Soil arsenic concentration spatial distribution inversion method, device and computer equipment

Abstract

This application relates to a method, apparatus, and computer equipment for inverting the spatial distribution of soil arsenic concentration. The method includes: acquiring a satellite hyperspectral image; constructing a key feature hyperspectral index for each pixel based on the spectral data corresponding to each pixel in the satellite hyperspectral image; and performing inversion processing on the key feature hyperspectral index of each pixel using an inversion model to obtain a spatial distribution inversion image of soil arsenic concentration corresponding to the satellite hyperspectral image. The pixel value corresponding to any pixel in the spatial distribution inversion image of soil arsenic concentration is used to characterize the arsenic concentration of the soil corresponding to that pixel. This method can improve the detection efficiency of soil arsenic concentration.

Description

Technical Field

[0001] This application relates to the field of environmental remote sensing monitoring technology, and in particular to a method, apparatus and computer equipment for inverting the spatial distribution of soil arsenic concentration. Background Technology

[0002] Healthy soil is essential for ensuring the safety of human living environments and food safety, and is a crucial foundation for ecological civilization. However, due to the long-term extensive development of industry and agriculture, large amounts of pollutants have entered the soil through various pathways, posing a serious threat to soil environmental quality. High-precision soil pollution surveys are the cornerstone for achieving efficient and accurate soil pollution remediation.

[0003] Traditional soil pollution investigations are mainly conducted through on-site sampling, laboratory testing, and geostatistical analysis. This often requires a significant amount of manpower, resources, and time, and the investigation process is complex and lacks real-time responsiveness. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, apparatus, and computer equipment for inverting the spatial distribution of soil arsenic concentration that can improve the detection efficiency of soil arsenic concentration, in response to the above-mentioned technical problems.

[0005] In a first aspect, this application provides a method for inverting the spatial distribution of soil arsenic concentration, the method comprising:

[0006] Acquire satellite hyperspectral images;

[0007] Based on the spectral data corresponding to each pixel in the satellite hyperspectral image, construct the key feature hyperspectral index of each pixel;

[0008] The key feature hyperspectral index of each pixel is inverted using an inversion model to obtain a spatial distribution inversion image of soil arsenic concentration corresponding to the satellite hyperspectral image. The pixel value corresponding to any pixel in the spatial distribution inversion image of soil arsenic concentration is used to characterize the arsenic concentration of the soil corresponding to that pixel.

[0009] In one embodiment, the inversion model includes a prediction model and an image generation module. The step of inverting the key feature hyperspectral index of each pixel using the inversion model to obtain an inverted image of the spatial distribution of soil arsenic concentration corresponding to the satellite hyperspectral image includes:

[0010] The hyperspectral index of the key feature of each pixel is predicted using the prediction model to obtain the arsenic concentration corresponding to each pixel.

[0011] The image generation module inverts the arsenic concentration corresponding to each pixel to obtain the spatial distribution inversion image of soil arsenic concentration corresponding to the satellite hyperspectral image.

[0012] In one embodiment, the method further includes:

[0013] Obtain sample satellite hyperspectral images of the study area and the arsenic concentration of the soil at each sampling point in the sample satellite hyperspectral images;

[0014] Based on the sample spectral data corresponding to each sampling point in the sample satellite hyperspectral image, construct the sample spectral index corresponding to each sampling point;

[0015] For any of the sampling points, determine the hyperspectral index of the key features of the sample from the sample spectral index corresponding to the sampling point;

[0016] The hyperspectral index of the key feature of the sample corresponding to the sampling point is predicted by the initial prediction model to obtain the predicted arsenic concentration corresponding to the sampling point.

[0017] Based on the predicted arsenic concentration corresponding to the sampling point and the arsenic concentration corresponding to the soil at the sampling point, the initial prediction model is constructed, and the prediction model is obtained.

[0018] In one embodiment, acquiring the sample satellite hyperspectral image includes:

[0019] Acquire satellite hyperspectral images of the study area;

[0020] The hyperspectral image of the satellite is corrected to obtain a sample satellite hyperspectral image.

[0021] In one embodiment, constructing the sample spectral index corresponding to each sampling point based on the sample spectral data corresponding to each sampling point in the sample satellite hyperspectral image includes:

[0022] The sample spectral data corresponding to each sampling point in the sample satellite hyperspectral image are subjected to noise reduction processing to obtain the noise-reduced sample spectral data.

[0023] Based on the sample spectral data and / or the denoised sample spectral data, the sample spectral index corresponding to each sampling point is constructed using a single-band construction method and / or a dual-band construction method.

[0024] In one embodiment, determining the hyperspectral index of key sample features from the sample spectral index corresponding to any one of the sampling points includes:

[0025] For any of the sampling points, the correlation coefficient between the spectral index of the sample corresponding to each sampling point and the arsenic concentration of the soil at each sampling point is determined based on the spectral index of the sample corresponding to each sampling point and the arsenic concentration of the soil at each sampling point.

[0026] The spectral index of the sample whose absolute value of the correlation coefficient is greater than or equal to the correlation coefficient threshold is taken as the hyperspectral index of the key feature of the sample.

[0027] In one embodiment, the method further includes:

[0028] Determine the target spectral bands and target construction methods used to construct the hyperspectral index of the key features of the sample;

[0029] The step of constructing a key feature hyperspectral index for each pixel based on the spectral data corresponding to each pixel in the satellite hyperspectral image includes:

[0030] For any pixel in the satellite hyperspectral image, a key feature hyperspectral index is constructed based on the target spectral band in the spectral data corresponding to the pixel and the target construction method.

[0031] Secondly, this application also provides a device for inverting the spatial distribution of soil arsenic concentration. The device includes:

[0032] The first acquisition module is used to acquire satellite hyperspectral images;

[0033] The first construction module is used to construct the key feature hyperspectral index of each pixel based on the spectral data corresponding to each pixel in the satellite hyperspectral image;

[0034] The inversion module is used to invert the key feature hyperspectral index of each pixel through an inversion model to obtain the spatial distribution inversion image of soil arsenic concentration corresponding to the satellite hyperspectral image. The pixel value corresponding to any pixel in the spatial distribution inversion image of soil arsenic concentration is used to characterize the arsenic concentration of the soil corresponding to the pixel.

[0035] In one embodiment, the inversion model includes a prediction model and an image generation module, and the inversion model is further configured to:

[0036] The hyperspectral index of the key feature of each pixel is predicted using the prediction model to obtain the arsenic concentration corresponding to each pixel.

[0037] The image generation module inverts the arsenic concentration corresponding to each pixel to obtain the spatial distribution inversion image of soil arsenic concentration corresponding to the satellite hyperspectral image.

[0038] In one embodiment, the device further includes:

[0039] The second acquisition module is used to acquire sample satellite hyperspectral images of the study area and the arsenic concentration of soil at each sampling point in the sample satellite hyperspectral images;

[0040] The second construction module is used to construct the sample spectral index corresponding to each sampling point based on the sample spectral data corresponding to each sampling point in the sample satellite hyperspectral image;

[0041] The first determining module is used to determine the hyperspectral index of key features of a sample from the sample spectral index corresponding to any of the sampling points;

[0042] The prediction module is used to predict the hyperspectral index of the key feature of the sample corresponding to the sampling point through an initial prediction model, so as to obtain the predicted arsenic concentration corresponding to the sampling point.

[0043] The training module is used to train the initial prediction model based on the predicted arsenic concentration corresponding to the sampling point and the arsenic concentration corresponding to the soil at the sampling point, so as to obtain the prediction model.

[0044] In one embodiment, the second acquisition module is further configured to:

[0045] Acquire satellite hyperspectral images of the study area;

[0046] The hyperspectral image of the satellite is corrected to obtain a sample satellite hyperspectral image.

[0047] In one embodiment, the second building module is further configured to:

[0048] The sample spectral data corresponding to each sampling point in the sample satellite hyperspectral image are subjected to noise reduction processing to obtain the noise-reduced sample spectral data.

[0049] Based on the sample spectral data and / or the denoised sample spectral data, the sample spectral index corresponding to each sampling point is constructed using a single-band construction method and / or a dual-band construction method.

[0050] In one embodiment, the first determining module is further configured to:

[0051] For any of the sampling points, the correlation coefficient between the spectral index of the sample corresponding to each sampling point and the arsenic concentration of the soil at each sampling point is determined based on the spectral index of the sample corresponding to each sampling point and the arsenic concentration of the soil at each sampling point.

[0052] The spectral index of the sample whose absolute value of the correlation coefficient is greater than or equal to the correlation coefficient threshold is taken as the hyperspectral index of the key feature of the sample.

[0053] In one embodiment, the device further includes:

[0054] The second determining module is used to determine the target spectral band and target construction method for constructing the hyperspectral index of the key features of the sample.

[0055] The first building module is also used for:

[0056] For any pixel in the satellite hyperspectral image, a key feature hyperspectral index is constructed based on the target spectral band in the spectral data corresponding to the pixel and the target construction method.

[0057] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the above-described method for inverting the spatial distribution of soil arsenic concentration.

[0058] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of the above-described method for inverting the spatial distribution of soil arsenic concentration.

[0059] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the steps of the above-described method for inverting the spatial distribution of soil arsenic concentration.

[0060] The aforementioned method, apparatus, and computer equipment for inverting the spatial distribution of soil arsenic concentration, after acquiring satellite hyperspectral images, constructs key feature hyperspectral indices for each pixel based on the spectral data corresponding to each pixel in the satellite hyperspectral images. Then, through an inversion model, the key feature hyperspectral indices of each pixel are inverted to obtain a spatial distribution inversion image of soil arsenic concentration corresponding to the satellite hyperspectral image. The pixel value corresponding to any pixel in the spatial distribution inversion image of soil arsenic concentration is used to characterize the arsenic concentration in the soil corresponding to that pixel. According to the method, apparatus, and computer equipment for inverting the spatial distribution of soil arsenic concentration provided in this application, the inversion model can be used to invert real-time acquired satellite hyperspectral images to obtain a spatial distribution inversion image of soil arsenic concentration in real time. This reduces manual intervention, thus reducing the consumption of manpower, material resources, and time, lowering the complexity of the detection process, and improving the efficiency and accuracy of predicting arsenic concentration in soil. Attached Figure Description

[0061] Figure 1 This is a flowchart illustrating the method for inverting the spatial distribution of soil arsenic concentration in one embodiment;

[0062] Figure 2 This is a flowchart illustrating the method for inverting the spatial distribution of soil arsenic concentration in one embodiment;

[0063] Figure 3 This is a flowchart illustrating the method for inverting the spatial distribution of soil arsenic concentration in one embodiment;

[0064] Figure 4 This is a flowchart illustrating the method for inverting the spatial distribution of soil arsenic concentration in one embodiment;

[0065] Figure 5 This is a flowchart illustrating the method for inverting the spatial distribution of soil arsenic concentration in one embodiment;

[0066] Figure 6 This is a flowchart illustrating the method for inverting the spatial distribution of soil arsenic concentration in one embodiment;

[0067] Figure 7 This is a schematic diagram of a method for inverting the spatial distribution of soil arsenic concentration in one embodiment;

[0068] Figures 8a-8b This is a schematic diagram of a method for inverting the spatial distribution of soil arsenic concentration in one embodiment;

[0069] Figures 9a-9b This is a schematic diagram of a method for inverting the spatial distribution of soil arsenic concentration in one embodiment;

[0070] Figure 10 This is a structural block diagram of a soil arsenic concentration spatial distribution inversion device in one embodiment;

[0071] Figure 11 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0072] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0073] In one embodiment, such as Figure 1 As shown, a method for inverting the spatial distribution of soil arsenic concentration is provided. This embodiment illustrates the application of this method to a terminal. It is understood that this method can also be applied to a server, and to a system including both a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0074] Step 102: Acquire satellite hyperspectral images.

[0075] In this embodiment of the application, satellite hyperspectral images corresponding to the study area where soil arsenic concentration testing is to be conducted can be acquired. For example, hyperspectral images of the area to be tested can be found on the Natural Resources Remote Sensing Satellite Cloud Service Platform, images with zero or minimal cloud cover in the area to be tested can be selected, and these images can be downloaded as the satellite hyperspectral images corresponding to the area to be tested.

[0076] Step 104: Construct the key feature hyperspectral index of each pixel based on the spectral data corresponding to each pixel in the satellite hyperspectral image.

[0077] In this embodiment, after obtaining a satellite hyperspectral image, a key feature hyperspectral index for each pixel can be constructed based on the spectral data corresponding to each pixel in the satellite hyperspectral image (spectral data includes multiple bands that make up the pixel; the spectral data mentioned in this embodiment are all hyperspectral data, and will not be specifically described in the following embodiments). For example, the key feature hyperspectral index of the pixel can be constructed using single-band and / or dual-band methods. For instance, single-band construction methods may include, but are not limited to, ln(a), 1 / a, 1 / ln(a), ln(1 / a), etc.; dual-band construction methods may include, but are not limited to, a×b, a / b, etc. The methods are (ab) / (a+b), where a and b are used to represent any band in the spectral data.

[0078] Step 106: The key feature hyperspectral index of each pixel is inverted using the inversion model to obtain the inversion image of the spatial distribution of soil arsenic concentration corresponding to the satellite hyperspectral image. The pixel value corresponding to any pixel in the inversion image of the spatial distribution of soil arsenic concentration is used to characterize the arsenic concentration of the soil corresponding to the pixel.

[0079] In this embodiment, after obtaining the key feature hyperspectral index corresponding to each pixel in the satellite hyperspectral image, the key feature hyperspectral index corresponding to each pixel can be input into the inversion model for inversion processing to obtain the inversion image of the spatial distribution of soil arsenic concentration corresponding to the satellite hyperspectral image. For example, the inversion model can predict the arsenic concentration corresponding to each pixel based on the key feature hyperspectral index of each pixel, and then generate the inversion image of the spatial distribution of soil arsenic concentration corresponding to the satellite hyperspectral image based on the arsenic concentration of each pixel and the location information of each pixel in the satellite hyperspectral image.

[0080] It should be noted that the inversion model is a pre-built machine learning model for predicting soil arsenic concentration. For example, the inversion model can be a model pre-built based on multiple linear regression, decision tree regression, random forest regression, support vector machine regression, partial least squares regression, principal component regression, neural network regression, etc. In this embodiment, no specific limitation is made on the network structure of the inversion model.

[0081] The soil arsenic concentration spatial distribution inversion method provided in this application involves acquiring a satellite hyperspectral image, constructing a key feature hyperspectral index for each pixel based on the spectral data corresponding to each pixel in the satellite hyperspectral image, and then performing inversion processing on the key feature hyperspectral index of each pixel using an inversion model to obtain a soil arsenic concentration spatial distribution inversion image corresponding to the satellite hyperspectral image. The pixel value corresponding to any pixel in the soil arsenic concentration spatial distribution inversion image is used to characterize the arsenic concentration in the soil corresponding to that pixel. According to the soil arsenic concentration spatial distribution inversion method provided in this application, the inversion model can perform inversion processing on the real-time acquired satellite hyperspectral image to obtain the soil arsenic concentration spatial distribution inversion image in real time. This reduces manual intervention, thus reducing the consumption of manpower, material resources, and time, lowering the complexity of the detection process, and improving the efficiency and accuracy of predicting arsenic concentration in soil.

[0082] In one embodiment, the inversion model may include a prediction model and an image generation module, referring to... Figure 2 In step 106 above, the key feature hyperspectral index of each pixel is inverted using an inversion model to obtain an inverted image of the spatial distribution of soil arsenic concentration corresponding to the satellite hyperspectral image, including:

[0083] Step 202: The key feature hyperspectral index of each pixel is predicted using a prediction model to obtain the arsenic concentration corresponding to each pixel.

[0084] Step 204: The arsenic concentration corresponding to each pixel is inverted through the image generation module to obtain the spatial distribution inversion image of soil arsenic concentration corresponding to the satellite hyperspectral image.

[0085] In this embodiment of the application, the inversion model may include a prediction model and an image generation module. The prediction model is used to predict the arsenic concentration corresponding to each pixel in the satellite hyperspectral image, and the image generation module can generate a corresponding spatial distribution inversion image of soil arsenic concentration based on the arsenic concentration corresponding to each pixel in the satellite hyperspectral image.

[0086] The key feature hyperspectral index corresponding to each pixel in the satellite hyperspectral image can be input into the prediction model, and the output of the prediction model is the arsenic concentration corresponding to each pixel. Furthermore, the image generation module can generate an inversion image of the spatial distribution of soil arsenic concentration corresponding to the satellite hyperspectral image based on the arsenic concentration corresponding to each pixel.

[0087] In one embodiment, refer to Figure 3 The above methods may also include:

[0088] Step 302: Obtain the sample satellite hyperspectral image of the study area and the arsenic concentration of the soil at each sampling point in the sample satellite hyperspectral image;

[0089] Step 304: Based on the sample spectral data corresponding to each sampling point in the sample satellite hyperspectral image, construct the sample spectral index corresponding to each sampling point;

[0090] Step 306: For any of the sampling points, determine the hyperspectral index of the key features of the sample from the sample spectral index corresponding to the sampling point;

[0091] Step 308: The hyperspectral index of the key feature of the sample corresponding to the sampling point is predicted by the initial prediction model to obtain the predicted arsenic concentration corresponding to the sampling point.

[0092] Step 310: Based on the predicted arsenic concentration at the sampling point and the arsenic concentration in the soil at the sampling point, construct an initial prediction model to obtain the prediction model.

[0093] In this embodiment, the prediction model can be constructed using methods such as multiple linear regression, decision tree regression, random forest regression, support vector machine regression, partial least squares regression, principal component regression, and neural network regression. The specific prediction process will not be described in detail in this embodiment. The following uses the neural network regression method to construct the prediction model as an example to illustrate the construction process of the prediction model in this embodiment.

[0094] In this embodiment, sampling points can be divided within the study area, and soil samples can be taken from these points. For example, a systematic sampling method can be used to deploy sampling points within the area to be tested. During sampling, the latitude and longitude of each sampling point can be determined using GPS (Global Positioning System). After pre-processing the soil samples (for example, pre-processing may include drying, removing weeds, gravel, and other impurities, grinding, and sieving), the arsenic concentration corresponding to the soil sample at each sampling point can be determined. For example, the arsenic concentration in the soil can be determined using inductively coupled plasma mass spectrometry (ICP-MS).

[0095] You can search for satellite hyperspectral images of the study area taken at the sampling time on the Natural Resources Remote Sensing Satellite Cloud Service Platform, select images with zero or minimal cloud cover for download, and perform image correction and other processing on the downloaded satellite hyperspectral images. Then, use the corrected satellite hyperspectral images as sample satellite hyperspectral images.

[0096] It should be noted that the sample satellite hyperspectral image and the satellite hyperspectral image in the aforementioned embodiments can be the same hyperspectral image or different hyperspectral images. This application does not make specific limitations on this.

[0097] Obtain the corresponding sample spectral data of each sampling point in the sample satellite hyperspectral image. For the sample spectral data corresponding to any sampling point, the sample spectral index corresponding to the sampling point can be constructed according to the spectral band corresponding to the sample spectral data. The specific method of constructing the sample spectral index can refer to the method of constructing the key feature hyperspectral index in the previous embodiment, and will not be repeated here in the embodiment of this application.

[0098] For any given sampling point, multiple sample spectral indices can be constructed. Based on the correlation between each sample spectral index and the arsenic concentration at the sampling point, a sample spectral index significantly correlated with the soil arsenic concentration can be selected as the key feature hyperspectral index of the sample. The key feature hyperspectral index corresponding to the sampling point is used as input information for the initial prediction model. After the initial prediction model processes the key feature hyperspectral index, the predicted arsenic concentration corresponding to the sampling point can be obtained. The model loss of the initial prediction model is calculated by comparing the predicted arsenic concentration with the arsenic concentration in the soil at the sampling point. If the model loss does not meet the training requirements (e.g., the model loss is greater than or equal to a preset loss threshold), the model parameters of the initial prediction model are adjusted, and iterative training continues until the model loss of the initial prediction model meets the training requirements (e.g., the model loss is less than a preset loss threshold). Training is then stopped, and the trained prediction model is obtained.

[0099] After training to obtain the prediction model, an inversion model can be constructed using the prediction model and the image generation module, and then the satellite hyperspectral image can be inverted using the inversion model.

[0100] Based on the soil arsenic concentration spatial distribution inversion method provided in this application, an inversion model can be constructed through the trained prediction model, and the inversion model can be used to invert the real-time acquired satellite hyperspectral images to obtain the soil arsenic concentration spatial distribution inversion image in real time. This can reduce human intervention, that is, reduce the consumption of manpower, material resources and time, reduce the complexity of the detection process, and improve the efficiency and accuracy of arsenic concentration prediction in soil.

[0101] In one embodiment, refer to Figure 4 As shown, step 302, acquiring sample satellite hyperspectral images of the study area, may include:

[0102] Step 402: Obtain satellite hyperspectral images of the study area;

[0103] Step 404: Correct the satellite hyperspectral image to obtain the sample satellite hyperspectral image.

[0104] In this embodiment of the application, after acquiring a satellite hyperspectral image, the satellite hyperspectral image can be corrected, and the corrected satellite hyperspectral image can be used as a sample satellite hyperspectral image. For example, the correction processing for the satellite hyperspectral image can include correction methods such as geometric correction and radiometric correction. Geometric correction and radiometric correction can eliminate the reflected radiation of substances such as water and carbon dioxide in the atmosphere, reducing the impact of these noises on the spectrum.

[0105] For example, in the process of geometric correction of satellite hyperspectral images, ground control points can be selected, and pixel spatial coordinate transformation of satellite hyperspectral images can be performed based on ground control points. Then, pixel grayscale resampling can be used to reconstruct the pixel values ​​of each pixel in the image to obtain the geometrically corrected satellite hyperspectral image.

[0106] During the radiometric correction of geometrically corrected satellite hyperspectral images, radiometric calibration can be performed on the geometrically corrected satellite hyperspectral images. Then, the initial pixel values ​​in the geometrically corrected satellite hyperspectral images can be converted into actual surface reflectance through FLAASH atmospheric correction. Actual surface reflectance is used to characterize the percentage of light radiation energy reflected after sunlight shines on ground objects relative to the total radiation energy.

[0107] Based on the spatial distribution inversion method of soil arsenic concentration provided in this application, by correcting the acquired satellite hyperspectral images and using the corrected sample satellite hyperspectral images to construct a prediction model, the influence of noise such as reflected radiation from substances such as water and carbon dioxide in the atmosphere on the spectrum can be reduced, thereby improving the prediction accuracy of the prediction model.

[0108] In one embodiment, refer to Figure 5 As shown, in step 304, based on the sample spectral data corresponding to each sampling point in the sample satellite hyperspectral image, the sample spectral index corresponding to each sampling point is constructed, including:

[0109] Step 502: Denoise the sample spectral data corresponding to each sampling point in the sample satellite hyperspectral image to obtain the denoised sample spectral data;

[0110] Step 504: Based on the sample spectral data and / or the denoised sample spectral data, construct the sample spectral index corresponding to each sampling point using a single-band construction method and / or a dual-band construction method.

[0111] In this embodiment, noise reduction methods such as SG (Savitzky-Golay convolution smoothing), moving average smoothing, variable standardization, multivariate scattering correction, first derivative transformation, and second derivative transformation can be used to reduce the noise of hyperspectral data at each sampling point.

[0112] For example, noise reduction can be achieved using a moving average smoothing method, as detailed in Formula (I) below.

[0113]

[0114] Where i is used to identify the i-th wavelength segment, x i,MAS x is the moving average smoothed value of the sampling point in the i-th wavelength segment. i Used to characterize the spectral value of the sampling point at the i-th wavelength, where 2w+1 is the window number.

[0115] For example, noise reduction is performed using the SG (Savitzky-Golay convolution smoothing method), as detailed in Formula (II) below.

[0116]

[0117] in, Used to characterize the coefficients of the polynomial obtained by fitting using the least squares method. The mean obtained by the SG convolution smoothing method is used to characterize the mean, j represents any value in [-w, w], and w represents the sliding window.

[0118] For example, noise reduction can be achieved using variable standardization, as detailed in Formula (III) below.

[0119]

[0120] Where x is used to characterize the original spectrum corresponding to the sampling point. Used to characterize the average spectral reflectance of each band, where p is the wavelength point of the spectral curve, and x... SNV This is used to characterize the noise reduction value of the spectrum after standardization and averaging through variables, where n is the total number of spectral bands and i is the i-th band among the n spectral bands.

[0121] For example, noise reduction is achieved using multivariate scattering correction, as detailed in formulas (iv) to (vi) below.

[0122]

[0123]

[0124]

[0125] Formula (iv) is used to calculate the average spectrum. Formula (5) is a linear regression equation, and Formula (6) is used to obtain the spectrum to be corrected. The average spectrum is used to characterize the spectrum, n is used to characterize the sample size, pi and bi can be obtained by linear regression fitting, and X i(MSC)Used to characterize spectral values ​​after multivariate scattering correction.

[0126] The first-order derivative transformation and the second-order derivative transformation described above can be implemented using the finite difference method, which will not be elaborated further in the embodiments of this application.

[0127] After denoising the sample spectral data, denoised sample spectral data can be obtained. Based on the sample spectral data and / or the denoised sample spectral data, the sample spectral index corresponding to each sampling point can be constructed using a single-band construction method and / or a dual-band construction method. The specific single-band construction method and dual-band construction method can be referred to the relevant descriptions in the foregoing embodiments, and will not be repeated here in the embodiments of this application.

[0128] By performing single-band transformations or pairwise band combinations on sample spectral data and / or denoised sample spectral data through mathematical transformations and combinations, single-band and dual-band sample spectral indices can be formed, which can greatly expand spectral information and enhance the ability of spectral data to identify ground objects, thereby improving the ability to predict soil arsenic concentration based on spectral data.

[0129] In one embodiment, refer to Figure 6 As shown, in step 306, for any sampling point, determining the hyperspectral index of the key features of the sample from the sample spectral index corresponding to the sampling point may include:

[0130] Step 602: For any sampling point, determine the correlation coefficient between the spectral index of each sample and the arsenic concentration in the soil at each sampling point, based on the spectral index of the sample corresponding to the sampling point and the arsenic concentration in the soil at each sampling point.

[0131] Step 604: The spectral index of the sample whose absolute value of the correlation coefficient is greater than or equal to the correlation coefficient threshold is taken as the hyperspectral index of the key feature of the sample.

[0132] In this embodiment of the application, for any sampling point, the correlation coefficient between the spectral index of each sample corresponding to the sampling point and the soil arsenic concentration can be determined by the Pearson coefficient. Based on the correlation coefficient between the spectral index of each sample and the soil arsenic concentration, the hyperspectral index of the key feature of the sample can be determined from the spectral index of the sample. For example, the spectral index of the sample with an absolute value of the correlation coefficient greater than or equal to the correlation coefficient threshold can be used as the hyperspectral index of the key feature of the sample. The correlation coefficient threshold is a preset value, and the specific value can be determined according to the prediction accuracy requirements. For example, the correlation coefficient threshold can be preset to 0.15.

[0133] For example, the correlation coefficient between the spectral index of each sample and the arsenic concentration in the soil at each sampling point can be determined based on the spectral index of the sample corresponding to the sampling point and the arsenic concentration in the soil at each sampling point, as shown in Formula (VII).

[0134]

[0135] Where r is used to characterize the correlation coefficient between the sample spectral index and the soil arsenic concentration, m is the total number of sampling points, q is used to characterize the q-th sampling point, and x q Let be the sample spectral index of the q-th sampling point. y is used to characterize the mean spectral index of a sample at each sampling point. q Let x be the soil arsenic concentration at the q-th sampling point. q Used to characterize the average arsenic concentration in the soil at each sampling point.

[0136] Based on the spatial distribution inversion method of soil arsenic concentration provided in this application, the key feature hyperspectral index of the sample that is significantly correlated with the soil arsenic concentration can be selected from the constructed sample spectral indices to build a prediction model. This can reduce the amount of computation, save computation time, and improve the prediction accuracy of the prediction model.

[0137] In one embodiment, the above method may further include:

[0138] Determine the target spectral bands and target construction methods used to construct the hyperspectral index of key sample features;

[0139] In this embodiment of the application, step 104, which involves constructing a key feature hyperspectral index for each pixel based on the spectral data corresponding to each pixel in the satellite hyperspectral image, may specifically include:

[0140] For any pixel in a satellite hyperspectral image, a hyperspectral index of key features of the pixel is constructed based on the target spectral band in the spectral data corresponding to the pixel and the target construction method.

[0141] In this embodiment of the application, during the training phase of the prediction model, when determining the hyperspectral index of key features of a sample from the sample spectral indices, the spectral band used to determine the hyperspectral index of key features of the sample and the construction method used to construct the hyperspectral index of key features of the sample can be respectively used as the target spectral band and the target construction method. Then, for any pixel in the satellite hyperspectral image, the target construction method can be used to process the target spectral band in the spectral data corresponding to that pixel to construct the hyperspectral index of key features of that pixel.

[0142] For example, in the construction stage of the prediction model, if the hyperspectral index of the key feature of the sample is the ratio of band 1 to band 3, then band 1 and band 3 can be determined as the target spectral bands, and construction method a / b can be determined as the target construction method. Then, for any pixel in the satellite hyperspectral image, the key feature hyperspectral index of the pixel can be obtained by calculating the ratio of band 1 to band 3 of the pixel.

[0143] To enable those skilled in the art to better understand the embodiments of this application, the embodiments of this application are described below through specific examples.

[0144] Reference Figure 7 As shown, during the network training phase, sampling points can be set up within the site using a systematic sampling method, in accordance with the "Technical Guidelines for Contaminated Site Investigation" (HJ25.1-2014) and the "Technical Guidelines for Site Environmental Monitoring" (HJ 25.2-2014). After the samples are brought back to the laboratory and pretreated, the arsenic concentration in the soil at each sampling point is determined by inductively coupled plasma mass spectrometry.

[0145] Obtain hyperspectral images of the samples. Hyperspectral images of the study area near the sampling time can be found on the Natural Resources Remote Sensing Satellite Cloud Service Platform. Select "Gaofen-5" images with zero or minimal cloud cover in the study area and download them for later use.

[0146] The downloaded hyperspectral images were corrected to obtain sample hyperspectral images. After geometric correction, radiometric correction was performed on the hyperspectral images. Radiometric calibration of the hyperspectral images was performed, converting the pixel units to μw / cm²·sr·nm. Then, FLAASH atmospheric correction was performed. Based on the study area and the acquisition time of the satellite imagery, "Sub-ArcticSummer" was selected as the aerosol model. The pixel spectral curves before atmospheric correction can be referenced. Figure 8a As shown, the atmospherically corrected pixel spectral curves can be referenced. Figure 8b As shown.

[0147] After correction, noise reduction can be performed on the hyperspectral data corresponding to the sampling points. The sample spectral data of the sampling points can be extracted from the sample hyperspectral image using Python's GDAL package, and noise reduction can be performed on the sample hyperspectral data of each sampling point using methods such as SG convolution smoothing, moving average smoothing, variable standardization, multivariate scattering correction, first derivative transformation, and second derivative transformation.

[0148] For the denoised sample hyperspectral data, single-band and dual-band spectral data construction methods can be used to construct the sample spectral indices corresponding to the sampling points. For example, the single-band and dual-band construction methods mentioned in the foregoing embodiments can be used to construct 19.45×102 spectral indices based on the 305 denoised spectral bands. 4 Spectral index of each sample.

[0149] Key hyperspectral indices can be selected from multiple sample spectral indices. Correlation analysis is used to determine the response relationship between each sample spectral index and soil arsenic concentration, and sample spectral indices with absolute correlation coefficients greater than the correlation coefficient threshold are selected as key hyperspectral indices.

[0150] A random forest-based inversion model is constructed. The sampling points obtained in the preceding process are randomly divided into training and test sets. One of the most important hyperparameters in random forest is the number of decision trees. This parameter can be optimized using cross-validation on the training set. The optimal number of decision trees is input into the random forest model, while the remaining parameters are set to default. The random forest model is then trained on the training set. The test set is input into the random forest model to predict the predicted arsenic concentration value for each point. The correlation coefficient R and mean squared error MSE between the predicted arsenic concentration values ​​and the actual arsenic concentration values ​​at the sampling points can be used to evaluate the prediction results of the random forest model. The calculation methods for both are as follows.

[0151] The correlation coefficient R between the predicted arsenic concentration and the actual arsenic concentration can be found in the following formula (VIII).

[0152]

[0153] The mean square error can be calculated using the following formula (IX).

[0154]

[0155] In the formula, m is the number of sampling points in the test set, and y q,pre Output the predicted arsenic concentration value for the q-th sampling point for the random forest model. To predict the average value of arsenic concentration, y q This represents the true arsenic concentration value at the q-th sampling point. This represents the average actual arsenic concentration.

[0156] Assuming the random forest model has an R-value of 0.73 and an MSE of 19671.51, meeting the training requirements, a trained prediction model is obtained. Based on this prediction model and the image generation module, an inversion model can be constructed. This inversion model can predict the spatial distribution of soil arsenic concentration in satellite hyperspectral images. For example, the comparison results of the random forest model with Kriging interpolation under different sampling units are shown below. Figure 9a and Figure 9b As shown ( Figure 9a and Figure 9b (Left side: Random Forest; Right side: Kriging Interference) Figure 9a The comparison results for R are as follows. Figure 9b The comparison results are for MSE.

[0157] During the inference phase, the spatial distribution of soil arsenic concentration can be retrieved from satellite hyperspectral images using an inversion model. Spectral data for each pixel in the satellite hyperspectral image can be extracted. After denoising the spectral data for each pixel, a key feature hyperspectral index is constructed for each pixel. This key feature hyperspectral index is then input into the inversion model to output the retrieved soil arsenic concentration distribution image corresponding to the satellite hyperspectral image.

[0158] The soil arsenic concentration spatial distribution inversion method provided in this application directly inverts the soil arsenic concentration distribution using satellite hyperspectral images. It expands the spectral data and improves the correlation between satellite hyperspectral images and soil arsenic concentration by employing single-band and dual-band hyperspectral index construction methods. Through correlation analysis, spectral indices with high responsiveness to soil concentration are selected as key feature variables for modeling. The quantitative relationship between spectral data and soil arsenic concentration is analyzed using random forest. This method eliminates the need for laboratory spectral acquisition of soil samples, making it simpler and more economical. Mathematical transformations and combinations enhance the correlation between spectral data and soil arsenic concentration, optimizing model accuracy from the source. The wide coverage and rich temporal data of satellite hyperspectral images facilitate large-area, multi-temporal soil arsenic concentration inversion.

[0159] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0160] Based on the same inventive concept, this application also provides a soil arsenic concentration spatial distribution inversion device for implementing the above-mentioned soil arsenic concentration spatial distribution inversion method. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more soil arsenic concentration spatial distribution inversion device embodiments provided below can be found in the limitations of the soil arsenic concentration spatial distribution inversion method above, and will not be repeated here.

[0161] In one embodiment, such as Figure 10As shown, a soil arsenic concentration spatial distribution inversion method device 1000 is provided, including: a first acquisition module 1002, a first construction module 1004, and an inversion module 1006, wherein:

[0162] The first acquisition module 1002 is used to acquire satellite hyperspectral images;

[0163] The first construction module 1004 is used to construct the key feature hyperspectral index of each pixel based on the spectral data corresponding to each pixel in the satellite hyperspectral image;

[0164] The inversion module 1006 is used to perform inversion processing on the key feature hyperspectral index of each pixel through an inversion model to obtain the spatial distribution inversion image of soil arsenic concentration corresponding to the satellite hyperspectral image. The pixel value corresponding to any pixel in the spatial distribution inversion image of soil arsenic concentration is used to characterize the arsenic concentration of the soil corresponding to the pixel.

[0165] The aforementioned soil arsenic concentration spatial distribution inversion device, after acquiring satellite hyperspectral images, constructs key feature hyperspectral indices for each pixel based on the spectral data corresponding to each pixel in the satellite hyperspectral images. It then uses an inversion model to invert these key feature hyperspectral indices, obtaining a soil arsenic concentration spatial distribution inversion image corresponding to the satellite hyperspectral image. The pixel value corresponding to any pixel in the soil arsenic concentration spatial distribution inversion image is used to characterize the arsenic concentration in the soil corresponding to that pixel. According to the soil arsenic concentration spatial distribution inversion device provided in this application, the inversion model can invert real-time acquired satellite hyperspectral images to obtain a soil arsenic concentration spatial distribution inversion image in real time. This reduces manual intervention, thus lowering the consumption of manpower, resources, and time, reducing the complexity of the detection process, and improving the efficiency and accuracy of predicting arsenic concentration in soil.

[0166] In one embodiment, the inversion model includes a prediction model and an image generation module, wherein the inversion module 1006 is further configured to:

[0167] The hyperspectral index of the key feature of each pixel is predicted using the prediction model to obtain the arsenic concentration corresponding to each pixel.

[0168] The image generation module inverts the arsenic concentration corresponding to each pixel to obtain the spatial distribution inversion image of soil arsenic concentration corresponding to the satellite hyperspectral image.

[0169] In one embodiment, the above-described apparatus further includes:

[0170] The second acquisition module is used to acquire sample satellite hyperspectral images of the study area and the arsenic concentration of soil at each sampling point in the sample satellite hyperspectral images;

[0171] The second construction module is used to construct the sample spectral index corresponding to each sampling point based on the sample spectral data corresponding to each sampling point in the sample satellite hyperspectral image;

[0172] The first determining module is used to determine the hyperspectral index of key features of a sample from the sample spectral index corresponding to any of the sampling points;

[0173] The prediction module is used to predict the hyperspectral index of the key feature of the sample corresponding to the sampling point through an initial prediction model, so as to obtain the predicted arsenic concentration corresponding to the sampling point.

[0174] The training module is used to train the initial prediction model based on the predicted arsenic concentration corresponding to the sampling point and the arsenic concentration corresponding to the soil at the sampling point, so as to obtain the prediction model.

[0175] In one embodiment, the second acquisition module is further configured to:

[0176] Acquire satellite hyperspectral images of the study area;

[0177] The hyperspectral image of the satellite is corrected to obtain a sample satellite hyperspectral image.

[0178] In one embodiment, the second building module is further configured to:

[0179] The sample spectral data corresponding to each sampling point in the sample satellite hyperspectral image are subjected to noise reduction processing to obtain the noise-reduced sample spectral data.

[0180] Based on the sample spectral data and / or the denoised sample spectral data, the sample spectral index corresponding to each sampling point is constructed using a single-band construction method and / or a dual-band construction method.

[0181] In one embodiment, the first determining module is further configured to:

[0182] For any of the sampling points, the correlation coefficient between the spectral index of the sample corresponding to each sampling point and the arsenic concentration of the soil at each sampling point is determined based on the spectral index of the sample corresponding to each sampling point and the arsenic concentration of the soil at each sampling point.

[0183] The spectral index of the sample whose absolute value of the correlation coefficient is greater than or equal to the correlation coefficient threshold is taken as the hyperspectral index of the key feature of the sample.

[0184] In one embodiment, the device further includes:

[0185] The second determining module is used to determine the target spectral band and target construction method for constructing the hyperspectral index of the key features of the sample.

[0186] The first construction module 1004 is further configured to:

[0187] For any pixel in the satellite hyperspectral image, a key feature hyperspectral index is constructed based on the target spectral band in the spectral data corresponding to the pixel and the target construction method.

[0188] Each module in the aforementioned soil arsenic concentration spatial distribution inversion device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0189] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 11 As shown, the computer device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When executed by the processor, the computer program implements a method for inverting the spatial distribution of soil arsenic concentration. The display screen can be an LCD screen or an e-ink display screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0190] Those skilled in the art will understand that Figure 11 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0191] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0192] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0193] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0194] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0195] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0196] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0197] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for inverting the spatial distribution of soil arsenic concentration, characterized in that, The method includes: Acquire satellite hyperspectral images; The target spectral band and target construction method for constructing the key feature hyperspectral index of the sample are determined. For any pixel in the satellite hyperspectral image, the key feature hyperspectral index of the pixel is constructed according to the target spectral band and the target construction method in the spectral data corresponding to the pixel. The target construction method includes single-band and / or dual-band construction methods. The key feature hyperspectral index of each pixel is inverted using an inversion model to obtain a spatial distribution inversion image of soil arsenic concentration corresponding to the satellite hyperspectral image. The pixel value corresponding to any pixel in the spatial distribution inversion image of soil arsenic concentration is used to characterize the arsenic concentration of the soil corresponding to that pixel.

2. The method according to claim 1, characterized in that, The inversion model includes a prediction model and an image generation module. The process of inverting the key feature hyperspectral index of each pixel using the inversion model to obtain the spatial distribution inversion image of soil arsenic concentration corresponding to the satellite hyperspectral image includes: The hyperspectral index of the key feature of each pixel is predicted using the prediction model to obtain the arsenic concentration corresponding to each pixel. The image generation module inverts the arsenic concentration corresponding to each pixel to obtain the spatial distribution inversion image of soil arsenic concentration corresponding to the satellite hyperspectral image.

3. The method according to claim 2, characterized in that, The method further includes: Obtain sample satellite hyperspectral images of the study area and the arsenic concentration of the soil at each sampling point in the sample satellite hyperspectral images; Based on the sample spectral data corresponding to each sampling point in the sample satellite hyperspectral image, construct the sample spectral index corresponding to each sampling point; For any of the sampling points, determine the hyperspectral index of the key features of the sample from the sample spectral index corresponding to the sampling point; The hyperspectral index of the key feature of the sample corresponding to the sampling point is predicted by the initial prediction model to obtain the predicted arsenic concentration corresponding to the sampling point. Based on the predicted arsenic concentration corresponding to the sampling point and the arsenic concentration corresponding to the soil at the sampling point, the initial prediction model is constructed, and the prediction model is obtained.

4. The method according to claim 3, characterized in that, The acquisition of sample satellite hyperspectral images of the study area includes: Acquire satellite hyperspectral images of the study area; The hyperspectral image of the satellite is corrected to obtain a sample satellite hyperspectral image.

5. The method according to claim 3 or 4, characterized in that, The step of constructing a sample spectral index corresponding to each sampling point based on the sample spectral data corresponding to each sampling point in the sample satellite hyperspectral image includes: The sample spectral data corresponding to each sampling point in the sample satellite hyperspectral image are subjected to noise reduction processing to obtain the noise-reduced sample spectral data. Based on the sample spectral data and / or the denoised sample spectral data, the sample spectral index corresponding to each sampling point is constructed using a single-band construction method and / or a dual-band construction method.

6. The method according to claim 5, characterized in that, For any of the sampling points, determining the key hyperspectral index of the sample from the sample spectral index corresponding to the sampling point includes: For any of the sampling points, the correlation coefficient between the spectral index of the sample corresponding to each sampling point and the arsenic concentration of the soil at each sampling point is determined based on the spectral index of the sample corresponding to each sampling point and the arsenic concentration of the soil at each sampling point. The spectral index of the sample whose absolute value of the correlation coefficient is greater than or equal to the correlation coefficient threshold is taken as the hyperspectral index of the key feature of the sample.

7. The method according to claim 1, characterized in that, The key feature hyperspectral data constructed using the single-band construction method include: ln(a), 1 / a, 1 / ln(a), ln(1 / a); the key feature hyperspectral data constructed using the dual-band construction method include: a×b, a / b, , , , , where a and b are used to characterize any band in the spectral data.

8. A soil arsenic concentration spatial distribution inversion device, characterized in that, The device includes: The first acquisition module is used to acquire satellite hyperspectral images; The first construction module is used to determine the target spectral band and target construction method for constructing the key feature hyperspectral index of the sample. For any pixel in the satellite hyperspectral image, the key feature hyperspectral index of the pixel is constructed according to the target spectral band and the target construction method in the spectral data corresponding to the pixel. The target construction method includes single-band and / or dual-band construction methods. The inversion module is used to invert the key feature hyperspectral index of each pixel through an inversion model to obtain the spatial distribution inversion image of soil arsenic concentration corresponding to the satellite hyperspectral image. The pixel value corresponding to any pixel in the spatial distribution inversion image of soil arsenic concentration is used to characterize the arsenic concentration of the soil corresponding to the pixel.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images