevation model; C i K i represents the curvature value of the i th pixel point in the curvature map of the high-resolution digital elevation model; G i represents the gradient value of the i th pixel point in the curvature map of the low-resolution digital elevation model in the corresponding pixel point in the corresponding feature map; C max C represents the maximum value of the curvature values of all pixel points in the curvature map of the high-resolution digital elevation model; T i T i represents the elevation value of the i th pixel point in the valley line elevation map of the high-resolution digital elevation model; G i represents the gradient value of the i th pixel point in the valley line elevation map of the low-resolution digital elevation model in the corresponding pixel point in the corresponding feature map; T max T represents the maximum value of the elevation values of all pixel points in the valley line elevation map of the high-resolution digital elevation model.

[0018] Further, the training of the convolutional neural network using the training data set to obtain the feature map extraction neural network comprises the following specific method:

[0019] The training data set is divided into a training set, a validation set and a test set according to a ratio of 7:2:1, and the convolutional neural network is trained to obtain the feature map extraction neural network; wherein all data blocks in the gradient map, the curvature map and the valley line elevation map corresponding to each low-resolution digital elevation model in the training data set are respectively taken as the input of the feature map extraction neural network, and the feature map extraction neural network outputs the feature map corresponding to the gradient map, the curvature map and the valley line elevation map of the low-resolution digital elevation model respectively.

[0020] Further, the method for determining the type of the feature data block according to the feature data block contained in the feature map and the elevation values of the pixel points in the data block corresponding to the feature data block comprises the following specific method:

[0021] The gradient of all pixel points in the feature data block of the feature map is calculated, and the ratio of the average value of the gradient of all pixel points in the feature data block of the feature map to the standard deviation is denoted as the gradient variation coefficient of the feature data block;

[0022] The curvature of all pixel points in the feature data block of the feature map is calculated, and the skewness of the curvature of all pixel points in the feature data block of the feature map is denoted as the curvature skewness of the feature data block;

[0023] determining the number of pixel points with elevation value not being 0 in the data block corresponding to the feature data block as the elevation non-zero number of the feature data block, determining the number of all pixel points in the data block corresponding to the feature data block as the elevation total number of the feature data block, and determining the ratio of the elevation non-zero number to the elevation total number as the valley line density of the feature data block;

[0024] determining the type of the feature data block according to the gradient variation coefficient, the curvature skewness and the valley line density of the feature data block.

[0025] Further, the method of determining the type of the feature data block according to the gradient variation coefficient, the curvature skewness and the valley line density of the feature data block comprises the following specific steps:

[0026] marking the type of the feature data block with the gradient variation coefficient greater than 0.7 as a steep slope;

[0027] marking the type of the feature data block with the gradient variation coefficient less than 0.7, the curvature skewness less than -0.4 and the valley line density greater than 0.35 as a river junction area;

[0028] marking the type of the feature data block with the gradient variation coefficient less than 0.4, the curvature skewness less than -0.3 and the valley line density greater than 0.1 and less than 0.35 as a gentle valley bottom;

[0029] marking the type of the feature data block with the gradient variation coefficient greater than 0.4 and less than 0.6, the curvature skewness greater than 0.5 and the valley line density less than 0.05 as a ridge;

[0030] marking the type of all the remaining feature data blocks as a common area.

[0031] Further, the method of assigning the gradient weight, the curvature weight and the elevation value weight of the feature data block respectively comprises the following specific steps:

[0032] when the type of the feature data block is marked as a steep slope, assigning the gradient weight, the curvature weight and the elevation value weight corresponding to the gradient variation coefficient, the curvature skewness and the valley line density of the feature data block respectively as 0.7, 0.2 and 0.1;

[0033] when the type of the feature data block is a river junction area, assigning the gradient weight, the curvature weight and the elevation value weight corresponding to the gradient variation coefficient, the curvature skewness and the valley line density of the feature data block respectively as 0.2, 0.2 and 0.6;

[0034] When the category mark of the feature data block is marked as a gentle valley, the gradient weight, curvature weight and elevation value weight corresponding to the gradient variation coefficient, curvature skewness and valley line density of the feature data block are respectively assigned as 0.1, 0.45 and 0.45;

[0035] When the category mark of the feature data block is marked as a mountain ridge, the gradient weight, curvature weight and elevation value weight corresponding to the gradient variation coefficient, curvature skewness and valley line density of the feature data block are respectively assigned as 0.45, 0.45 and 0.1;

[0036] When the category mark of the feature data block is marked as a general area, the gradient weight, curvature weight and elevation value weight corresponding to the gradient variation coefficient, curvature skewness and valley line density of the feature data block are respectively assigned as 0.33, 0.33 and 0.34.

[0037] Further, the feature difference function is specifically:

[0038]

[0039] In the formula, indicates the feature difference function; argmin{} indicates the minimum value function argmin function; indicates the L2 norm; X indicates the low-resolution digital elevation model of the region to be generated; Y indicates the reconstructed digital elevation model of the region to be generated; Y↓ indicates the reconstructed digital elevation model of the region to be generated after downsampling; ω u,1 indicates the gradient weight of the u-th feature data block; ω u,2 indicates the curvature weight of the u-th feature data block; ω u,3 indicates the elevation value weight of the u-th feature data block; N indicates the number of data blocks divided by the gradient graph of the low-resolution digital elevation model of the region to be generated; G u,1 indicates the u-th feature data block in the gradient graph of the low-resolution digital elevation model of the region to be generated; G u,2 indicates the u-th feature data block in the curvature graph of the low-resolution digital elevation model of the region to be generated; G u,3 indicates the u-th feature data block in the valley line elevation graph of the low-resolution digital elevation model of the region to be generated; indicates the u-th feature data block in the gradient graph of the reconstructed digital elevation model of the region to be generated; indicates the u-th feature data block in the curvature graph of the reconstructed digital elevation model of the region to be generated; indicates the u-th feature data block in the valley line elevation graph of the reconstructed digital elevation model of the region to be generated.

[0040] Further, the feature difference function is taken as the objective function of the least square method, and the reconstruction high-resolution digital elevation data of the to-be-generated area is obtained by using the least square method, including the specific method that:

[0041] The feature difference function is taken as the objective function of the least square method, and the feature difference function is taken as the objective function of the least square method, and the optimal solution is sought by using the least square method, so that the difference between the obtained reconstruction digital elevation model of the to-be-generated area and the low-resolution digital elevation model of the to-be-generated area is minimized, and the reconstruction high-resolution digital elevation data of the to-be-generated area is obtained.

[0042] The beneficial effects of the present application are:

[0043] The present application collects the high-resolution digital elevation model and the low-resolution digital elevation model of the area with the same geological type as the to-be-generated area, and extracts the gradient graph, the curvature graph and the valley line elevation graph according to the corresponding high-resolution and low-resolution digital elevation models of the area with the same address type as the to-be-generated area, wherein the gradient graph, the curvature graph and the valley line elevation graph extract the features of the corresponding high-resolution and low-resolution digital elevation models of the same area from the spatial intensity variation, the morphological distribution law and the terrain connectivity caused by the terrain slope of the address type of the to-be-generated area; then, the loss function is constructed according to the gradient loss, the curvature loss and the elevation loss of the valley line respectively, and the convolutional neural network is trained using the training data set and the loss function to obtain the feature map extraction neural network, which can extract the feature map of the gradient graph, the curvature graph and the valley line elevation graph of the low-resolution digital elevation model of the to-be-generated area; further, the geological features of each feature data block are analyzed according to the features extracted from the feature map, the types of the feature data blocks are determined, and different weights are assigned to the differences between each feature data block and the corresponding data block according to the types of the feature data blocks, and the optimal solution is obtained by using the least square method to obtain the reconstruction high-resolution digital elevation data; finally, the digital elevation model is generated according to the reconstruction high-resolution digital elevation data, and the reconstruction digital elevation model of the to-be-generated area is obtained, solving the problem that the influence of the digital terrain features on the reconstruction process is not considered in the process of reconstructing the high-resolution digital elevation model from the low-resolution digital elevation model, resulting in poor reconstruction effect. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments or the prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0045] Figure 1A high-resolution digital elevation model generation processing method flowchart provided by an embodiment of the present application;

[0046] Figure 2 A data block acquisition flowchart provided by an embodiment of the present application. DETAILED DESCRIPTION

[0047] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0048] Please refer to Figure 1 which shows a high-resolution digital elevation model generation processing method flowchart provided by an embodiment of the present application. The method comprises the following steps:

[0049] Step S001, collect a low-resolution digital elevation model of a region to be generated of a high-resolution digital elevation model, a preset number of high-resolution digital elevation models of regions with the same address type as the region to be generated and a low-resolution digital elevation model, acquire a gradient map, a curvature map and a valley line elevation map according to all the high-resolution digital elevation models and the low-resolution digital elevation models of the corresponding regions, and divide data blocks to construct a training data set.

[0050] Through the USGS website, a low-resolution digital elevation model of a region to be generated of a high-resolution digital elevation model is extracted, and a plurality of high-resolution digital elevation models and low-resolution digital elevation models of corresponding regions are collected.

[0051] It should be noted that the corresponding regions of the corresponding high-resolution digital elevation models and low-resolution digital elevation models should be of the same geological type as the region to be generated of the high-resolution digital elevation model.

[0052] Preferably, in an embodiment of the present application, 500 high-resolution digital elevation models and low-resolution digital elevation models of regions with the same geological type as the region to be generated are collected; the geological types of all the regions collected in the present embodiment and the region to be generated are all valleys.

[0053] Compared with other geological types of regions, the valley terrain has more obvious landform features. The overall shape of the valley is a long strip of depression, accompanied by ridges on both sides, and complex water flow paths are distributed; the valley side height is steep, and the slope mutation degree is large, the valley bottom is narrow and relatively flat, and there are natural landforms such as streams in the valley bottom, which presents a concave shape; the ridges on both sides of the valley generally present a gradually widening and thickening shape from top to bottom, and are convex curved surfaces due to the accumulation effect; there are main valleys and straight streams in the valley, forming a continuous dendritic network.

[0054] According to these morphological characteristics of the valley, the gradient feature is used to describe the slope change, which can accurately quantify the spatial intensity change caused by the terrain slope, and effectively describe the mutation caused by the transition of the valley side and the valley bottom; the curvature feature can accurately describe the morphological features of the naturally formed concave valley bottom and convex ridge, and distinguish and quantify the concave-convex morphological distribution rule; according to the valley line feature, the complete skeleton of the valley drainage network is constrained, which can ensure the terrain connectivity in the reconstruction process, and solve the problem of small straight stream breakage and destruction of the continuous topological structure of the valley in the low-resolution digital elevation model.

[0055] The Sobel operator is used to obtain the gradient map of each corresponding high-resolution digital elevation model and low-resolution digital elevation model of the same region; the ArcGIS tool is used to obtain the curvature map of each corresponding high-resolution digital elevation model and low-resolution digital elevation model of the same region; the valley line in each corresponding high-resolution digital elevation model and low-resolution digital elevation model of the same region is obtained by hydrological analysis, and the elevation map corresponding to the high-resolution digital elevation model and the low-resolution digital elevation model is obtained by the geographic information system (GIS) software, and the pixel points not contained in the valley line in the high-resolution digital elevation model and the pixel points contained in the valley line are identified in the elevation map, the elevation value of the pixel points not contained in the valley line identified in the elevation map is assigned as 0, and the elevation value of the pixel points contained in the valley line in the high-resolution digital elevation model and the elevation map is retained, the valley line elevation map of the high-resolution digital elevation model is obtained, the elevation value of the pixel points not contained in the valley line identified in the elevation map is assigned as 0, and the elevation value of the pixel points contained in the valley line in the low-resolution digital elevation model and the low-resolution digital elevation model is retained, and the valley line elevation map of the low-resolution digital elevation model is obtained.

[0056] At this point, the gradient map, curvature map and valley line elevation map of the corresponding high-resolution digital elevation model and low-resolution digital elevation model of the same region are obtained.

[0057] All the gradient maps, curvature maps and valley line elevation maps of the corresponding high-resolution digital elevation model and low-resolution digital elevation model of the same region are divided into equal-sized data blocks, and the data set composed of all the data blocks is recorded as the training data set.

[0058] Preferably, in an embodiment of the present application, each data block is a 128*128 data block, and the data block acquisition flow chart is as shown in Figure 2

[0059] At this point, the training data set and the low-resolution digital elevation model of the region to be generated are obtained.

[0060] In step S002, a feature value loss function is established according to the training data set, the feature value loss function is used as a loss function, the convolutional neural network is trained using the training data set, a feature map extraction neural network is obtained, the feature map extraction neural network is used to obtain the gradient map, the curvature map and the feature map of the valley line elevation map of the low-resolution digital elevation model of the region to be generated, and all feature maps of the region to be generated are divided into equal-sized feature data blocks, wherein each feature data block corresponds to a data block.

[0061] The feature value loss function is established according to the training data set, specifically, the feature value loss function L is:

[0062]

[0063] In the formula, L represents the feature value loss function; n represents the number of pixel points contained in the gradient map of the high-resolution digital elevation model; G i represents the gradient value of the i-th pixel point in the gradient map of the high-resolution digital elevation model; represents the gradient value of the corresponding pixel point in the corresponding feature map of the i-th pixel point in the gradient map of the low-resolution digital elevation model; G max represents the maximum value of the gradient values of all pixel points in the gradient map of the high-resolution digital elevation model; C i represents the curvature value of the i-th pixel point in the curvature map of the high-resolution digital elevation model; represents the gradient value of the corresponding pixel point in the corresponding feature map of the i-th pixel point in the curvature map of the low-resolution digital elevation model; C max represents the maximum value of the curvature values of all pixel points in the curvature map of the high-resolution digital elevation model; T i represents the elevation value of the i-th pixel point in the valley line elevation map of the high-resolution digital elevation model; represents the gradient value of the corresponding pixel point in the corresponding feature map of the i-th pixel point in the valley line elevation map of the low-resolution digital elevation model; T max represents the maximum value of the elevation values of all pixel points in the valley line elevation map of the high-resolution digital elevation model.

[0064] In the formula, the predicted value of the pixel gradient value, the predicted value of the curvature value and the predicted value of the elevation value are respectively the values of the corresponding pixel points in the corresponding feature maps of the gradient map, the curvature map and the valley line elevation map of the low-resolution digital elevation model.​

[0065] It can be understood that the number of pixel points contained in the gradient map, the curvature map and the valley line elevation map of the high-resolution digital elevation model is the same; the eigenvalue loss function is composed of gradient loss, curvature loss and valley line elevation loss, specifically, corresponding to the gradient loss, corresponding to the curvature loss, corresponding to the valley line elevation loss.

[0066] The eigenvalue loss function is used as the loss function, and the convolutional neural network is trained using the training data set to obtain the feature map extraction neural network. Specifically, the training data set is divided into a training set, a validation set and a test set according to a ratio of 7:2:1, and the convolutional neural network is trained; all data blocks in the gradient map, the curvature map and the valley line elevation map corresponding to each low-resolution digital elevation model in the training data set are respectively input into the convolutional neural network, and the convolutional neural network outputs the feature map corresponding to the gradient map, the curvature map and the valley line elevation map of the low-resolution digital elevation model; the convolutional neural network is connected through the excitation layer between adjacent convolutional layers, the excitation layer adopts the ReLU function, the initial learning rate is set to 0.001, and the learning rate decay strategy is adopted, and the learning rate is reduced to one fifth of the original every 1000 training rounds.

[0067] It is known that the process of training the convolutional neural network is not described again.

[0068] It can be understood that inputting all data blocks corresponding to the gradient map of a low-resolution digital elevation model into the feature map extraction neural network can obtain the feature map corresponding to the gradient map of the low-resolution digital elevation model; inputting all data blocks corresponding to the curvature map of a low-resolution digital elevation model into the feature map extraction neural network can obtain the feature map corresponding to the curvature map of the low-resolution digital elevation model; inputting all data blocks corresponding to the valley line elevation map of a low-resolution digital elevation model into the feature map extraction neural network can obtain the feature map corresponding to the valley line elevation map of the low-resolution digital elevation model. The pixel value of each pixel point in the feature map corresponding to the gradient map, the curvature map and the valley line elevation map of the low-resolution digital elevation model is respectively used to measure the change characteristics of the gradient change, the curvature change and the elevation value change of the pixel point.

[0069] According to the low-resolution digital elevation model of the to-be-generated region, the gradient map, the curvature map and the valley line elevation map of the low-resolution digital elevation model of the to-be-generated region are obtained, and the gradient map, the curvature map and the valley line elevation map of the to-be-generated region are input into the feature map extraction neural network to obtain the feature map of the gradient map, the curvature map and the valley line elevation map of the to-be-generated region, respectively.

[0070] The feature maps of the gradient map, the curvature map and the valley line elevation map of the region to be generated are divided into equal-sized feature data blocks, each of which corresponds to a data block of the low-resolution digital elevation model of the region to be generated.

[0071] Thus, the feature maps of the gradient map, the curvature map and the valley line elevation map of the region to be generated are obtained, as well as the feature data blocks of all the feature maps.

[0072] In step S003, the type of the feature data block is determined according to the feature data block contained in the feature map and the elevation values of the pixel points in the data block corresponding to the feature data block, and the gradient weight, the curvature weight and the elevation value weight of the feature data block are respectively assigned values. The feature difference function is determined according to the difference between the low-resolution digital elevation model of the region to be generated and the high-resolution digital elevation model generated according to the low-resolution digital elevation model, and the gradient weight, the curvature weight and the elevation value weight of the feature data block. The feature difference function is used as the objective function of the least square method, and the least square method is used to obtain the reconstructed high-resolution digital elevation data of the region to be generated. The digital elevation model is generated according to the reconstructed high-resolution digital elevation data, and the reconstructed digital elevation model of the region to be generated is obtained.

[0073] The gradient variation coefficient, the curvature skewness and the valley line density of the feature data block are respectively determined according to the feature data block contained in the feature map and the elevation values of the pixel points in the data block corresponding to the feature data block.

[0074] The gradients of all the pixel points in the feature data block of the feature map are calculated, and the ratio of the average value to the standard deviation of the gradients of all the pixel points in the feature data block of the feature map is denoted as the gradient variation coefficient of the feature data block. The curvatures of all the pixel points in the feature data block of the feature map are calculated, and the skewness of the curvatures of all the pixel points in the feature data block of the feature map is denoted as the curvature skewness of the feature data block. Since each feature data block corresponds to a data block of the low-resolution digital elevation model of the region to be generated, the data block corresponding to the feature data block is determined. The number of pixel points with non-zero elevation values contained in the data block corresponding to the feature data block is denoted as the non-zero elevation number of the feature data block. The number of all the pixel points contained in the data block corresponding to the feature data block is denoted as the total elevation number of the feature data block. The ratio of the non-zero elevation number to the total elevation number of the feature data block is denoted as the valley line density of the feature data block.

[0075] The type of the feature data block is determined according to the gradient variation coefficient, the curvature skewness and the valley line density of the feature data block.

[0076] The characteristic data blocks with a gradient variation coefficient greater than 0.7 are marked as steep slopes; the characteristic data blocks with a gradient variation coefficient less than 0.7, a curvature skewness less than -0.4, and a valley line density greater than 0.35 are marked as river confluences; the characteristic data blocks with a gradient variation coefficient less than 0.4, a curvature skewness less than -0.3, and a valley line density greater than 0.1 and less than 0.35 are marked as gentle valley bottoms; the characteristic data blocks with a gradient variation coefficient greater than 0.4 and less than 0.6, a curvature skewness greater than 0.5, and a valley line density less than 0.05 are marked as ridges; all other characteristic data blocks are marked as ordinary areas.

[0077] According to the type of feature data block, as well as the gradient variation coefficient, curvature skewness and valley line density of the feature data block, the gradient weight, curvature weight and elevation value weight of the feature data block are assigned respectively.

[0078] When the type of the characteristic data block is marked as a steep slope, the gradient weight, curvature weight and elevation weight corresponding to the gradient variation coefficient, curvature skewness and valley line density of the characteristic data block are assigned to 0.7, 0.2 and 0.1 respectively; when the type of the characteristic data block is a river confluence area, the gradient weight, curvature weight and elevation weight corresponding to the gradient variation coefficient, curvature skewness and valley line density of the characteristic data block are assigned to 0.2, 0.2 and 0.6 respectively; when the type of the characteristic data block is marked as a gentle valley bottom, the gradient weight, curvature skewness and valley line density of the characteristic data block are assigned to The corresponding gradient weight, curvature weight and elevation value weight are assigned to 0.1, 0.45 and 0.45 respectively; when the type of the feature data block is marked as a ridge, the gradient weight, curvature weight and elevation value weight corresponding to the gradient variation coefficient, curvature skewness and valley line density of the feature data block are assigned to 0.45, 0.45 and 0.1 respectively; when the type of the feature data block is marked as an ordinary area, the gradient weight, curvature weight and elevation value weight corresponding to the gradient variation coefficient, curvature skewness and valley line density of the feature data block are assigned to 0.33, 0.33 and 0.34 respectively.

[0079] According to the difference between the low-resolution digital elevation model of the area to be generated and the high-resolution digital elevation model generated based on the low-resolution digital elevation model, as well as the gradient weight, curvature weight and elevation value weight of the feature data block, a feature difference function is determined. Specifically, the feature difference function is:

[0080]

[0081] Where, represents the characteristic difference function; argmin{} represents the minimum function argmin function; denotes L2 norm; X denotes the low resolution digital elevation model of the region to be generated; Y denotes the reconstructed digital elevation model of the region to be generated; Y↓ denotes the down-sampled reconstructed digital elevation model of the region to be generated, which has the same resolution as the low resolution digital elevation model of the region to be generated; ω u,1 denotes the gradient weight of the u-th feature data block; ω u,2 denotes the curvature weight of the u-th feature data block; ω u,3 denotes the elevation value weight of the u-th feature data block; N denotes the number of data blocks divided by the gradient map of the low resolution digital elevation model of the region to be generated; G u,1 denotes the u-th feature data block in the gradient map of the low resolution digital elevation model of the region to be generated; G u,2 denotes the u-th feature data block in the curvature map of the low resolution digital elevation model of the region to be generated; G u,3 denotes the u-th feature data block in the valley line elevation map of the low resolution digital elevation model of the region to be generated; denotes the u-th feature data block in the gradient map of the reconstructed digital elevation model of the region to be generated; denotes the u-th feature data block in the curvature map of the reconstructed digital elevation model of the region to be generated; denotes the u-th feature data block in the valley line elevation map of the reconstructed digital elevation model of the region to be generated.

[0082] The feature difference function is taken as the objective function of the least square method, and the least square method is used to seek the optimal solution, and the feature map of the region to be generated is optimized by inversion, so that the difference between the down-sampled result of the reconstructed digital elevation model of the region to be generated and the low resolution digital elevation model of the region to be generated is minimized, and the reconstructed high resolution digital elevation data of the region to be generated is obtained.

[0083] The reconstructed high resolution digital elevation data of the region to be generated is processed by using ArcGIS to obtain the reconstructed digital elevation model of the region to be generated, that is, the high resolution digital elevation model of the region to be generated is generated according to the low resolution digital elevation model of the region to be generated.

[0084] Wherein, ArcGIS is a geographic information system software platform, which can be used to generate high resolution digital elevation model according to digital elevation data, and using ArcGIS to generate high resolution digital elevation model and using least square method to seek optimal solution are all known technologies, which will not be described in detail.

[0085] Thus, the generation of high resolution digital elevation model is realized.

[0086] The above description is only the preferred embodiment of the present application, and is not used to limit the present application, any modification, equivalent replacement, improvement, etc. made within the principle of the present application should be included in the protection scope of the present application.

Claims

1. A high-resolution digital elevation model generation processing method characterized by, The method comprises the following steps: The method comprises the following steps: According to the gradient weight, the curvature weight and the height value weight of the feature data block, and the difference between the low-resolution digital elevation model of the to-be-generated region and the high-resolution digital elevation model generated according to the low-resolution digital elevation model, the feature difference function is determined, the feature difference function is taken as the objective function of the least square method, and the least square method is used to obtain the reconstructed high-resolution digital elevation data of the to-be-generated region, the digital elevation model is generated according to the reconstructed high-resolution digital elevation data, and the reconstructed digital elevation model of the to-be-generated region is obtained. The method for obtaining the gradient graph, the curvature graph and the valley line height graph comprises the following steps:

2. The method of claim 1, wherein The gradient graph, the curvature graph and the height graph of each high-resolution digital elevation model and low-resolution digital elevation model corresponding to the same region are obtained respectively; The valley lines in each high-resolution digital elevation model and low-resolution digital elevation model corresponding to the same region are obtained respectively; The pixel points not contained in the high-resolution digital elevation model and the pixel points contained in the high-resolution digital elevation model are identified in the height graph, the height value of the pixel points not contained in the height graph is assigned as 0, the height value of the pixel points contained in the high-resolution digital elevation model and the height graph is retained, and the valley line height graph of the high-resolution digital elevation model is obtained; the height value of the pixel points not contained in the height graph is assigned as 0, the height value of the pixel points contained in the low-resolution digital elevation model and the height graph is retained, and the valley line height graph of the low-resolution digital elevation model is obtained. The method for obtaining the data block and the training data set comprises the following steps:

3. The method of claim 2, wherein The gradient graph, the curvature graph and the valley line height graph of all high-resolution digital elevation models and low-resolution digital elevation models corresponding to the same region are divided into data blocks with a preset size; The data set composed of all data blocks is recorded as the training data set. The feature value loss function is specifically 4. The method of claim 1, wherein, ​ In the formula, L represents an eigenvalue loss function; n represents the number of pixel points contained in a gradient graph of a high-resolution digital elevation model; G i represents a gradient value of an i-th pixel point in the gradient graph of the high-resolution digital elevation model; represents a gradient value of a corresponding pixel point in a corresponding feature graph of an i-th pixel point in a gradient graph of a low-resolution digital elevation model; G max represents a maximum value of gradient values of all pixel points in the gradient graph of the high-resolution digital elevation model; C i represents a curvature value of an i-th pixel point in a curvature graph of a high-resolution digital elevation model; represents a gradient value of a corresponding pixel point in a corresponding feature graph of an i-th pixel point in a curvature graph of a low-resolution digital elevation model; C max represents a maximum value of curvature values of all pixel points in the curvature graph of the high-resolution digital elevation model; T i represents an elevation value of an i-th pixel point in a valley line elevation graph of a high-resolution digital elevation model; represents a gradient value of a corresponding pixel point in a corresponding feature graph of an i-th pixel point in a valley line elevation graph of a low-resolution digital elevation model; T max represents a maximum value of elevation values of all pixel points in the valley line elevation graph of the high-resolution digital elevation model.

5. The method of claim 1, wherein, The method comprises the following steps of: According to the proportion of 7:2:1, the training data set is divided into a training set, a verification set and a test set, the convolutional neural network is trained, and a feature map extraction neural network is obtained; wherein all data blocks in the gradient map, the curvature map and the valley line elevation map corresponding to each low-resolution digital elevation model in the training data set are respectively taken as the input of the feature map extraction neural network, and the feature map extraction neural network outputs the feature map corresponding to the gradient map, the curvature map and the valley line elevation map of the low-resolution digital elevation model.

6. The method of claim 1, wherein, The method comprises the following steps of: The gradient of all pixel points in the feature data block of the feature map is calculated, and the ratio of the average value of the gradient of all pixel points in the feature data block of the feature map to the standard deviation is taken as the gradient variation coefficient of the feature data block; The curvature of all pixel points in the feature data block of the feature map is calculated, and the skewness of the curvature of all pixel points in the feature data block of the feature map is taken as the curvature skewness of the feature data block; The number of pixel points with a height value of 0 in the data block corresponding to the feature data block is taken as the non-zero height number of the feature data block, the total number of all pixel points in the data block corresponding to the feature data block is taken as the total height number of the feature data block, and the ratio of the non-zero height number to the total height number is taken as the valley line density of the feature data block; The type of the feature data block is determined according to the gradient variation coefficient, the curvature skewness and the valley line density of the feature data block.

7. The method of claim 6, wherein The method comprises the following steps of: The type of the feature data block is marked as steep slope when the gradient variation coefficient is greater than 0.7; The type of the feature data block is marked as river junction area when the gradient variation coefficient is less than 0.7, the curvature skewness is less than-0.4, and the valley line density is greater than 0.35; The type of the feature data block is marked as gentle valley bottom when the gradient variation coefficient is less than 0.4, the curvature skewness is less than-0.3, and the valley line density is greater than 0.1 and less than 0.35; The type of the feature data block is marked as ridge when the gradient variation coefficient is greater than 0.4 and less than 0.6, the curvature skewness is greater than 0.5, and the valley line density is less than 0.05; The types of all the remaining feature data blocks are marked as ordinary areas.

8. The method of claim 7, wherein, The method comprises the following steps of: When the type of the feature data block is marked as steep slope, the gradient weight, the curvature weight and the height value weight corresponding to the gradient variation coefficient, the curvature skewness and the valley line density of the feature data block are respectively assigned as 0.7, 0.2 and 0.1; When the kind of the feature data block is a river confluence area, the gradient weight, the curvature weight and the elevation value weight corresponding to the gradient variation coefficient, the curvature skewness and the valley line density of the feature data block are respectively assigned as 0.2, 0.2 and 0.6; When the kind of the feature data block is a gentle valley bottom, the gradient weight, the curvature weight and the elevation value weight corresponding to the gradient variation coefficient, the curvature skewness and the valley line density of the feature data block are respectively assigned as 0.1, 0.45 and 0.45; When the kind of the feature data block is a ridge, the gradient weight, the curvature weight and the elevation value weight corresponding to the gradient variation coefficient, the curvature skewness and the valley line density of the feature data block are respectively assigned as 0.45, 0.45 and 0.1; When the kind of the feature data block is a common area, the gradient weight, the curvature weight and the elevation value weight corresponding to the gradient variation coefficient, the curvature skewness and the valley line density of the feature data block are respectively assigned as 0.33, 0.33 and 0.

34.

9. The method of claim 8, wherein, The feature difference function is specifically: In the formula, denotes the characteristic difference function; argmin{} represents the minimum value function argmin function; represents the L2 norm; X represents the low-resolution digital elevation model of the region to be generated; Y represents the reconstructed digital elevation model of the region to be generated; Y↓ represents the down-sampled reconstructed digital elevation model of the region to be generated; ω u,1 represents the gradient weight of the u-th feature data block; ω u,2 represents the curvature weight of the u-th feature data block; ω u,3 represents the elevation value weight of the u-th feature data block; N represents the number of data blocks divided by the gradient map of the low-resolution digital elevation model of the region to be generated; G u,1 represents the u-th feature data block in the gradient map of the low-resolution digital elevation model of the region to be generated; G u,2 u-th feature data block in a curvature map representing a low resolution digital elevation model of the region to be generated; G u,3 u-th feature data block in a valley line elevation map representing a low resolution digital elevation model of the region to be generated; u-th feature data block in a gradient map representing a reconstructed digital elevation model of the region to be generated; u-th feature data block in a curvature map representing a reconstructed digital elevation model of the region to be generated; u-th feature data block in a valley line elevation map representing a reconstructed digital elevation model of the region to be generated.

10. The method of claim 9, wherein, The feature difference function is used as an objective function of the least square method, and the specific method for obtaining the reconstructed high-resolution digital elevation data of the to-be-generated area by using the least square method comprises: The feature difference function is used as an objective function of the least square method, and the specific method for obtaining the reconstructed high-resolution digital elevation data of the to-be-generated area by using the least square method comprises: The feature difference function is used as an objective function of the least square method, and the specific method for obtaining the reconstructed high-resolution digital elevation data of the to-be-generated area by using the least square method comprises:

Images