Optimization method for pixel spectral database and related device

Abstract

The application relates to the technical field of data processing, and provides a pixel spectral database optimization method and related devices.The method comprises the following steps: acquiring an initial pixel spectral database; performing screening on initial pixel spectral data in the initial pixel spectral database by adopting a linear separability quantitative analysis method to obtain a first screened pixel spectral database subset; performing screening on the first screened pixel spectral database subset by adopting a nonlinear separability method to obtain a second screened pixel spectral database subset, and obtaining an optimized pixel spectral database.The pixel spectral database can be screened twice from the linear angle of low latitude and the nonlinear angle of high latitude, redundant data and noise in the pixel spectral database are removed, the pixel spectral database is optimized, and the data quality of the pixel spectral database is improved.

Description

Technical Field

[0001] This application relates to the field of data processing technology, specifically to a pixel spectral database optimization method and related apparatus. Background Technology

[0002] In the fields of remote sensing image processing and hyperspectral data analysis, pixel spectral databases play a crucial role. However, with the continuous advancement of spectral data acquisition technology, the initial pixel spectral databases often contain a large amount of redundant data and noise, posing a significant challenge to subsequent data processing and analysis.

[0003] Existing pixel spectral database optimization methods mainly rely on simple statistical analysis and threshold screening. Although they can remove some redundant data to a certain extent, they often ignore the complex relationships and intrinsic connections between spectral data, resulting in a large amount of noise and interference information still remaining in the optimized database. Summary of the Invention

[0004] This application provides a pixel spectral database optimization method and related apparatus, which can perform two screenings on the pixel spectral database from a low-latitude linear perspective and a high-latitude nonlinear perspective, eliminating redundant data and noise in the pixel spectral database, thereby optimizing the pixel spectral database.

[0005] A first aspect of this application provides a pixel spectral database optimization method, the method comprising:

[0006] Obtain the initial pixel spectral database;

[0007] The initial pixel spectral data in the initial pixel spectral database were screened using the linear separability quantitative analysis method to obtain a subset of the pixel spectral database after the first screening.

[0008] The nonlinear separability method is used to filter the subset of the pixel spectral database after the first screening, resulting in the subset of the pixel spectral database after the second screening, thus obtaining the optimized pixel spectral database.

[0009] Furthermore, the initial pixel spectral data in the initial pixel spectral database were filtered using a linear separability quantitative analysis method to obtain a subset of the pixel spectral database after the first filtering, including:

[0010] Based on the initial pixel spectral data, confirm the reference spectral data;

[0011] The SAM values ​​of the initial pixel spectral data and the corresponding reference spectral data are calculated using the spectral angle mapping method to obtain the initial pixel spectral SAM values.

[0012] The initial pixel spectral SAM value is normalized to obtain the pixel spectral metric value;

[0013] The SAM threshold was determined using the principle of spectral similarity.

[0014] The initial pixel spectral data is filtered based on the pixel spectral metric, reference spectral data, and SAM threshold to obtain a subset of the pixel spectral database after the first filtering.

[0015] Furthermore, the initial pixel spectral data is filtered based on the pixel spectral metric, reference spectral data, and SAM threshold to obtain a subset of the pixel spectral database after the first filtering, including:

[0016] The pixel spectral measurement value and the reference spectral data are compared, and the spectral data corresponding to the pixel spectral measurement value with the highest similarity and the reference spectral data are extracted.

[0017] Spectral data corresponding to pixel spectral measurement values ​​that do not meet the first condition are removed to obtain pixel spectral data after the first screening. Specifically, the first condition is that the spectral data corresponding to the extracted pixel spectral measurement value and the reference spectral data belong to the same category, the pixel spectral measurement value is the minimum value, and the pixel spectral measurement value is less than the SAM threshold.

[0018] Furthermore, based on the initial pixel spectral data, the reference spectral data is confirmed, including:

[0019] Extract initial pixel spectral label information from the initial pixel spectral data;

[0020] Based on the initial pixel spectral label information, the initial pixel spectral data is classified to obtain initial pixel spectral data of n categories;

[0021] Reference values ​​are calculated for the initial pixel spectral data of the n categories respectively to obtain the reference spectral data of the corresponding initial pixel spectral data.

[0022] Furthermore, a nonlinear separability method is used to further filter the subset of the pixel spectral database after the first screening, resulting in a subset of the pixel spectral database after the second screening, thus obtaining the optimized pixel spectral database, including:

[0023] Extract the k nearest neighbors of the target pixel spectral data and construct a neighborhood relationship graph, wherein the target pixel spectral data is arbitrarily selected pixel spectral data from the subset of the pixel spectral database after the first screening;

[0024] The edge weights of the neighborhood graph are calculated using the Gaussian kernel function;

[0025] The label information of the first optimized pixel spectral data is extracted, and the weighted label maximum value among the k nearest neighbors of the target pixel spectral data is calculated using the Kronecker function to obtain the pixel spectral data corresponding to the weighted label maximum value among the k nearest neighbors of the target pixel spectral data.

[0026] Target pixel spectral data that does not meet the second condition are removed, wherein the second condition is that if the label of the pixel spectral data corresponding to the maximum weighted label is compared with the label of the target pixel spectral data, the labels are consistent;

[0027] The process of repeatedly extracting the k nearest neighbors of the target pixel spectral data, constructing a neighborhood relationship graph, and eliminating target pixel spectral data that does not meet the second condition is repeated until all pixel points in the first optimized pixel spectral data are filtered out, resulting in a subset of the pixel spectral database after the second filtering, and thus the optimized pixel spectral database.

[0028] In this example, the initial pixel spectral data is first screened using a linear separability quantitative analysis method to remove redundant data and noise in the low-dimensional region of the pixel spectral database, resulting in a subset of pixel spectral data after the first screening. Then, a nonlinear separability method is used to screen the pixel spectral data after the first screening, resulting in a subset of pixel spectral data after the second screening. This subset serves as the optimized pixel spectral database, removing the remaining redundant data and noise in the high-dimensional region of the pixel spectral database, thereby optimizing the pixel spectral database and removing most of the redundant data and noise.

[0029] A second aspect of this application provides a pixel spectral database optimization apparatus, the apparatus comprising: a first acquisition unit, configured to acquire an initial pixel spectral database;

[0030] The first processing unit is used to filter the initial pixel spectral data in the initial pixel spectral database using the linear separability quantitative analysis method to obtain a subset of the pixel spectral database after the first filtering.

[0031] The second processing unit is used to filter the subset of the pixel spectral database after the first screening using the nonlinear separability method, to obtain the subset of the pixel spectral database after the second screening, and to obtain the optimized pixel spectral database.

[0032] Furthermore, in the aspect of using the linear separability quantitative analysis method to filter the initial pixel spectral data in the initial pixel spectral database to obtain a subset of the pixel spectral database after the first filtering, the first processing unit is used to:

[0033] Based on the initial pixel spectral data, confirm the reference spectral data;

[0034] The SAM values ​​of the initial pixel spectral data and the corresponding reference spectral data are calculated using the spectral angle mapping method to obtain the initial pixel spectral SAM values.

[0035] The initial pixel spectral SAM value is normalized to obtain the pixel spectral metric value;

[0036] The SAM threshold was determined using the principle of spectral similarity.

[0037] The initial pixel spectral data is filtered based on the pixel spectral metric, reference spectral data, and SAM threshold to obtain a subset of the pixel spectral database after the first filtering.

[0038] Furthermore, in the aspect of using a nonlinear separability method to filter the pixel spectral database subset after the first screening to obtain a second screening subset of the pixel spectral database, thus obtaining an optimized pixel spectral database, the second processing unit is used to:

[0039] Extract the k nearest neighbors of the target pixel spectral data and construct a neighborhood relationship graph, wherein the target pixel spectral data is arbitrarily selected pixel spectral data from the subset of the pixel spectral database after the first screening;

[0040] The edge weights of the neighborhood graph are calculated using the Gaussian kernel function;

[0041] The label information of the first optimized pixel spectral data is extracted, and the weighted label maximum value among the k nearest neighbors of the target pixel spectral data is calculated using the Kronecker function to obtain the pixel spectral data corresponding to the weighted label maximum value among the k nearest neighbors of the target pixel spectral data.

[0042] Target pixel spectral data that does not meet the second condition are removed, wherein the second condition is that if the label of the pixel spectral data corresponding to the maximum weighted label is compared with the label of the target pixel spectral data, the labels are consistent;

[0043] The process of repeatedly extracting the k nearest neighbors of the target pixel spectral data, constructing a neighborhood relationship graph, and eliminating target pixel spectral data that does not meet the second condition is repeated until all pixel points in the first optimized pixel spectral data are filtered out, resulting in a subset of the pixel spectral database after the second filtering, and thus the optimized pixel spectral database.

[0044] A third aspect of this application provides a terminal including a processor, an input device, an output device, and a memory, wherein the processor, input device, output device, and memory are interconnected, wherein the memory is used to store a computer program, the computer program including program instructions, and the processor is configured to invoke the program instructions to execute the step instructions of the pixel spectral database optimization method as described in the first aspect of this application.

[0045] A fourth aspect of this application provides a computer-readable storage medium storing a computer program for electronic data interchange, wherein the computer program causes a computer to perform some or all of the steps described in the pixel spectral database optimization method of the first aspect of this application. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This application provides a schematic diagram of the overall process steps for a pixel spectral database optimization method.

[0048] Figure 2 This application provides a schematic diagram illustrating the first screening steps of a pixel spectral database optimization method in an embodiment of the present application.

[0049] Figure 3 This application provides a schematic diagram illustrating the specific steps of the first screening in a pixel spectral database optimization method.

[0050] Figure 4 This application provides a schematic diagram illustrating the second screening step of a pixel spectral database optimization method in an embodiment of the present application.

[0051] Figure 5 An illustration of the initial pixel spectral library provided as an example of a pixel spectral database optimization method in this application embodiment;

[0052] Figure 6 An example of a reference spectrum illustrating a pixel spectral database optimization method provided in this application embodiment;

[0053] Figure 7 This application provides an example of a pixel spectral database optimization method, illustrating an optimized pixel spectral database.

[0054] Figure 8 A schematic diagram of a pixel spectral database optimization device is provided in this application embodiment;

[0055] Figure 9 This is a schematic diagram of the structure of a terminal provided in an embodiment of this application;

[0056] Figure label:

[0057] First acquisition unit-1, first processing unit-2, second processing unit-3. Detailed Implementation

[0058] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0059] The terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0060] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application can be combined with other embodiments.

[0061] To better understand the pixel spectral database optimization method provided in this application embodiment, the following is a brief introduction to the scenarios in which this method is applied. Pixel spectral quality is easily affected by remote sensing image quality, image processing, and the homogeneity of ground features, and currently there is still no effective quality optimization method. This leads to problems such as confusion, poor separability, and low quality in the acquired spectra, which easily causes error accumulation during analysis and modeling, severely restricting the quantitative survey and monitoring capabilities of natural resource elements.

[0062] The pixel spectral database optimization method is applied to the pixel spectral database optimization device. Figure 1 A schematic diagram illustrating the overall process steps of a pixel spectral database optimization method is shown. Figure 1 As shown, the method includes:

[0063] S1. Obtain the initial pixel spectral database. The initial pixel spectral database should contain label information and spectral information. The initial pixel spectral database can be an existing pixel spectral database, or it can be created by selecting a Region of Interest (ROI) in ENVI software.

[0064] S2. The initial pixel spectral data in the initial pixel spectral database is filtered using a linear separability quantitative analysis method to obtain a subset of the pixel spectral database after the first filtering. In this preferred embodiment, Figure 2 This diagram illustrates the first screening steps of a pixel spectral database optimization method, as shown below. Figure 2 As shown, step S2 includes:

[0065] S201. Based on the initial pixel spectral data, confirm the reference spectral data. Specifically, this includes:

[0066] S2011. Based on the characteristics of the initial pixel spectral data, classify the initial pixel spectral data to obtain n categories of initial pixel spectral data. In this embodiment, the similarity between pixel spectral data is used as a characteristic of pixel spectral data to classify pixel spectral data with the same similarity into the same type. Specifically, this includes:

[0067] Initial pixel spectral label information is extracted from the initial pixel spectral data.

[0068] Based on the initial pixel spectral label information, the initial pixel spectral data is classified to obtain initial pixel spectral data of n categories.

[0069] Reference values ​​are calculated for the initial pixel spectral data of the n categories respectively to obtain the reference spectral data of the corresponding initial pixel spectral data.

[0070] S2012. Calculate reference values ​​for the initial pixel spectral data of each of the n categories to obtain the reference spectral data for the corresponding initial pixel spectral data. Specifically, calculate the average spectral curve of the pixel spectral data for each category sequentially, using it as the reference spectral data for that category. The formula for calculating the reference spectral data is as follows:

[0071]

[0072] Where X represents the spectral information of the initial pixel spectral library, which is a D*N dimensional matrix, D is the number of spectral bands in the initial pixel spectral library, and N is the total number of pixel spectra. The spectral information for the reference spectrum is a D*C matrix, where C is the number of land cover categories. c represents any land cover category, c = {1, 2, 3, ..., C}, and Nc represents the number of pixels in the c-th land cover category in the initial pixel spectral library. Xn represents the spectrum of any pixel in the c-th land cover category, and Xn is a D*1 matrix, n = {1, 2, 3, ..., Nc}. Y represents the tag information for the initial pixel spectrum, a 1*N matrix, where Yn represents the tag of the n-th pixel spectrum in the initial pixel spectrum, n = {1, 2, 3, ..., N}. The reference spectrum's label information is a 1*C dimensional matrix. Yc represents the label of the c-th land cover category; for simplified calculation, Yc = {1,2,3,...,C}. If the initial pixel spectral library also contains temporal and spatial coordinate information, the reference spectrum can also be calculated by using a method that combines contemporaneous high-resolution remote sensing imagery, natural resource land cover type vector data, and the initial pixel spectral spatial overlay. The obtained reference spectrum contains one sample for each land cover category.

[0073] S202. Calculate the SAM values ​​of the initial pixel spectral data and the reference spectral data using the spectral angle mapping method to obtain the initial pixel spectral SAM value, and thus the pixel spectral SAM value F. Specifically, the formula for calculating the SAM value is as follows:

[0074]

[0075] Where F is the SAM value of the pixel spectrum, which is an N*C dimensional matrix, where N is the number of initial pixel spectra and C is the number of land cover categories. Fnc represents the SAM value between the initial pixel spectrum n and the reference spectrum c, where n = {1, 2, 3, ..., N} and c = {1, 2, 3, ..., C}.

[0076] S203. Normalize the initial pixel spectral SAM values ​​to obtain pixel spectral metric values. Following the minimum-maximum normalization calculation rule, calculate the normalized metric value F, ensuring it falls within the [0,1] interval. Specifically, the normalization formula is as follows:

[0077]

[0078] in, Fmin is the minimum value in matrix F, and Fmax is the maximum value in matrix F. It is an N*C dimensional matrix formed after performing MIN-MAX normalization on F.

[0079] S204. The SAM threshold F is determined using the spectral similarity principle. Following the principle that the smaller the SAM value of a spectrum within the same class, the stronger the spectral similarity, an intra-class SAM threshold T1 is set.

[0080] S205. The initial pixel spectral data is filtered based on the pixel spectral metric value, reference spectral data, and SAM threshold to obtain a subset of the pixel spectral database after the first filtering. In this preferred embodiment, Figure 3 The diagram illustrates the specific steps of the first screening in a pixel spectral database optimization method, including:

[0081] S2051. Compare the pixel spectral measurement value and the reference spectral data, and extract the spectral data and reference spectral data corresponding to the pixel spectral measurement value with the highest similarity.

[0082] S2052. Remove the spectral data corresponding to the spectral measurement values ​​of pixels that do not meet the first condition, and obtain the pixel spectral data after the first screening. Specifically, the first condition is that the spectral data corresponding to the extracted pixel spectral measurement values ​​and the reference spectral data belong to the same category, the pixel spectral measurement value is the minimum value, and the pixel spectral measurement value is less than the SAM threshold. Specifically, the following formula is used for discrimination:

[0083] right It is necessary to clarify before making a judgment Is it a similar SAM value or a dissimilar SAM value?

[0084]

[0085] Where S is an N*C dimensional matrix. When Snc = 1, it means that the label information of the nth initial spectrum and the cth reference spectrum are the same. When Snc = 0, it means that the label information of the nth initial spectrum and the cth reference spectrum are different.

[0086] make Compare Is it If the minimum value is found, then determine whether Snc is equal to 1, and... If the pixel is less than or equal to the threshold T1 and meets the above conditions, then the pixel is retained. After each pixel is determined, the refined initial pixel spectrum forms a subset A.

[0087] A represents the spectral information of the spectral subset selected through the above steps. It is a D*M dimensional matrix, where D is the number of spectral bands and M is the number of selected pixel spectra. YA represents the label information of the spectral subset, which is a 1*M dimensional matrix. YAi represents the label of the i-th pixel spectrum in A, where i = {1, 2, 3, ..., M}.i ∈{1,2,3,...,C}.

[0088] S3. The nonlinear separability method is used to filter the subset of the pixel spectral database after the first screening to obtain the subset of the pixel spectral database after the second screening, thus obtaining the optimized pixel spectral database.

[0089] In this preferred embodiment, Figure 4 This diagram illustrates the second screening step in a pixel spectral database optimization method. It should be noted that the first screening removes noise from low-dimensional low-spectral data, while the second screening removes noise from high-dimensional hyperspectral data. Hyperspectral data possesses high-dimensional nonlinear structural characteristics, and manifold learning can effectively uncover the inherent geometric structure of nonlinear data, revealing low-dimensional manifolds and better capturing the key features of low-dimensional spectral mappings. Figure 4 As shown, step S3 includes:

[0090] S301. Extract the k nearest neighbors of the target pixel spectral data and construct a neighborhood relationship graph. The target pixel spectral data is arbitrarily selected from the first optimized pixel spectral data. k is any given number. For each A... i (i = 1, ..., M), determine its nearest neighbors. Let S represent the nearest neighbors of Ai, arranged in ascending order. The neighborhood graph G(V, E) has Ai as a vertex and a spectral divergence of S. SID (A i A j By using edges, a symmetrical and sparse graph structure can be constructed. In this sparse graph structure, each cell is typically connected to only a small number of other cells by edges.

[0091]

[0092] Where, p i =A i / ΣA,q j =A j / ΣA, Ai, Aj are any points in the pixel spectral subset A after the first optimization, S SID (A i A j Let p and q be the spectral information divergence of Ai and Aj, respectively. p = (p1, p2, ..., pq) L )q=(q1,q2,...,q L ), where L is the number of bands in the pixel spectrum. It is the relative entropy of Aj with respect to Ai. It is the relative entropy of Ai with respect to Aj. Where p l =Ail / ∑A i ,q l =A jl / ∑A j , l∈[1,...,L].

[0093] S302. Calculate the edge weights of the neighborhood graph using the Gaussian kernel function method. Specifically, due to the constructed neighborhood graph, the k-nearest neighbors... The degree of contribution to Ai often varies, and this contribution is usually related to the similarity between data points, represented by edge weights. The formula for calculation using the Gaussian kernel function method is as follows:

[0094]

[0095] Where Ai is the spectral data of a pixel arbitrarily selected from the first optimized pixel spectral data, Aj is the nearest neighbor of Ai, and ||Aj| ... i -A j ∥ 2 This represents the distance between Ai and its nearest neighbor Aj; δ is the scale parameter. The edge weights for non-nearest neighbors are set to 0.

[0096] S303. Extract the label information of the first optimized pixel spectral data, and use the Kronecker function to calculate the weighted label maximum value among the k nearest neighbors of the target pixel spectral data to obtain the pixel spectral data corresponding to the weighted label maximum value among the k nearest neighbors of the target pixel spectral data.

[0097] Specifically, the formula for calculating the maximum weighted label value is as follows:

[0098]

[0099] Where δ is the Kronecker function, when YA i If the expression equals c, output 1; otherwise, output 0.

[0100] S304. Remove target pixel spectral data that does not meet the second condition, wherein the second condition is that if the label of the pixel spectral data corresponding to the maximum weighted label value is compared with the label of the target pixel spectral data, the labels are consistent. Specifically, it is determined that if YA i * =YA i If the expression is positive, it indicates that Ai is reconstructable on a nonlinear scale, so keep Ai; otherwise, discard it.

[0101] S305. Repeat the steps of extracting the k nearest neighbors of the target pixel spectral data, constructing a neighborhood relationship graph, and eliminating the target pixel spectral data that does not meet the second condition, until all pixel points in the first optimized pixel spectral data are filtered out, and a subset of the pixel spectral database after the second filtering is obtained.

[0102] S4. Summarize the pixel spectral database subsets after the second screening to obtain the second-optimized pixel spectral subset B as the optimized pixel spectral database.

[0103] In this example, the following case is used as a specific application scenario: the Zhuhai-1 hyperspectral remote sensing image is used as the base image, and preprocessing work such as atmospheric correction and geometric correction is performed on the image.

[0104] The first step involved using ArcGIS 10.6 software to initially screen eight types of land cover samples, including rice and corn, in Nanchuan District, Chongqing, and delineate the pixel sampling area. The boundaries of the sampling area were then refined by referring to a high-precision remote sensing image map, ensuring that the area contained as many clean pixels as possible. Using the Vector to ROI tool in ENVI 5.3's Region of intersect, the vector range of the sampling area was converted into raster data of regions of interest (ROIs) corresponding to the hyperspectral remote sensing image space, and attribute fields were labeled for each raster. Simultaneously, the ROI separability tool was used to perform separability analysis on the ROIs. Based on the analysis results, ROIs with a separability less than 1.8 were re-refined spectrally to ensure that the selected ROIs were as small, numerous, and precise as possible. After multiple rounds of data refinement, an initial pixel spectral library was formed. The number of spectra for each land cover type varied depending on the size of the divided sampling area, with rice having the most initial spectra (1010) and pear trees the fewest (75). The initial pixel spectral library contained a total of 3927 spectra, as shown in the table below.

[0105]

[0106] The initial pixel spectral library is shown as follows Figure 5 As shown.

[0107] The second step involved calculating the reference spectrum for each land cover using the averaging method described below. This work was implemented using MATLAB software.

[0108]

[0109] The obtained reference spectra contain one sample for each land cover category. The sample data are displayed using ENVI 5.3 software; see below for details. Figure 6 As shown.

[0110] The third step involves using the Spectral Angle Mapping (SAM) method to measure the linear separability between spectral data, and conducting a quantitative analysis of the linear separability between the initial pixel spectrum and the reference spectrum. This work was implemented using MATLAB software.

[0111] First, the SAM value F of the initial pixel spectrum and the reference spectrum is calculated using the spectral angle mapping method. Then, the normalized metric value is obtained using the min-max normalization method. This ensures the metric value is within the range [0,1]. Finally, a class threshold T1 = 0.1 is set, and the points with the highest similarity between the initial spectrum and the reference spectrum are compared to determine if they satisfy the condition of being in the same class and... If the value is less than T1, retain the pixels that meet the conditions to form the filtered spectral subset A.

[0112] The filtered spectral subset A contains 3043 spectra, after which spectra with low intra-class similarity were removed.

[0113] The number of rice varieties has been greatly refined. The spectral count of each category is shown in the table below.

[0114]

[0115] The fourth step involves constructing a graph based on the Laplacian eigenmap (LE)-based manifold learning algorithm, utilizing the Laplacian graph to measure the nonlinear separability between spectral data. This work was implemented using MATLAB software.

[0116] First, set the nearest neighbor k=8 and construct the neighborhood graph G(V,E). Then, use the Gaussian kernel function to calculate the edge weights W of the graph, set the scale parameter of the Gaussian kernel function δ=0.01, and use the Kronecker function to calculate the weighted label maximum value among the nearest neighbors based on the label information of the pixel spectral subset. Retain the pixel spectrum that is consistent with its label to obtain the second-optimized pixel spectral subset B.

[0117] The spectral subset B after secondary optimization contains 3019 spectra, eliminating pixels that are nonlinearly irreconstructable. Spectra for types such as pear trees and bamboo forests were further optimized.

[0118]

[0119] The optimized pixel spectral library is shown below. Figure 7 As shown.

[0120] Please see Figure 8 , Figure 8 This application provides a schematic diagram of the structure of a pixel spectral database optimization device. For example... Figure 8 As shown, it includes:

[0121] The first acquisition unit 1 is used to acquire the initial pixel spectral database.

[0122] The first processing unit 2 is used to filter the initial pixel spectral data in the initial pixel spectral database using the linear separability quantitative analysis method to obtain a subset of the pixel spectral database after the first screening.

[0123] The second processing unit 3 is used to filter the subset of the pixel spectral database after the first screening using the nonlinear separability method, to obtain the subset of the pixel spectral database after the second screening, and to obtain the optimized pixel spectral database.

[0124] In this example, the initial pixel spectral data is first screened using a linear separability quantitative analysis method to remove redundant data and noise in the low-dimensional region of the pixel spectral database, resulting in a subset of pixel spectral data after the first screening. Then, a nonlinear separability method is used to screen the pixel spectral data after the first screening, resulting in a subset of pixel spectral data after the second screening. This subset serves as the optimized pixel spectral database, removing the remaining redundant data and noise in the high-dimensional region of the pixel spectral database, thereby optimizing the pixel spectral database and removing most of the redundant data and noise.

[0125] In one possible implementation, in the aspect of filtering the initial pixel spectral data in the initial pixel spectral database using a linear separability quantitative analysis method to obtain a subset of the pixel spectral database after the first filtering, the first processing unit is configured to:

[0126] Based on the initial pixel spectral data, confirm the reference spectral data;

[0127] The SAM value of the reference spectral data corresponding to the initial pixel spectral data is calculated using the spectral angle mapping method to obtain the initial pixel spectral SAM value.

[0128] The initial pixel spectral SAM value is normalized to obtain the pixel spectral metric value;

[0129] The SAM threshold was determined using the principle of spectral similarity.

[0130] The initial pixel spectral data is filtered based on the pixel spectral metric, reference spectral data, and SAM threshold to obtain a subset of the pixel spectral database after the first filtering.

[0131] In one possible implementation, in the aspect of filtering the pixel spectral database subset after the first screening using a nonlinear separability method to obtain a second screening subset of the pixel spectral database, thus obtaining an optimized pixel spectral database, the second processing unit is used for:

[0132] Extract the k nearest neighbors of the target pixel spectral data and construct a neighborhood relationship graph, wherein the target pixel spectral data is arbitrarily selected pixel spectral data from the subset of the pixel spectral database after the first screening;

[0133] The edge weights of the neighborhood graph are calculated using the Gaussian kernel function;

[0134] The label information of the first optimized pixel spectral data is extracted, and the weighted label maximum value among the k nearest neighbors of the target pixel spectral data is calculated using the Kronecker function to obtain the pixel spectral data corresponding to the weighted label maximum value among the k nearest neighbors of the target pixel spectral data.

[0135] Target pixel spectral data that does not meet the second condition are removed, wherein the second condition is that if the label of the pixel spectral data corresponding to the maximum weighted label is compared with the label of the target pixel spectral data, the labels are consistent;

[0136] The process of repeatedly extracting the k nearest neighbors of the target pixel spectral data, constructing a neighborhood relationship graph, and eliminating target pixel spectral data that does not meet the second condition is repeated until all pixel points in the first optimized pixel spectral data are filtered out, resulting in a subset of the pixel spectral database after the second filtering, and thus the optimized pixel spectral database.

[0137] For examples consistent with the above embodiments, please refer to... Figure 9 , Figure 9 A schematic diagram of a terminal structure provided in an embodiment of this application is shown in the figure. It includes a processor, an input device, an output device, and a memory. The processor, input device, output device, and memory are interconnected. The memory is used to store a computer program, which includes program instructions. The processor is configured to call the program instructions. The program includes instructions for performing the following steps.

[0138] Obtain the initial pixel spectral database;

[0139] The initial pixel spectral data in the initial pixel spectral database were screened using the linear separability quantitative analysis method to obtain a subset of the pixel spectral database after the first screening.

[0140] The nonlinear separability method is used to filter the subset of the pixel spectral database after the first screening, resulting in the subset of the pixel spectral database after the second screening, thus obtaining the optimized pixel spectral database.

[0141] In this example, the initial pixel spectral data is first screened using a linear separability quantitative analysis method to remove redundant data and noise in the low-dimensional region of the pixel spectral database, resulting in a subset of pixel spectral data after the first screening. Then, a nonlinear separability method is used to screen the pixel spectral data after the first screening, resulting in a subset of pixel spectral data after the second screening. This subset serves as the optimized pixel spectral database, removing the remaining redundant data and noise in the high-dimensional region of the pixel spectral database, thereby optimizing the pixel spectral database and removing most of the redundant data and noise.

[0142] The above mainly describes the solutions of the embodiments of this application from the perspective of the method execution process. It is understood that, in order to achieve the above functions, the terminal includes the corresponding hardware structure and / or software modules for executing each function. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments provided herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0143] This application embodiment can divide the terminal into functional units according to the above method example. For example, each function can be divided into a separate functional unit, or two or more functions can be integrated into one processing unit. The integrated unit can be implemented in hardware or as a software functional unit. It should be noted that the unit division in this application embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0144] This application also provides a computer storage medium storing a computer program for electronic data interchange, which causes a computer to perform some or all of the steps of any of the pixel spectral database optimization methods described in the above method embodiments.

[0145] This application also provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program that causes a computer to perform some or all of the steps of any of the pixel spectral database optimization methods described in the above method embodiments.

[0146] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0147] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0148] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical or other forms.

[0149] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0150] Furthermore, the functional units in the various embodiments of the application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software program module.

[0151] If the integrated unit is implemented as a software program module and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0152] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage device, which may include: a flash drive, a read-only memory, a random access memory, a magnetic disk, or an optical disk, etc.

[0153] The embodiments of this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for optimizing a pixel spectral database, characterized in that, include: Obtain the initial pixel spectral database; The initial pixel spectral data in the initial pixel spectral database were filtered using a linear separability quantitative analysis method, resulting in a subset of the pixel spectral database after the first filtering, including: Based on the initial pixel spectral data, confirm the reference spectral data; The SAM values ​​of the initial pixel spectral data and the reference spectral data are calculated using the spectral angle mapping method to obtain the initial pixel spectral SAM value. The initial pixel spectral SAM value is normalized to obtain the pixel spectral metric value; The initial pixel spectral data is filtered based on the pixel spectral metric, reference spectral data, and SAM threshold to obtain a subset of the pixel spectral database after the first filtering. The nonlinear separability method is used to filter the subset of the pixel spectral database after the first screening, resulting in a subset of the pixel spectral database after the second screening, thus obtaining the optimized pixel spectral database, which includes: Extract the k nearest neighbors of the target pixel spectral data and construct a neighborhood relationship graph, wherein the target pixel spectral data is arbitrarily selected pixel spectral data from the subset of the pixel spectral database after the first screening; The edge weights of the neighborhood graph are calculated using the Gaussian kernel function; The label information of the first optimized pixel spectral data is extracted, and the weighted label maximum value among the k nearest neighbors of the target pixel spectral data is calculated using the Kronecker function to obtain the pixel spectral data corresponding to the weighted label maximum value among the k nearest neighbors of the target pixel spectral data. Target pixel spectral data that does not meet the second condition are removed, wherein the second condition is that if the label of the pixel spectral data corresponding to the maximum weighted label is compared with the label of the target pixel spectral data, the labels are consistent; The process of repeatedly extracting the k nearest neighbors of the target pixel spectral data, constructing a neighborhood relationship graph, and eliminating target pixel spectral data that does not meet the second condition is repeated until all pixel points in the first optimized pixel spectral data are filtered out, resulting in a subset of the pixel spectral database after the second filtering, and thus the optimized pixel spectral database.

2. The pixel spectral database optimization method according to claim 1, characterized in that, The initial pixel spectral data is filtered based on the pixel spectral metric, reference spectral data, and SAM threshold to obtain a subset of the pixel spectral database after the first filtering, including: The pixel spectral measurement value and the reference spectral data are compared, and the spectral data corresponding to the pixel spectral measurement value with the highest similarity and the reference spectral data are extracted. Spectral data corresponding to pixel spectral measurement values ​​that do not meet the first condition are removed to obtain pixel spectral data after the first screening. Specifically, the first condition is that the spectral data corresponding to the extracted pixel spectral measurement value and the reference spectral data belong to the same category, the pixel spectral measurement value is the minimum value, and the pixel spectral measurement value is less than the SAM threshold.

3. The pixel spectral database optimization method according to claim 1, characterized in that, Based on the initial pixel spectral data, the reference spectral data is confirmed, including: Extract initial pixel spectral label information from the initial pixel spectral data; Based on the initial pixel spectral label information, the initial pixel spectral data is classified to obtain initial pixel spectral data of n categories; Reference values ​​are calculated for the initial pixel spectral data of the n categories respectively to obtain the reference spectral data of the corresponding initial pixel spectral data.

4. A pixel spectral database optimization device, characterized in that, include: The first acquisition unit is used to acquire the initial pixel spectral database; The first processing unit is used to filter the initial pixel spectral data in the initial pixel spectral database using a linear separability quantitative analysis method, to obtain a subset of the pixel spectral database after the first filtering, including: Based on the initial pixel spectral data, confirm the reference spectral data; The SAM values ​​of the initial pixel spectral data and the reference spectral data are calculated using the spectral angle mapping method to obtain the initial pixel spectral SAM value. The initial pixel spectral SAM value is normalized to obtain the pixel spectral metric value; The initial pixel spectral data is filtered based on the pixel spectral metric, reference spectral data, and SAM threshold to obtain a subset of the pixel spectral database after the first filtering. The second processing unit is used to filter the pixel spectral database subset after the first screening using a nonlinear separability method, to obtain the pixel spectral database subset after the second screening, resulting in an optimized pixel spectral database, including: Extract the k nearest neighbors of the target pixel spectral data and construct a neighborhood relationship graph, wherein the target pixel spectral data is arbitrarily selected pixel spectral data from the subset of the pixel spectral database after the first screening; The edge weights of the neighborhood graph are calculated using the Gaussian kernel function; The label information of the first optimized pixel spectral data is extracted, and the weighted label maximum value among the k nearest neighbors of the target pixel spectral data is calculated using the Kronecker function to obtain the pixel spectral data corresponding to the weighted label maximum value among the k nearest neighbors of the target pixel spectral data. Target pixel spectral data that does not meet the second condition are removed, wherein the second condition is that if the label of the pixel spectral data corresponding to the maximum weighted label is compared with the label of the target pixel spectral data, the labels are consistent; The process of repeatedly extracting the k nearest neighbors of the target pixel spectral data, constructing a neighborhood relationship graph, and eliminating target pixel spectral data that does not meet the second condition is repeated until all pixel points in the first optimized pixel spectral data are filtered out, resulting in a subset of the pixel spectral database after the second filtering, and thus the optimized pixel spectral database.

5. A terminal, characterized in that, The device includes a processor, an input device, an output device, and a memory, which are interconnected. The memory stores a computer program, which includes program instructions. The processor is configured to invoke the program instructions to execute the pixel spectral database optimization method as described in any one of claims 1-3.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, the computer program including program instructions that, when executed by a processor, cause the processor to perform the pixel spectral database optimization method as described in any one of claims 1-3.

Images