A Hyperspectral Cadastral Surveying and Mapping Method, System, Device and Storage Medium

Abstract

The present application provides a hyperspectral cadastral mapping method, system, device and storage medium, which can obtain hyperspectral images of a specified area; determine the vector proximity of spectral vectors between every two pixel points, and then segment to obtain multiple image segmentation domains; for each image segmentation domain, obtain the reference spectral values of all pixel points, so as to obtain the spectral similarity between each pixel point and the landform, determine the structural feature vectors of each pixel point in the image segmentation domain through all the spectral similarities, and obtain all the structural feature vectors; obtain all sets of neighborhood pixel points, and determine the spectral difference entropy between each pixel point and the corresponding set of neighborhood pixel points; determine the feature separation degree of each pixel point through the structural feature vector and spectral difference entropy of each pixel point, and perform segmentation based on all the feature separation degrees, so as to achieve cadastral annotation. By adopting the solution of the present application, spectral redundancy in hyperspectral images can be avoided, so as to improve the ability to distinguish surface objects.

Description

Technical Field

[0001] This application relates to the technical field of cadastral surveying and mapping. More specifically, this application relates to a hyperspectral cadastral surveying and mapping method, system, device, and storage medium. Background Art

[0002] Cadastral surveying and mapping involves the measurement and recording of land boundaries, areas, shapes, and positions, etc. Its main purpose is to ensure the clarity of land ownership, the legality of land use, and the rational allocation of land resources. Cadastral surveying and mapping provides basic data support for land management departments, helps decision-makers make scientific land planning and management decisions, and is of great significance for the effective management and sustainable utilization of land resources. It is a key tool for achieving scientific management and decision-making.

[0003] Hyperspectral cadastral surveying and mapping combines hyperspectral remote sensing technology and advanced cadastral surveying and mapping methods, aiming to provide more detailed and accurate land information. This surveying and mapping technology uses hyperspectral sensors to obtain spectral data of the earth's surface, thereby obtaining rich information about ground object types, states, and characteristics, etc. It mainly uses hyperspectral sensors to conduct remote sensing measurements on the earth's surface to obtain spectral information of each pixel. Among them, hyperspectral sensors can capture a large amount of spectral data in different spectral bands (usually from the visible light to the near-infrared region). These data can reflect the spectral characteristics of surface substances. In the existing hyperspectral cadastral surveying and mapping process, due to the strong correlation between the bands of hyperspectral images, the inter-spectral correlation coefficient of hyperspectral images is often too large, resulting in spectral redundancy, reducing the ability of classification algorithms to distinguish surface objects, and thus reducing the accuracy of cadastral surveying and mapping. Therefore, how to avoid spectral redundancy in hyperspectral images to improve the ability to distinguish surface objects has become a difficult problem faced by the industry. Summary of the Invention

[0004] This application provides a hyperspectral cadastral surveying and mapping method, system, device, and storage medium, which can avoid spectral redundancy in hyperspectral images to improve the ability to distinguish surface objects.

[0005] In a first aspect, this application provides a hyperspectral cadastral surveying and mapping method, including the following steps:

[0006] Start hyperspectral cadastral surveying and mapping to obtain a hyperspectral image of a specified area;

[0007] Determine the vector approximation degree between the spectral vectors of every two pixel points in the hyperspectral image, and approximately segment the hyperspectral image according to all vector approximation degrees to obtain a plurality of image segmentation domains;

[0008] For each image segmentation domain, obtain the reference spectral values of the image segmentation domain based on the geomorphic information of the specified area, and then determine the spectral similarity between each pixel point in the image segmentation domain and the geomorphology of the specified area according to the reference spectral values. Construct the features of all pixel points in the image segmentation domain through all the spectral similarities, obtain the structural feature vectors of each pixel point in the image segmentation domain, and then obtain the structural feature vectors of each pixel point in each image segmentation domain;

[0009] Obtain the neighborhood pixel point sets of each pixel point in the hyperspectral image, and determine the spectral difference entropy between each pixel point and the corresponding neighborhood pixel point set according to the spectral distribution characteristics of the hyperspectral image;

[0010] Determine the feature separation degree of each pixel point in the hyperspectral image through the structural feature vector corresponding to each pixel point and the spectral difference entropy corresponding to each pixel point. Perform secondary segmentation on the hyperspectral image based on the feature separation degrees of all pixel points, and perform cadastral annotation on the specified area according to the segmentation results.

[0011] In some embodiments, determining the vector approach degree of the spectral vectors between every two pixel points in the hyperspectral image specifically includes:

[0012] Select a pixel point in the hyperspectral image as the selected pixel point, and use the remaining pixel points in the hyperspectral image after selecting the selected pixel point as the pixel point transition set;

[0013] Determine the vector approach degree of the spectral vectors between the selected pixel point and each pixel point in the pixel point transition set;

[0014] Select a pixel point from all the pixel points in the pixel point transition set as the new selected pixel point, and use the remaining pixel points in the pixel point transition set after selecting the new selected pixel point as the new pixel point transition set;

[0015] Determine the vector approach degree of the spectral vectors between the new selected pixel point and each pixel point in the new pixel point transition set;

[0016] And so on until all the pixel points in the hyperspectral image are selected, and then obtain the vector approach degree of the spectral vectors between every two pixel points in the hyperspectral image.

[0017] In some embodiments, approximately segmenting the hyperspectral image according to all the vector approach degrees to obtain multiple image segmentation domains specifically includes:

[0018] Determine the approach degree matrix of the hyperspectral image according to all the vector approach degrees;

[0019] Segment the hyperspectral image according to the similarity matrix to obtain multiple image segmentation domains.

[0020] In some embodiments, obtaining the reference spectral value of the image segmentation domain based on the geomorphic information of a specified area specifically includes:

[0021] Determine the significant bands corresponding to the geomorphic information of the specified area;

[0022] Obtain the significant spectral values of each pixel point in the image segmentation domain under the significant bands;

[0023] Determine the reference spectral value of the image segmentation domain through all the significant spectral values.

[0024] In some embodiments, determining the spectral similarity between each pixel point in the image segmentation domain and the geomorphology of the specified area according to the reference spectral value specifically includes:

[0025] Determine the characteristic spectral value of each pixel point in the image segmentation domain;

[0026] Determine the standard spectral value of each pixel point in the image segmentation domain through the characteristic spectral value of each pixel point and the reference spectral value;

[0027] Determine the spectral similarity between each pixel point in the image segmentation domain and the geomorphology of the specified area according to all the standard spectral values.

[0028] In some embodiments, performing feature construction on all pixel points in the image segmentation domain through all the spectral similarities to obtain the structural feature vector of each pixel point in the image segmentation domain specifically includes:

[0029] Extract feature pixel points from all pixel points in the image segmentation domain through all the spectral similarities;

[0030] Determine the spectral change rate between adjacent bands in the spectral vector of the feature pixel points;

[0031] Extract multiple feature bands from all bands according to each spectral change rate;

[0032] Select a pixel point in the image segmentation domain as the selected pixel point;

[0033] Construct the spectral values of the selected pixel point under each feature band into the structural feature vector of the selected pixel point;

[0034] Continue to determine the structural feature vectors of the remaining pixel points.

[0035] In some embodiments, determining the feature separation degree of each pixel point in the hyperspectral image through the structural feature vector corresponding to each pixel point and the spectral difference entropy corresponding to each pixel point specifically includes:

[0036] Select a pixel point as the selected pixel point;

[0037] Determine the feature variability according to the structural feature vector of the selected pixel points;

[0038] Determine the feature separation degree of the selected pixel points in the hyperspectral image according to the spectral difference entropy of the selected pixel points and the feature variability;

[0039] Continue to determine the feature separation degree of the remaining pixel points in the hyperspectral image, so as to obtain the feature separation degree of each pixel point in the hyperspectral image.

[0040] In a second aspect, the present application provides a hyperspectral cadastral surveying and mapping system, including:

[0041] An acquisition module, configured to acquire a hyperspectral image of a specified area after starting hyperspectral cadastral surveying and mapping;

[0042] A processing module, configured to determine the vector proximity degree of the spectral vectors between every two pixel points in the hyperspectral image, and approximately segment the hyperspectral image according to all the vector proximity degrees to obtain a plurality of image segmentation domains;

[0043] The processing module is further configured to, for each image segmentation domain, obtain a reference spectral value of the image segmentation domain based on the landform information of the specified area, and then determine the spectral similarity between each pixel point in the image segmentation domain and the landform of the specified area according to the reference spectral value, and perform feature construction on all pixel points in the image segmentation domain through all the spectral similarities to obtain the structural feature vectors of each pixel point in the image segmentation domain, and then obtain the structural feature vectors of each pixel point in each image segmentation domain;

[0044] The processing module is further configured to obtain the neighborhood pixel point set of each pixel point in the hyperspectral image, and determine the spectral difference entropy between each pixel point and the corresponding neighborhood pixel point set according to the spectral distribution characteristics of the hyperspectral image;

[0045] An execution module, configured to determine the feature separation degree of each pixel point in the hyperspectral image through the structural feature vector corresponding to each pixel point and the spectral difference entropy corresponding to each pixel point, perform secondary segmentation on the hyperspectral image based on the feature separation degree of each pixel point, and perform cadastral annotation on the specified area according to the segmentation result.

[0046] In a third aspect, the present application provides a computer device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, the steps of the above-mentioned hyperspectral cadastral surveying and mapping method are implemented.

[0047] In a fourth aspect, the present application provides a computer-readable storage medium storing a computer program, which when executed by a processor implements the steps of the above-described hyperspectral cadastral mapping method.

[0048] The technical solutions provided by the disclosed embodiments of the present application have the following beneficial effects:

[0049] In the hyperspectral cadastral mapping method, system, device, and storage medium provided by the present application, after starting hyperspectral cadastral mapping, a hyperspectral image of a specified area is obtained; the vector approximation degree between the spectral vectors of every two pixel points in the hyperspectral image is determined, and the hyperspectral image is approximately segmented according to all the vector approximation degrees to obtain a plurality of image segmentation domains; for each image segmentation domain, a reference spectral value of the image segmentation domain is obtained based on the landform information of the specified area, and then the spectral similarity between each pixel point in the image segmentation domain and the landform of the specified area is determined according to the reference spectral value. The structural feature vectors of each pixel point in the image segmentation domain are constructed through all the spectral similarities, and then the structural feature vectors of each pixel point in each image segmentation domain are obtained; the neighborhood pixel point set of each pixel point in the hyperspectral image is obtained, and the spectral difference entropy between each pixel point and the corresponding neighborhood pixel point set is determined according to the spectral distribution characteristics of the hyperspectral image; the feature separation degree of each pixel point in the hyperspectral image is determined through the structural feature vector corresponding to each pixel point and the spectral difference entropy corresponding to each pixel point. The hyperspectral image is secondarily segmented based on the feature separation degrees of the pixel points, and cadastral annotation of the specified area is performed according to the segmentation result.

[0050] It can be seen that in this application, first, hyperspectral cadastral surveying and mapping is initiated. After obtaining the hyperspectral image of a specified area, the hyperspectral image is segmented into multiple image segmentation domains by the vector proximity of the spectral vectors between every two pixel points in the hyperspectral image. Then, for each image segmentation domain, a representative spectral value (reference spectral value) for classification of the image segmentation domain is obtained based on the geomorphic information of the specified area. Furthermore, the matching degree (spectral similarity) between each pixel point in the image segmentation domain and the geomorphology of the specified area is determined according to the reference spectral value. The most representative pixel points in the hyperspectral image are extracted through all the spectral similarities, and key characteristic bands are selected based on the spectral change rate. Then, the spectral values under these characteristic bands are converted into the structural feature vectors of each pixel point, thereby avoiding considering the spectral values on all bands and avoiding spectral redundancy caused by the strong correlation between different bands. After obtaining the set of neighborhood pixel points of each pixel point in the hyperspectral image, the degree of spectral characteristic difference between each pixel point and the corresponding set of neighborhood pixel points is determined. Furthermore, the discrimination ability (feature separation degree) of each pixel point in the hyperspectral image relative to its neighborhood is determined according to the spectral values (structural feature vectors) of each pixel point on all characteristic bands and the degree of spectral characteristic difference between each pixel point and the corresponding set of neighborhood pixel points. Thus, the hyperspectral image is secondarily segmented according to each feature separation degree, and cadastral annotation is performed on the specified area according to the segmentation result. In summary, the solution of this application can avoid spectral redundancy in the hyperspectral image and improve the discrimination ability for surface objects. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 is a schematic flowchart of a hyperspectral cadastral surveying and mapping method according to some embodiments of the present application;

[0052] Figure 2 is a schematic flowchart of determining a reference spectral value according to some embodiments of the present application;

[0053] Figure 3 is a schematic flowchart of determining a feature separation degree according to some embodiments of the present application;

[0054] Figure 4 is a structural block diagram of a hyperspectral cadastral surveying and mapping system according to some embodiments of the present application;

[0055] Figure 5 is an internal structural diagram of a computer device for implementing the hyperspectral cadastral surveying and mapping method according to some embodiments of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0056] The core of this application is to obtain the hyperspectral image of a specified area after starting hyperspectral cadastral surveying; determine the vector approximation degree of the spectral vectors between every two pixel points in the hyperspectral image, and approximately segment the hyperspectral image according to all the vector approximation degrees to obtain multiple image segmentation domains; for each image segmentation domain, obtain the reference spectral value of the image segmentation domain based on the geomorphic information of the specified area, and then determine the spectral similarity between each pixel point in the image segmentation domain and the geomorphology of the specified area according to the reference spectral value, and construct the feature of all pixel points in the image segmentation domain through all the spectral similarities to obtain the structural feature vector of each pixel point in the image segmentation domain, and then obtain the structural feature vector of each pixel point in each image segmentation domain; obtain the set of neighborhood pixel points of each pixel point in the hyperspectral image, and determine the spectral difference entropy between each pixel point and the corresponding set of neighborhood pixel points according to the spectral distribution characteristics of the hyperspectral image; determine the feature separation degree of each pixel point in the hyperspectral image through the structural feature vector corresponding to each pixel point and the spectral difference entropy corresponding to each pixel point, perform secondary segmentation on the hyperspectral image based on the feature separation degree of each pixel point, and perform cadastral annotation on the specified area according to the segmentation result; it can avoid spectral redundancy in the hyperspectral image to improve the ability to distinguish surface objects.

[0057] To better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings of the specification and specific implementation manners. Refer to Figure 1 , this figure is a schematic flowchart of a hyperspectral cadastral surveying method shown in some embodiments of the present application. The hyperspectral cadastral surveying method 100 mainly includes the following steps:

[0058] In step 101, start hyperspectral cadastral surveying and obtain the hyperspectral image of a specified area.

[0059] Specifically, after starting hyperspectral cadastral surveying, a hyperspectral sensor can be installed on a drone device, and the drone can fly over the specified area to collect hyperspectral information, and a hyperspectral image of the specified area can be constructed according to the collected information.

[0060] It should be noted that the hyperspectral sensor described in this application is of the Specim AisaFENIX model, which provides a band range of 400 - 2500nm and is suitable for imaging from visible light to short-wave infrared. In addition, the hyperspectral image is composed of multiple pixel points, where each pixel point corresponds to a spectral vector, and the spectral vector contains the spectral values of the corresponding pixel point in all bands of the hyperspectral sensor.

[0061] In step 102, the vector proximity of the spectral vectors between every two pixel points in the hyperspectral image is determined, and the hyperspectral image is approximately segmented according to all the vector proximities to obtain a plurality of image segmentation domains.

[0062] In some embodiments, the determination of the vector proximity of the spectral vectors between every two pixel points in the hyperspectral image can be implemented by the following steps:

[0063] Select a pixel point in the hyperspectral image as the selected pixel point, and take the remaining pixel points in the hyperspectral image after selecting the selected pixel point as the pixel point transition set;

[0064] Determine the vector proximity of the spectral vectors of the selected pixel point and each pixel point in the pixel point transition set;

[0065] Select a pixel point from all the pixel points in the pixel point transition set as the new selected pixel point, and take the remaining pixel points in the pixel point transition set after selecting the new selected pixel point as the new pixel point transition set;

[0066] Determine the vector proximity of the spectral vectors of the new selected pixel point and each pixel point in the new pixel point transition set;

[0067] And so on until all the pixel points in the hyperspectral image are selected, thereby obtaining the vector proximity of the spectral vectors between every two pixel points in the hyperspectral image.

[0068] In specific implementation, the determination of the vector proximity of the spectral vectors of the selected pixel point and each pixel point in the pixel point transition set can be implemented by the following method, that is: First, determine the magnitudes of the spectral vectors of the selected pixel point and all the pixel points in the pixel point transition set. Then, select a pixel point in the pixel point transition set as the selected transition pixel point. Secondly, take the result of dividing the dot product of the spectral vectors of the selected pixel point and the selected transition pixel point by the magnitudes of the spectral vectors of the selected pixel point and the selected transition pixel point as the vector proximity between the selected pixel point and the selected transition pixel point, and continue to determine the vector proximity between the selected pixel point and the remaining selected transition pixel points, so as to obtain the vector proximity of the spectral vectors of the selected pixel point and all the pixel points in the pixel point transition set.

[0069] It should be noted that in this application, the vector proximity characterizes the similarity degree of the spectral characteristics between the corresponding two pixel points. The larger the vector proximity, the higher the similarity degree of the spectral characteristics between the corresponding two pixel points, and the smaller the vector proximity, the lower the similarity degree of the spectral characteristics between the corresponding two pixel points.

[0070] In some embodiments, the approximation segmentation of the hyperspectral image according to all the vector proximities to obtain a plurality of image segmentation domains can be implemented by the following steps:

[0071] Determine the proximity matrix of the hyperspectral image according to the proximities of all vectors;

[0072] Segment the hyperspectral image according to the similarity matrix to obtain multiple image segmentation regions.

[0073] It should be noted that in this application, after setting a unit matrix with elements on the diagonal of size K*K being 1, all vector proximities are filled into it, and the filled matrix is used as the proximity matrix in this application, where K is the total number of pixel points in the hyperspectral image. In some embodiments, the proximity matrix is expressed as:

[0074]

[0075] where V is the proximity matrix, C(1, 2) represents the vector proximity between the first pixel point and the second pixel point in the hyperspectral image, C(1, K - 1) represents the vector proximity between the first pixel point and the (K - 1)-th pixel point in the hyperspectral image, and so on. The vector proximity in the first row and the K-th column of the proximity matrix is expressed as C(1, K), C(2, 1) represents the vector proximity in the second row and the first column of the proximity matrix, and so on. The vector proximity in the (K - 1)-th row and the first column of the proximity matrix is expressed as C(K - 1, 1), and the vector proximity in the K-th row and the (K - 1)-th column is expressed as C(K, K - 1).

[0076] It should be noted that the approximate segmentation in this application refers to different image regions (i.e., image segmentation regions) divided in the hyperspectral image according to the spectral characteristics of different pixel points. The approximate segmentation can achieve a rough classification of the hyperspectral image. In addition, the image segmentation regions characterize different ground object types in the hyperspectral image, such as vegetation, water bodies, bare soil, and urban buildings, etc. Specifically, when implementing, segmenting the hyperspectral image according to the similarity matrix to obtain multiple image segmentation regions can be achieved by the following method: First, determine the Laplacian matrix of the proximity matrix, then, use the Laplacian matrix as the similarity matrix of spectral clustering in the prior art, perform eigen-decomposition on the Laplacian matrix to obtain multiple eigenvectors, then classify all the eigenvectors, and segment each pixel point in the hyperspectral image into multiple categories according to the classification results. Finally, the image composed of pixel points of the corresponding category is used as an image segmentation region, thereby obtaining multiple image segmentation regions.

[0077] In step 103, for each image segmentation domain, based on the geomorphic information of the specified area, the reference spectral value of the image segmentation domain is obtained. Then, according to the reference spectral value, the spectral similarity between each pixel point in the image segmentation domain and the geomorphology of the specified area is determined. Through all the spectral similarities, the feature construction of all pixel points in the image segmentation domain is carried out to obtain the structural feature vector of each pixel point in the image segmentation domain, and then the structural feature vector of each pixel point in each image segmentation domain is obtained.

[0078] In some embodiments, refer to Figure 2 As shown, this figure is a schematic flowchart of determining the reference spectral value according to some embodiments of the present application. Obtaining the reference spectral value of the image segmentation domain based on the geomorphic information of the specified area can be achieved by the following steps:

[0079] First, in 1031, determine the significant bands corresponding to the geomorphic information of the specified area;

[0080] Then, in 1032, obtain the significant spectral values of each pixel point in the image segmentation domain in the significant bands;

[0081] Finally, in 1033, determine the reference spectral value of the image segmentation domain through all the significant spectral values.

[0082] It should be noted that the geomorphic information of the specified area refers to the geomorphic type information of the specified area, such as different types of water bodies, bare soil, and vegetation. Among them, each geomorphic type will exhibit different spectral characteristics in the hyperspectral image. For example, the vegetation geomorphic type has a higher reflectance in the near-infrared band, and the water body geomorphic type has a significant absorption rate in the short-infrared band. In the present application, the bands with significant absorption rate and higher reflectance for the geomorphic type of the specified area are selected as the significant bands corresponding to the geomorphic information of the specified area. In other embodiments, other methods can also be used to determine, which is not limited here.

[0083] It should be noted that the significant spectral value described in this application refers to the spectral value of a pixel at a specific significant band. The significant band refers to a band that has a significant effect on the classification of land cover types in the image segmentation domain. The reference spectral value is a representative spectral value extracted from the spectral data of all pixels in the specified area or the image segmentation domain at the significant band. It is used for subsequent classification, comparison, and analysis to help determine the spectral characteristics of a certain area and serve as a benchmark for comparing the spectral characteristics of this area with other areas. Therefore, in specific implementation, the significant spectral values of all pixels in the image segmentation domain at the significant band can be obtained in the following way: that is, the spectral values of all pixels in the image segmentation domain at the significant band are used as the significant spectral values corresponding to each pixel. In other embodiments, other methods can also be used to determine it, which will not be elaborated here. Additionally, as a preferred embodiment, the reference spectral value of all pixels in the image segmentation domain can be determined through all the significant spectral values in the following way: that is, the standard deviation of the significant spectral values corresponding to all pixels in the image segmentation domain is used as the reference spectral value of the image segmentation domain. In other embodiments, other methods can also be used to determine it, which is not limited here.

[0084] In some embodiments, the spectral similarity between each pixel in the image segmentation domain and the landform of the specified area can be determined according to the reference spectral value by the following steps:

[0085] Determine the characteristic spectral value of each pixel in the image segmentation domain;

[0086] Determine the standard spectral value of each pixel in the image segmentation domain through the characteristic spectral value of each pixel and the reference spectral value;

[0087] Determine the spectral similarity between each pixel in the image segmentation domain and the landform of the specified area according to all the standard spectral values.

[0088] In specific implementation, the characteristic spectral value of each pixel in the image segmentation domain can be determined in the following way: that is, the average value of the spectral values of each pixel in the image segmentation domain at all bands is used as the characteristic spectral value corresponding to the pixel. In other embodiments, other methods can also be used to determine it, which is not limited here.

[0089] In specific implementation, the standard spectral value of each pixel point in the image segmentation domain can be determined by the characteristic spectral value of each pixel point and the reference spectral value in the following manner: that is, the result of taking the square root of the square of the difference between the characteristic spectral value of each pixel point and the reference spectral value is used as the standard spectral value of the corresponding pixel point. In other embodiments, other methods can also be used for determination, which is not limited here; as a preferred embodiment, the spectral similarity between each pixel point in the image segmentation domain and the landform of a specified area can be determined according to all the standard spectral values in the following manner: that is, first, obtain the maximum standard spectral value and the minimum standard spectral value among all the standard spectral values. Secondly, after using the maximum standard spectral value and the minimum standard spectral value as inputs, the results corresponding to the characteristic spectral values of each pixel point are mapped between 0 and 1 by normalization in the prior art, and then the value obtained by subtracting the result of mapping the characteristic spectral value of each pixel point between 0 and 1 from 1 is used to obtain the spectral similarity between each pixel point and the landform of the specified area. In other embodiments, other methods can also be used for determination, which is not limited here.

[0090] It should be noted that the spectral similarity described in this application characterizes the matching degree between the spectral characteristics of the pixel point and the landform of the specified area. The greater the spectral similarity, the higher the matching degree between the spectral characteristics of the pixel point and the landform of the specified area; the smaller the spectral similarity, the lower the matching degree between the spectral characteristics of the pixel point and the landform of the specified area. The value range of the spectral similarity is [0, 1].

[0091] In some embodiments, the structural feature vector of each pixel point in the image segmentation domain can be obtained by constructing features for all pixel points in the image segmentation domain through all the spectral similarities, which can be implemented by the following steps:

[0092] Extract feature pixel points from all pixel points in the image segmentation domain through all the spectral similarities;

[0093] Determine the spectral change rate between adjacent bands in the spectral vector of the feature pixel points;

[0094] Extract multiple feature bands from all bands according to each spectral change rate;

[0095] Select a pixel point in the image segmentation domain as the selected pixel point;

[0096] Construct the spectral values of the selected pixel point in each feature band into the structural feature vector of the selected pixel point;

[0097] Continue to determine the structural feature vectors of the remaining pixel points.

[0098] It should be noted that the characteristic pixel points refer to the pixel points corresponding to the highest spectral similarity. By selecting the pixel points with the largest spectral similarity as the characteristic pixel points in this application, the most representative spectral information in the corresponding image segmentation domain can be captured. Therefore, in specific implementation, the characteristic pixel points can be extracted from all pixel points in the image segmentation domain through all spectral similarities in the following manner, that is: First, obtain the maximum spectral similarity from all spectral similarities. Then, use the pixel points corresponding to the maximum spectral similarity as the characteristic pixel points. In other embodiments, other methods can also be used to determine, which are not limited here.

[0099] It should be noted that the spectral change rate reflects the change speed of the spectral vectors of the characteristic pixel points between different bands, and can be used to identify and distinguish the spectral characteristics of different ground object types. The larger the spectral change rate, the greater the spectral characteristic change in the corresponding band region, and the smaller the spectral change rate, the smaller the spectral characteristic change in the corresponding band region. As a preferred embodiment, the spectral change rate between each adjacent band in the spectral vector of the characteristic pixel points can be determined in the following manner, that is: First, obtain the spectral values of all bands in the spectral vector of the characteristic pixel points. Then, use the result of subtracting the spectral values between all adjacent bands and dividing by the absolute value of the difference between adjacent bands as the spectral change rate between the corresponding adjacent bands. In other embodiments, other methods can also be used to determine, which are not limited here. As a preferred embodiment, multiple characteristic bands can be extracted from all bands according to each spectral change rate in the following manner, that is: First, obtain the maximum spectral change rate from all spectral change rates. Then, obtain the multiple bands corresponding to the maximum spectral change rate. Finally, use the bands corresponding to the maximum spectral change rate as the characteristic bands in this application, so as to obtain multiple characteristic bands. In addition, the characteristic bands refer to the bands corresponding to the largest spectral characteristic change in the spectral vector of the characteristic pixel points, that is: Analyzing the characteristic bands can effectively identify the landform type.

[0100] It should be noted that the feature construction in this application refers to extracting the most representative pixel points in the hyperspectral image through all spectral similarities, selecting key feature bands based on the spectral change rate, and then converting the spectral values under these feature bands into the structural feature vectors of each pixel point. The feature construction can improve the efficiency and accuracy of the model, while reducing spectral redundancy. The structural feature vector refers to a vector constructed by arranging the spectral values of each pixel point under the selected feature bands in the order of the feature bands, which represents the spectral characteristics of each pixel point under the most representative bands (feature bands). In this application, the spectral characteristics of each pixel point under all feature bands are analyzed through the structural feature vector, thus avoiding spectral redundancy caused by strong correlations between different bands. As a preferred embodiment, constructing the spectral values of the selected pixel points under each feature band into the structural feature vector of the selected pixel point can be achieved by the following method, that is: First, obtain the spectral values of the selected pixel point under each feature band. Second, sort all the spectral values according to the magnitudes of the corresponding feature bands and use them as vector elements, thereby obtaining the structural feature vector of the selected pixel point.

[0101] In step 104, obtain the set of neighboring pixel points of each pixel point in the hyperspectral image, and determine the spectral difference entropy between each pixel point and the corresponding set of neighboring pixel points according to the spectral distribution characteristics of the hyperspectral image.

[0102] Specifically, obtaining the set of neighboring pixel points of each pixel point in the hyperspectral image can be achieved by the following method, that is: First, select a pixel point as the selected pixel point. Then, based on the 8-direction search method in the prior art, obtain 8 directions at the selected pixel point, which are directly above, directly below, directly left, directly right, upper left, lower left, upper right, and lower right. Then, select a pixel point adjacent to the selected pixel point in the corresponding direction as a neighboring pixel point of the selected pixel point. Thus, the set composed of the obtained multiple neighboring pixel points is used as the set of neighboring pixel points of the selected pixel point, and continue to determine the set of neighboring pixel points of the remaining pixel points, thereby obtaining multiple neighboring pixel points corresponding to each pixel point. In other embodiments, it can also be determined by other methods, which will not be elaborated here.

[0103] In some embodiments, determining the spectral difference entropy between each pixel point and the corresponding set of neighboring pixel points according to the spectral distribution characteristics of the hyperspectral image can be achieved by the following steps:

[0104] Select a pixel point as the selected pixel point;

[0105] Determine the spectral angular distance between the selected pixel point and each of the neighboring pixel points in the set of neighboring pixel points corresponding to the selected pixel point according to the spectral vector of the selected pixel point and the spectral vectors of each of the neighboring pixel points in the set of neighboring pixel points corresponding to the selected pixel point. Among them, all the spectral angular distances are the spectral distribution characteristics of the hyperspectral image;

[0106] Determine the spectral relative distribution between the selected pixel point and each of the neighboring pixel points in the set of neighboring pixel points corresponding to the selected pixel point according to all the spectral angular distances;

[0107] Determine the spectral difference entropy between the selected pixel point and the set of neighboring pixel points corresponding to the selected pixel point through all the spectral relative distributions;

[0108] Continue to determine the spectral difference entropy between the remaining pixel points and the set of neighboring pixel points corresponding to them.

[0109] When specifically implemented, determining the spectral angular distance between the selected pixel point and each of the neighboring pixel points in the set of neighboring pixel points corresponding to the selected pixel point according to the spectral vector of the selected pixel point and the spectral vectors of each of the neighboring pixel points in the set of neighboring pixel points corresponding to the selected pixel point can be implemented in the following way, that is: First, select a neighboring pixel point from the set of neighboring pixel points corresponding to the selected pixel point. Then, determine the modulus length of the spectral vector of the selected pixel point and the modulus length of the spectral vector of this neighboring pixel point. Second, after determining the dot product of the spectral vector of the selected pixel point and the spectral vector of this neighboring pixel point, divide the obtained result by the product of the modulus length of the spectral vector of the selected pixel point and the modulus length of the spectral vector of this neighboring pixel point. Finally, take the arccosine of the obtained result as the spectral angular distance between the selected pixel point and this neighboring pixel point. Repeat the above steps to determine the spectral angular distances between the selected pixel point and the remaining neighboring pixel points in the set of neighboring pixel points corresponding to the selected pixel point.

[0110] When specifically implemented, determining the spectral relative distribution between the selected pixel point and each of the neighboring pixel points in the set of neighboring pixel points corresponding to the selected pixel point according to all the spectral angular distances can be implemented in the following way, that is: First, take the spectral angular distance corresponding to each neighboring pixel point as the negative exponent of the natural constant e. Then, sum all the results. Second, select a neighboring pixel point from the set of neighboring pixel points corresponding to the selected pixel point, and take the quotient of the result corresponding to this neighboring pixel point and the sum result as the spectral relative distribution between the selected pixel point and this neighboring pixel point. Repeat the above steps to determine the spectral relative distributions between the selected pixel point and the remaining neighboring pixel points in the set of neighboring pixel points corresponding to the selected pixel point.

[0111] It should be noted that the spectral relative distribution characterizes the similarity degree of the spectral characteristics of the selected pixel point to each of the neighboring pixel points. The larger the spectral relative distribution, the higher the similarity degree of the spectral characteristics of the selected pixel point to each of the neighboring pixel points. The smaller the spectral relative distribution, the lower the similarity degree of the spectral characteristics of the selected pixel point to each of the neighboring pixel points.

[0112] It should be noted that the spectral difference entropy described in this application represents the degree of chaos of all spectral values in the spectral information around the selected pixel point. The difference degree of the spectral information around the selected pixel point can be reflected by the degree of chaos of all spectral values. The greater the spectral difference entropy, the greater the spectral characteristic difference between the selected pixel point and the neighboring pixel points; the smaller the spectral difference entropy, the smaller the spectral characteristic difference between the selected pixel point and the neighboring pixel points. As a preferred embodiment, the spectral difference entropy between the selected pixel point and the corresponding neighboring pixel point set can be determined by the following method through all spectral relative distributions, that is: First, select the spectral relative distribution corresponding to one neighboring pixel point in the neighboring pixel point set corresponding to the selected pixel point. Then, multiply the spectral relative distribution by the logarithmic function value of the spectral relative distribution. Repeat the above steps to determine the results corresponding to the remaining neighboring pixel points in the neighboring pixel point set. Finally, take the negative result of the sum of the results corresponding to all neighboring pixel points as the spectral difference entropy between the selected pixel point and the corresponding neighboring pixel point set.

[0113] In step 105, the feature separation degree of each pixel point in the hyperspectral image is determined by the structural feature vector corresponding to each pixel point and the spectral difference entropy corresponding to each pixel point. The hyperspectral image is secondarily segmented based on the feature separation degrees of each pixel point, and cadastral annotation is performed on the specified area according to the segmentation result.

[0114] In some embodiments, as shown in Figure 3 the figure is a schematic flowchart of determining the feature separation degree according to some embodiments of the present application. The feature separation degree of each pixel point in the hyperspectral image can be determined by the following steps through the structural feature vector corresponding to each pixel point and the spectral difference entropy corresponding to each pixel point:

[0115] First, in 1051, select a pixel point as the selected pixel point;

[0116] Secondly, in 1052, determine the feature variability according to the structural feature vector of the selected pixel point;

[0117] Then, in 1053, determine the feature separation degree of the selected pixel point in the hyperspectral image according to the spectral difference entropy of the selected pixel point and the feature variability;

[0118] Finally, in 1054, continue to determine the feature separation degrees of the remaining pixel points in the hyperspectral image, so as to obtain the feature separation degree of each pixel point in the hyperspectral image.

[0119] In specific implementation, determining the feature variability degree according to the structural feature vector of the selected pixel point can be achieved in the following manner, that is: taking the variance of all spectral values in the structural feature vector of the selected pixel point as the feature variability degree of the selected pixel point. In other embodiments, other methods can also be used to determine it, which is not limited here. Additionally, as a preferred embodiment, determining the feature separation degree of the selected pixel point in the hyperspectral image according to the spectral difference entropy and the feature variability degree of the selected pixel point can be achieved in the following manner, that is: taking the result obtained by dividing the feature variability degree by the spectral difference entropy of the selected pixel point as the feature separation degree of the selected pixel point in the hyperspectral image.

[0120] It should be noted that the feature variability degree in this application characterizes the difference degree of the spectral characteristics of the corresponding pixel point between different bands. The larger the feature variability degree, the higher the difference degree of the spectral characteristics of the corresponding pixel point between different bands; the smaller the feature variability degree, the lower the difference degree of the spectral characteristics of the corresponding pixel point between different bands. Additionally, the feature separation degree is a quantitative index of the discrimination ability of the corresponding pixel point in the hyperspectral image relative to its neighborhood. The larger the feature separation degree, it indicates that the difference between the spectral characteristics of the corresponding pixel point and its neighboring pixel points is greater, that is: it is easier to be recognized and classified; the smaller the feature separation degree, it indicates that the difference between the spectral characteristics of the corresponding pixel point and its neighboring pixel points is smaller, that is: it is not easy to be recognized and classified.

[0121] In some embodiments, performing secondary segmentation on the hyperspectral image based on the feature separation degree of each pixel point and performing cadastral annotation on a specified area according to the segmentation result can be achieved through the following steps:

[0122] Training a classification model through the feature separation degree of each pixel point;

[0123] Performing secondary segmentation on the hyperspectral image through the classification model to obtain the class label of each pixel point in the hyperspectral image;

[0124] Mapping the class labels of all pixel points back to the corresponding pixel positions of the hyperspectral image, thereby obtaining a cadastral annotation map of the same size as the hyperspectral image, so as to achieve cadastral annotation of the specified area.

[0125] It should be noted that pixel points with high feature separation are easier to distinguish from their neighborhoods, and the corresponding pixel points represent important ground feature in a specified area. Therefore, the secondary segmentation in this application refers to using the scikit-learn library in Python to take the feature separation of each pixel point as input, training a classification model, and classifying all pixel points in the hyperspectral image through a support vector machine in the prior art in combination with the classification model to obtain the class label of each pixel point in the hyperspectral image. The class label refers to the ground feature type of the corresponding pixel point, such as vegetation, water body, bare soil, urban buildings, etc.

[0126] In addition, on the other hand of this application, in some embodiments, this application provides a hyperspectral cadastral surveying and mapping system. Referring to Figure 4 , this figure is a structural block diagram of a hyperspectral cadastral surveying and mapping system shown according to some embodiments of this application. The hyperspectral cadastral surveying and mapping system 200 includes: an acquisition module 201, a processing module 202, and an execution module 203, which are described as follows:

[0127] The acquisition module 201. In this application, the acquisition module 201 is mainly used to acquire the hyperspectral image of a specified area after starting the hyperspectral cadastral surveying and mapping;

[0128] The processing module 202. In this application, the processing module 202 is mainly used to determine the vector proximity of the spectral vectors between every two pixel points in the hyperspectral image, and approximately segment the hyperspectral image according to all the vector proximities to obtain a plurality of image segmentation domains;

[0129] In addition, in this application, the processing module 202 is further used for each image segmentation domain to obtain the reference spectral value of the image segmentation domain based on the geomorphic information of the specified area, and then determine the spectral similarity between each pixel point in the image segmentation domain and the geomorphology of the specified area according to the reference spectral value, and construct the features of all pixel points in the image segmentation domain through all the spectral similarities to obtain the structural feature vectors of each pixel point in the image segmentation domain, and then obtain the structural feature vectors of each pixel point in each image segmentation domain;

[0130] In addition, in this application, the processing module 202 is further used to obtain the neighborhood pixel point set of each pixel point in the hyperspectral image, and determine the spectral difference entropy between each pixel point and the corresponding neighborhood pixel point set according to the spectral distribution characteristics of the hyperspectral image;

[0131] An execution module 203. In this application, the execution module 203 is mainly used to determine the feature separation degree of each pixel in the hyperspectral image based on the structural feature vectors corresponding to each pixel and the spectral difference entropy corresponding to each pixel, perform secondary segmentation on the hyperspectral image based on the feature separation degrees of each pixel, and perform cadastral annotation on a specified area according to the segmentation result.

[0132] In addition, this application also provides a computer device, which includes a memory and a processor. The memory stores code, and the processor is configured to obtain the code and execute the above-mentioned hyperspectral cadastral mapping method.

[0133] In some embodiments, refer to Figure 5 , this figure is the internal structure diagram of a computer device applying the hyperspectral cadastral mapping method according to some embodiments of this application. The hyperspectral cadastral mapping method in the above embodiments can be implemented by Figure 5 the computer device shown. This computer device 300 includes at least one processor 301, a communication bus 302, a memory 303, and at least one communication interface 304.

[0134] The processor 301 can be a general-purpose central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more for controlling the execution of the hyperspectral cadastral mapping method in this application.

[0135] The communication bus 302 is used to transmit information between the above components.

[0136] The memory 303 may be a read-only memory (ROM) or other type of static storage device that can store static information and instructions, a random access memory (RAM), or other type of dynamic storage device that can store information and instructions. It may also be an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM), or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), a magnetic disk, or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory 303 may exist independently and be connected to the processor 301 through the communication bus 302. The memory 303 may also be integrated with the processor 301.

[0137] Among them, the memory 303 is used to store the program code for executing the solution of this application and is controlled by the processor 301 to execute. The processor 301 is used to execute the program code stored in the memory 303. The program code may include one or more software modules. The hyperspectral cadastral mapping method in the above embodiments may be implemented by one or more software modules in the program code in the processor 301 and the memory 303.

[0138] The communication interface 304 uses any device such as a transceiver to communicate with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), etc.

[0139] In a specific implementation, as an embodiment, the computer device may include multiple processors, and each of these processors may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. Here, the processor may refer to one or more devices, circuits, and / or processing cores for processing data (such as computer program instructions).

[0140] The above computer device can be a general-purpose computer device or a special-purpose computer device. In specific implementations, the computer device can be a desktop computer, a laptop computer, a network server, a personal digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device, or an embedded device. The embodiments of the present application do not limit the type of the computer device.

[0141] In addition, the present application also provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the above hyperspectral cadastral mapping method is implemented.

[0142] In summary, in the hyperspectral cadastral mapping method, system, device, and storage medium disclosed in the embodiments of the present application, after starting hyperspectral cadastral mapping, a hyperspectral image of a specified area is obtained; the vector proximity degree between the spectral vectors of every two pixel points in the hyperspectral image is determined, and the hyperspectral image is approximately segmented according to all the vector proximity degrees to obtain a plurality of image segmentation domains; for each image segmentation domain, a reference spectral value of the image segmentation domain is obtained based on the geomorphic information of the specified area, and then the spectral similarity between each pixel point in the image segmentation domain and the geomorphology of the specified area is determined according to the reference spectral value, and the structural feature vectors of each pixel point in the image segmentation domain are obtained by constructing features for all the pixel points in the image segmentation domain through all the spectral similarities, and then the structural feature vectors of each pixel point in each image segmentation domain are obtained; the set of neighborhood pixel points of each pixel point in the hyperspectral image is obtained, and the spectral difference entropy between each pixel point and the corresponding set of neighborhood pixel points is determined according to the spectral distribution characteristics of the hyperspectral image; the feature separation degree of each pixel point in the hyperspectral image is determined through the structural feature vector corresponding to each pixel point and the spectral difference entropy corresponding to each pixel point, the hyperspectral image is secondarily segmented based on the feature separation degree of each pixel point, and cadastral annotation is performed on the specified area according to the segmentation result; spectral redundancy in the hyperspectral image can be avoided to improve the ability to distinguish surface objects.

[0143] Although the preferred embodiments of the present application have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications falling within the scope of the present application.

[0144] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application is also intended to include these modifications and variations.

Claims

1. A hyperspectral cadastral mapping method, characterized in that: The steps include: Start hyperspectral cadastral mapping to obtain hyperspectral images of designated areas; Determine the vector approximation of the spectral vector between every two pixel points in the hyperspectral image, and perform approximate segmentation on the hyperspectral image according to all the vector approximations to obtain a plurality of image segmentation domains; For each image segmentation domain, a reference spectral value of the image segmentation domain is obtained based on the topographic information of the specified area, and then the spectral similarity between each pixel point in the image segmentation domain and the topography of the specified area is determined according to the reference spectral value, and all the pixels in the image segmentation domain are feature constructed through all the spectral similarities to obtain the structural feature vector of each pixel point in the image segmentation domain, and then the structural feature vector of each pixel point in each image segmentation domain is obtained; Obtaining a neighborhood pixel point set of each pixel point in the hyperspectral image, and determining a spectral difference entropy between each pixel point and the corresponding neighborhood pixel point set according to spectral distribution characteristics of the hyperspectral image; The characteristic separation degree of each pixel in the hyperspectral image is determined by the structural feature vector corresponding to each pixel and the spectral difference entropy corresponding to each pixel. The hyperspectral image is segmented twice based on the characteristic separation degree of each pixel, and the designated area is marked according to the segmentation result.

2. The method according to claim 1, characterized in that Determining the vector approximation of the spectral vector between every two pixels in the hyperspectral image specifically includes: Selecting a pixel point in the hyperspectral image as a selected pixel point, and using the remaining pixel points in the hyperspectral image after selecting the selected pixel point as a pixel point transition set; Determining the degree of vector proximity between the selected pixel point and the spectral vectors of each pixel point in the pixel point transition set; Selecting a pixel point from all the pixel points in the pixel point transition set as a new selected pixel point, and using the remaining pixel points in the pixel point transition set after selecting the new selected pixel point as a new pixel point transition set; Determine the vector closeness between the spectral vectors of each pixel point in the new selected pixel point and the new pixel point transition set; The same process is repeated until all pixels in the hyperspectral image are selected, thereby obtaining the vector approximation of the spectral vector between every two pixels in the hyperspectral image.

3. The method according to claim 1, characterized in that The hyperspectral image is approximately segmented according to all vector approximations to obtain multiple image segmentation domains, specifically including: Determine a proximity matrix of the hyperspectral image according to the proximity of all vectors; The hyperspectral image is segmented according to the similarity matrix to obtain a plurality of image segmentation domains.

4. The method according to claim 1, characterized in that The reference spectral value of the image segmentation domain is obtained based on the topographic information of the specified area, specifically including: Determine the significant bands corresponding to the geomorphic information of the specified area; Obtaining the significant spectral value of each pixel point in the image segmentation domain under the significant band; The reference spectral value of the image segmentation domain is determined by all the significant spectral values.

5. The method according to claim 1, characterized in that Determining the spectral similarity between each pixel point in the image segmentation domain and the landform in the designated area according to the reference spectral value specifically includes: Determine the characteristic spectral value of each pixel in the image segmentation domain; Determine the standard spectrum value of each pixel point in the image segmentation domain by using the characteristic spectrum value of each pixel point and the reference spectrum value; The spectral similarity between each pixel in the image segmentation domain and the landform in the specified area is determined based on all standard spectral values.

6. The method according to claim 1, characterized in that Through all spectral similarities, all pixels in the image segmentation domain are feature constructed to obtain the structural feature vectors of each pixel in the image segmentation domain, including: Extract feature pixels from all pixels in the image segmentation domain through all spectral similarities; Determine the inter-spectral change rate between each adjacent band in the spectral vector of the characteristic pixel point; Extract multiple characteristic bands from all bands according to the change rates between each spectrum; Select a pixel point in the image segmentation domain as the selected pixel point; The spectral value of the selected pixel point in each characteristic band is constructed as a structural characteristic vector of the selected pixel point; Continue to determine the structural feature vectors of the remaining pixels.

7. The method according to claim 1, characterized in that Determining the feature separation of each pixel in the hyperspectral image by using the structural feature vector corresponding to each pixel and the spectral difference entropy corresponding to each pixel specifically includes: Select a pixel as the selected pixel; Determine the feature variation according to the structural feature vector of the selected pixel point; Determining the feature separation of the selected pixel point in the hyperspectral image according to the spectral difference entropy of the selected pixel point and the feature variability; Continue to determine the feature separation of the remaining pixels in the hyperspectral image, so as to obtain the feature separation of each pixel in the hyperspectral image.

8. A hyperspectral cadastral mapping system, characterized in that: include: The acquisition module is used to obtain the hyperspectral image of the specified area after starting the hyperspectral cadastral mapping; A processing module, used for determining the vector approximation of the spectral vector between every two pixel points in the hyperspectral image, and performing approximate segmentation on the hyperspectral image according to all the vector approximations to obtain a plurality of image segmentation domains; The processing module is further used to obtain, for each image segmentation domain, a reference spectral value of the image segmentation domain based on the landform information of the specified area, and then determine the spectral similarity between each pixel point in the image segmentation domain and the landform of the specified area according to the reference spectral value, perform feature construction on all the pixel points in the image segmentation domain through all the spectral similarities, obtain the structural feature vector of each pixel point in the image segmentation domain, and then obtain the structural feature vector of each pixel point in each image segmentation domain; The processing module is further used to obtain a neighborhood pixel point set of each pixel point in the hyperspectral image, and determine a spectral difference entropy between each pixel point and a corresponding neighborhood pixel point set according to a spectral distribution feature of the hyperspectral image; The execution module is used to determine the feature separation degree of each pixel in the hyperspectral image through the structural feature vector corresponding to each pixel and the spectral difference entropy corresponding to each pixel, perform secondary segmentation on the hyperspectral image based on the feature separation degree of each pixel, and perform cadastral marking on the designated area according to the segmentation result.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the hyperspectral cadastral mapping method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the hyperspectral cadastral mapping method according to any one of claims 1 to 7 are implemented.

Images