Key area ecological comprehensive assessment method based on natural resource survey monitoring data

Abstract

The invention relates to the technical field of ecological comprehensive evaluation, in particular to a key area ecological comprehensive evaluation method based on natural resource survey monitoring data, which comprises the following steps: performing timing remote sensing image acquisition on a target key area according to a fixed camera position to obtain a plurality of remote sensing images at different time, sorting the plurality of remote sensing images at different times according to a time sequence to obtain a remote sensing image sequence; and performing image enhancement on the remote sensing image sequence of the target key area to obtain a standard remote sensing image sequence, and outputting a target water body image sequence corresponding to the standard remote sensing image sequence through the trained image segmentation model. According to the method, through timed remote sensing image acquisition and time sequence analysis, dynamic ecological change information of a target area can be provided, timeliness and accuracy of an evaluation result are ensured, and through image enhancement and application of a trained model, water body change can be accurately identified, so that detailed water body ecological evaluation is carried out.

Description

Technical Field

[0001] The present invention relates to the technical field of ecological comprehensive assessment, and in particular to a key area ecological comprehensive assessment method based on natural resource survey and monitoring data. Background Art

[0002] Natural resources are also called natural resources. Natural resources are the general term for natural materials and energy that have been discovered by humans, input into the production process, or directly entered into the consumption process, becoming useful or able to give people a sense of comfort, thereby generating economic value to improve human current and future welfare. Natural resource surveys and monitoring include regular surveys, on-site monitoring, remote sensing surveys and monitoring, and statistical analysis. Monitoring of key water source areas is a very important working method.

[0003] Traditional methods usually rely on manual sampling and field surveys, which are cumbersome and time-consuming, and it is difficult to cover large areas. Field data collection and analysis often take a long time, resulting in ecological assessment results that are not timely enough to reflect rapidly changing environmental conditions. Traditional methods often require more resource inputs, including manpower, equipment, and materials, which makes the overall cost higher and is not conducive to large-scale or long-term ecological monitoring projects. Traditional data analysis usually relies on manual calculations and statistics, which makes it difficult to handle large-scale and complex data sets, and has lower efficiency and accuracy than modern data analysis techniques. Summary of the invention

[0004] The purpose of the present invention is to address the problems existing in the background technology and to propose a comprehensive ecological assessment method for key areas based on natural resource survey and monitoring data.

[0005] The technical solution of the present invention: a comprehensive ecological assessment method for key areas based on natural resource survey and monitoring data, comprising:

[0006] Performing timed remote sensing image acquisition of a target key area according to a fixed camera position to obtain a plurality of remote sensing images at different times, and sorting the plurality of remote sensing images at different times according to a time series to obtain a remote sensing image sequence;

[0007] Performing image enhancement on the remote sensing image sequence of the target key area to obtain a standard remote sensing image sequence, inputting the standard remote sensing image sequence into a trained image segmentation model, and outputting a target water body image sequence corresponding to the standard remote sensing image sequence through the trained image segmentation model;

[0008] Performing water body pixel number statistics on the target water body image sequence to obtain a water body pixel number sequence corresponding to the target water body image sequence, and performing a water body ecological assessment on the target key area based on the water body pixel number to obtain a water body ecological coefficient of the target key area;

[0009] Manually sampling multiple sampling points in the target key area to obtain soil pollution content at the multiple sampling points, and conducting soil ecological assessment on the target key area based on the soil pollution content at the multiple sampling points to obtain a soil ecological coefficient of the target key area;

[0010] Based on the remote sensing sensor, spectral information of the target key area is collected to obtain spectral information of the target key area, and based on the spectral information of the target key area, a vegetation coverage ecological assessment is performed on the target key area to obtain a vegetation coverage ecological coefficient of the target key area;

[0011] The comprehensive ecological coefficient of the target key area is obtained based on the water ecological coefficient, soil ecological coefficient and vegetation cover ecological coefficient of the target key area.

[0012] Preferably, performing image enhancement on the remote sensing image sequence of the target key area to obtain a standard remote sensing image sequence includes:

[0013] The remote sensing image is expressed as a color three-dimensional vector form, and the RGB space pixel ratio is obtained based on the color three-dimensional vector of the remote sensing image. The remote sensing image expression formula is as follows:

[0014] S(x,y)={R i ,G i ,B i};

[0015] Among them, S(x,y) represents the remote sensing image, {R i ,G i ,B i} represents the color three-dimensional vector of the i-th pixel of the remote sensing image;

[0016] The RGB space pixel ratio calculation formula is as follows:

[0017]

[0018] Among them, α represents the pixel ratio in RGB space;

[0019] A brightness gain surface is determined based on the RGB spatial pixel ratio of the remote sensing image, and the brightness of the remote sensing image is separated based on the brightness gain surface. The brightness separation formula is as follows:

[0020] B(x,y)=α(x,y)max(R(x,y),G(x,y),B(x,y));

[0021] Among them, B(x,y) represents the brightness of the remote sensing image, α(x,y) represents the brightness gain surface, and max(R(x,y),G(x,y),B(x,y)) represents the grayscale degree of the remote sensing image.

[0022] Preferably, performing image enhancement on the remote sensing image sequence of the target key area to obtain a standard remote sensing image sequence also includes:

[0023] The brightness of the separated remote sensing image is enhanced based on the Retinex algorithm, and the brightness enhancement formula is as follows:

[0024] EB(x,y,ε)=lgB(x,y)-lg[F(x,y,ε)*B(x,y)];

[0025] Among them, EB(x,y,ε) represents the enhanced brightness, ε represents the pixel scale, and F(x,y,ε) represents the center surrounding function of the remote sensing image by the Retinex algorithm;

[0026] The contrast of the remote sensing image is enhanced based on the enhanced brightness of the remote sensing image. The contrast enhancement formula is as follows:

[0027]

[0028] Among them, BL(x,y,ε) represents the enhanced contrast, EB avg (x, y, ε) represents the mean value of the enhanced brightness.

[0029] Preferably, the standard remote sensing image sequence is input into a trained image segmentation model, and the target water body image sequence corresponding to the standard remote sensing image sequence is output through the trained image segmentation model, including:

[0030] Inputting the standard remote sensing image into the image segmentation model;

[0031] For the pixel k of the input standard remote sensing image, its pixel value is x k , the corresponding true category label value is The pixel value x of the input standard remote sensing image k , forward propagated through the ISUNet network to obtain the C-dimensional output feature vector Wherein, C represents the total number of categories in the standard remote sensing image, and c represents the cth category in the total number of categories in the standard remote sensing image;

[0032] Based on the C-dimensional output feature vector Get the predicted category label of pixel k of the standard remote sensing image belonging to the cth category The probability calculation formula for the predicted category label of the pixel k of the standard remote sensing image belonging to the cth category is as follows:

[0033]

[0034] Among them, p c (x k ) represents the probability that pixel k of the standard remote sensing image belongs to the predicted category label of category c.

[0035] Preferably, the standard remote sensing image sequence is input into a trained image segmentation model, and the target water body image sequence corresponding to the standard remote sensing image sequence is output through the trained image segmentation model, further comprising:

[0036] Based on the probability that the pixel k of the standard remote sensing image belongs to the predicted category label of the cth category, the loss between the true value and the predicted value is calculated through the loss function, and it is determined whether the loss reaches the preset loss threshold. If it does not reach the preset loss threshold, the error is transmitted through the back propagation algorithm, and the Adam optimization algorithm is used to train the image segmentation model and update the network parameters. Otherwise, the image segmentation model training process is terminated. The loss function is as follows:

[0037]

[0038] Among them, Loss represents the loss between the true value and the predicted value, and N represents the total number of pixels in the standard remote sensing image;

[0039] After the image segmentation model training is completed, the standard remote sensing image is segmented, and the category corresponding to the maximum probability is obtained using the argmax function. The corresponding pixels are segmented based on the category, and the above operation is repeated until each pixel in the standard remote sensing image is traversed to obtain the target water body image. The calculation formula for the category corresponding to the maximum probability using the argmax function is as follows:

[0040] C k = rgmax(p c (x k ));

[0041] Among them, C k Represents the kth category among the total number of categories in the standard remote sensing image.

[0042] Preferably, performing a water body ecological assessment on the target key area based on the number of water body pixels to obtain a water body ecological coefficient of the target key area includes:

[0043] Acquire a water body pixel area sequence based on the water body pixel quantity sequence;

[0044] Fitting the water body pixel area sequence into a water body pixel area curve;

[0045] Comparing the water body pixel area curve with a preset standard water body pixel area curve to obtain an error between the two;

[0046] The error is compared with a preset error threshold to obtain the water ecological coefficient of the target key area.

[0047] Preferably, the soil ecological assessment of the target key area is performed based on the soil pollution content of the multiple sampling points to obtain the soil ecological coefficient of the target key area, including:

[0048] The pollution index of each pollution factor in the target key area is calculated, and the pollution index calculation formula of the pollution factor is as follows:

[0049]

[0050] Among them, P i represents the pollution index of the ith pollution factor in the target key area, U i represents the measured value of the i-th pollution factor in the target key area, S i represents the evaluation standard value of the i-th pollution factor in the target key area;

[0051] The comprehensive pollution index of each pollution factor in the target key area is calculated based on the maximum value and average value of the pollution index of each pollution factor in the target key area. The calculation formula of the comprehensive pollution index of the pollution factor is as follows:

[0052]

[0053] in, represents the comprehensive pollution index of the ith pollution factor at the nth sampling point in the target key area, It represents the maximum value of the pollution index of the ith pollution factor at the nth sampling point in the target key area. It represents the average value of the pollution index of the ith pollution factor at the nth sampling point in the target key area.

[0054] Preferably, the soil ecological assessment of the target key area is performed based on the soil pollution content of the multiple sampling points to obtain the soil ecological coefficient of the target key area, and further includes:

[0055] The soil ecological coefficient in the target key area is calculated based on the toxicity coefficient of each pollution factor and the comprehensive pollution index of each pollution factor in the target key area. The soil ecological coefficient calculation formula is as follows:

[0056]

[0057] Among them, RI represents the soil ecological coefficient in the target key area, T i represents the toxicity coefficient of the i-th pollution factor, and M represents the total number of sampling points in the target key area.

[0058] Preferably, performing a vegetation coverage ecological assessment on the target key area based on the spectral information of the target key area to obtain a vegetation coverage ecological coefficient of the target key area includes:

[0059] A pixel binary model is constructed, and the pixel binary model is as follows:

[0060] S=S V +S S ;

[0061] Where S represents the spectral information observed by the remote sensing sensor, S V Represents the spectral information of the green vegetation part, S S represents the spectral information of the part without vegetation coverage, and S V =S veg *f c , S veg represents the spectral information of pure pixels covered by vegetation, f c represents vegetation coverage, and S S =S soil *(1-f c ), S soil Represents the spectral information obtained from pure pixels covered entirely by soil;

[0062] The pixel binary model is solved based on the spectral information of the target key area to obtain the vegetation coverage.

[0063] Preferably, the vegetation coverage calculation formula is as follows:

[0064]

[0065] Compared with the prior art, the above technical solution of the present invention has the following beneficial technical effects:

[0066] The present invention can provide dynamic ecological change information of the target area through timed remote sensing image acquisition and time series analysis, ensure the timeliness and accuracy of the evaluation results, and can accurately identify water changes through image enhancement and the application of trained models, so as to carry out detailed water ecological evaluation. At the same time, through soil sampling and spectral information analysis, it can comprehensively evaluate the ecological status of soil and vegetation, and through the comprehensive water ecological coefficient, soil ecological coefficient and vegetation coverage ecological coefficient, it can obtain the overall ecological health status of the target area, help to make comprehensive environmental management and protection decisions, and provide quantitative data support by counting the number of water pixels and soil pollution content, so as to make the ecological evaluation results more objective and scientific. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 A schematic flow chart of the steps of an overall method in an embodiment of the present invention. DETAILED DESCRIPTION

[0068] Embodiment 1, as Figure 1 As shown, the key regional ecological comprehensive assessment method based on natural resource survey and monitoring data proposed in the present invention includes:

[0069] S1. Performing timed remote sensing image acquisition of a target key area according to a fixed camera position to obtain a plurality of remote sensing images at different times, and sorting the plurality of remote sensing images at different times according to a time series to obtain a remote sensing image sequence;

[0070] S2, performing image enhancement on the remote sensing image sequence of the target key area to obtain a standard remote sensing image sequence, inputting the standard remote sensing image sequence into the trained image segmentation model, and outputting the target water body image sequence corresponding to the standard remote sensing image sequence through the trained image segmentation model;

[0071] S3, counting the number of water body pixels of the target water body image sequence to obtain a water body pixel number sequence corresponding to the target water body image sequence, and conducting a water body ecological assessment on the target key area based on the number of water body pixels to obtain a water body ecological coefficient of the target key area;

[0072] S4. Manually sampling multiple sampling points in the target key area to obtain soil pollution content at the multiple sampling points, and conducting soil ecological assessment on the target key area based on the soil pollution content at the multiple sampling points to obtain a soil ecological coefficient for the target key area;

[0073] S5. Collecting spectral information of the target key area based on a remote sensing sensor to obtain spectral information of the target key area, and conducting a vegetation coverage ecological assessment of the target key area based on the spectral information of the target key area to obtain a vegetation coverage ecological coefficient of the target key area;

[0074] S6. Obtain the comprehensive ecological coefficient of the target key area based on the water ecological coefficient, soil ecological coefficient and vegetation cover ecological coefficient of the target key area.

[0075] In the present invention, fixed camera position refers to using remote sensing image acquisition equipment at a fixed position to regularly acquire images of the target area under the same conditions; remote sensing image sequence refers to time series data composed of remote sensing images acquired at different time points. These images can show the changes in the target area over time; the number of water body pixels refers to the total number of pixels identified as water bodies in the output of the image segmentation model. By counting these pixels, the coverage of the water body can be evaluated; manual sampling refers to selecting multiple sampling points on site to collect soil samples for analysis, and these samples are used to detect the content of pollutants in the soil.

[0076] Embodiment 2, the method for comprehensive ecological assessment of key areas based on natural resource survey and monitoring data proposed by the present invention, compared with embodiment 1, this embodiment further includes: performing image enhancement on the remote sensing image sequence of the target key area to obtain a standard remote sensing image sequence, including:

[0077] A1. Express the remote sensing image as a three-dimensional color vector. Obtain the RGB space pixel ratio based on the three-dimensional color vector of the remote sensing image. The remote sensing image expression formula is as follows:

[0078] S(x,y)={R i ,G i ,B i};

[0079] Among them, S(x,y) represents the remote sensing image, {R i ,G i ,B i} represents the color three-dimensional vector of the i-th pixel of the remote sensing image;

[0080] The calculation formula for the RGB space pixel ratio is as follows:

[0081]

[0082] Among them, α represents the pixel ratio in RGB space;

[0083] A2. Determine the brightness gain surface based on the RGB spatial pixel ratio of the remote sensing image, and separate the brightness of the remote sensing image based on the brightness gain surface. The brightness separation formula is as follows:

[0084] B(x,y)=α(x,y)max(R(x,y),G(x,y),B(x,y));

[0085] Among them, B(x,y) represents the brightness of the remote sensing image, α(x,y) represents the brightness gain surface, and max(R(x,y),G(x,y),B(x,y)) represents the grayscale degree of the remote sensing image.

[0086] In an optional embodiment, performing image enhancement on the remote sensing image sequence of the target key area to obtain a standard remote sensing image sequence also includes:

[0087] A3. Based on the Retinex algorithm, the brightness of the separated remote sensing image is enhanced. The brightness enhancement formula is as follows:

[0088] EB(x,y,ε)=lgB(x,y)-lg[F(x,y,ε)*B(x,y)];

[0089] Among them, EB(x,y,ε] represents the enhanced brightness, ε represents the pixel scale, and F(x,y,ε) represents the center surrounding function of the remote sensing image by the Retinex algorithm;

[0090] A4. The contrast of the remote sensing image is enhanced based on the brightness of the enhanced remote sensing image. The contrast enhancement formula is as follows:

[0091]

[0092] Among them, BL(x,y,ε) represents the enhanced contrast, EB avg (x, y, ε) represents the mean value of the enhanced brightness.

[0093] It should be noted that the three-dimensional color vector represents the red, green and blue components of the pixel in the image; the brightness gain surface is a function used to adjust the brightness of the image, which can make the brightness distribution of the image more consistent with the perception of the human eye; the Retinex algorithm is a technology used to enhance the brightness and color contrast of the image, which imitates the perception of the human eye to improve the visual quality of the image; contrast refers to the degree of difference between light and dark in the image, and enhancing the contrast can make the details in the image more obvious.

[0094] In an optional embodiment, the standard remote sensing image sequence is input into a trained image segmentation model, and the trained image segmentation model is used to output a target water body image sequence corresponding to the standard remote sensing image sequence, including:

[0095] B1. Input the standard remote sensing image into the image segmentation model;

[0096] B2. For pixel k of the input standard remote sensing image, its pixel value is x k , the corresponding true category label value is The pixel value x of the input standard remote sensing imagek , forward propagated through the ISUNet network to obtain the C-dimensional output feature vector Where C represents the total number of categories in the standard remote sensing image, and c represents the cth category in the total number of categories in the standard remote sensing image;

[0097] B3. Output feature vector based on C dimension Get the predicted category label of pixel k of the standard remote sensing image belonging to the cth category The probability calculation formula for the predicted category label of pixel k of the standard remote sensing image belonging to the cth category is as follows:

[0098]

[0099] Among them, p c (x k ) represents the probability that pixel k of the standard remote sensing image belongs to the predicted category label of category c.

[0100] It should be noted that ISUNet is a deep learning network for image processing, especially in image enhancement and super-resolution reconstruction tasks.

[0101] In an optional embodiment, the standard remote sensing image sequence is input into a trained image segmentation model, and the target water body image sequence corresponding to the standard remote sensing image sequence is output through the trained image segmentation model, further comprising:

[0102] B4. Based on the probability that pixel k of the standard remote sensing image belongs to the predicted category label of category c, the loss between the true value and the predicted value is calculated through the loss function to determine whether the loss reaches the preset loss threshold. If it does not reach the preset loss threshold, the error is transmitted through the back propagation algorithm, and the Adam optimization algorithm is used to train the image segmentation model and update the network parameters. Otherwise, the image segmentation model training process is terminated. The loss function is as follows:

[0103]

[0104] Among them, loss represents the loss between the true value and the predicted value, and N represents the total number of pixels in the standard remote sensing image;

[0105] B5. After the image segmentation model training is completed, the standard remote sensing image is segmented, and the category corresponding to the maximum probability is obtained using the argmax function. The corresponding pixels are segmented based on the category, and the above operation is repeated until each pixel in the standard remote sensing image is traversed to obtain the target water body image. The calculation formula for the category corresponding to the maximum probability using the argmax function is as follows:

[0106] C k= argmax(p c (x k ));

[0107] Among them, C k Represents the kth category among the total number of categories in the standard remote sensing image.

[0108] It should be noted that the backpropagation algorithm is a technology used to train neural networks, which minimizes the loss function by calculating gradients and updating network parameters; the Adam (Adaptive Moment Estimation) optimization algorithm is a commonly used optimization method that combines the advantages of gradient descent and momentum methods and can adaptively adjust the learning rate.

[0109] In an optional embodiment, a water body ecology assessment is performed on the target key area based on the number of water body pixels to obtain a water body ecology coefficient of the target key area, including:

[0110] C1. Obtaining a water body pixel area sequence based on a water body pixel quantity sequence;

[0111] C2, fitting the water body pixel area sequence into a water body pixel area curve;

[0112] C3, comparing the water body pixel area curve with a preset standard water body pixel area curve to obtain an error between the two;

[0113] C4. Compare the error with the preset error threshold to obtain the water ecological coefficient of the target key area.

[0114] It should be noted that the water body pixel area sequence refers to the actual area corresponding to the number of water body pixels recorded in time series, which can be calculated by the resolution and number of pixels of the image; the water body pixel area curve refers to the curve generated by fitting the water body pixel area sequence data, which is used to represent the trend of water body area change over time; the standard water body pixel area curve refers to the preset ideal water body pixel area curve, which is usually defined based on reference data or theoretical models and is used for comparison with the actual curve; the error refers to the difference between the actual water body pixel area curve and the standard water body pixel area curve, which reflects the deviation between the actual water body change and the expected change; the error threshold refers to the set maximum allowable error value, which is used to judge whether the actual data is within the acceptable range. Exceeding this threshold may mean abnormal ecological conditions; the water body ecological coefficient refers to an indicator obtained by comparing the actual error with the preset error threshold, which indicates the water body ecological status in the target key area. A high ecological coefficient may indicate that the water body is in good health.

[0115] In an optional embodiment, soil ecological assessment is performed on the target key area based on the soil pollution content of multiple sampling points to obtain the soil ecological coefficient of the target key area, including:

[0116] D1. Calculate the pollution index of each pollution factor in the target key area. The calculation formula of the pollution index of the pollution factor is as follows:

[0117]

[0118] Among them, P i represents the pollution index of the ith pollution factor in the target key area, U i represents the measured value of the i-th pollution factor in the target key area, S i represents the evaluation standard value of the i-th pollution factor in the target key area;

[0119] D2. Based on the maximum and average values ​​of the pollution index of each pollution factor in the target key area, the comprehensive pollution index of each pollution factor in the target key area is calculated. The calculation formula of the comprehensive pollution index of the pollution factor is as follows:

[0120]

[0121] in, represents the comprehensive pollution index of the ith pollution factor at the nth sampling point in the target key area, It represents the maximum value of the pollution index of the ith pollution factor at the nth sampling point in the target key area. It represents the average value of the pollution index of the ith pollution factor at the nth sampling point in the target key area.

[0122] In an optional embodiment, the soil ecological assessment of the target key area is performed based on the soil pollution content of the multiple sampling points to obtain the soil ecological coefficient of the target key area, further comprising:

[0123] D3. Calculate the soil ecological coefficient in the target key area based on the toxicity coefficient of each pollution factor and the comprehensive pollution index of each pollution factor in the target key area. The soil ecological coefficient calculation formula is as follows:

[0124]

[0125] Among them, RI represents the soil ecological coefficient in the target key area, T i represents the toxicity coefficient of the i-th pollution factor, and M represents the total number of sampling points in the target key area.

[0126] In an optional embodiment, a vegetation coverage ecological assessment is performed on the target key area based on the spectral information of the target key area to obtain a vegetation coverage ecological coefficient of the target key area, including:

[0127] E1. Construct a pixel binary model, which is as follows:

[0128] S=S V +S S ;

[0129] Where S represents the spectral information observed by the remote sensing sensor, S V Represents the spectral information of the green vegetation part, S S represents the spectral information of the part without vegetation coverage, and S V =S veg *f c , S veg represents the spectral information of pure pixels covered by vegetation, f c represents vegetation coverage, and S S =S soil *(1-f c ), S soil Represents the spectral information obtained from pure pixels covered entirely by soil;

[0130] E2. Solve the pixel binary model based on the spectral information of the target key area to obtain vegetation coverage.

[0131] It should be noted that the pixel binary model is a method used in remote sensing image analysis, especially in land cover classification and change detection. This model is used to divide each pixel in the remote sensing image into two categories: target (object of interest) and background (non-target).

[0132] In an optional embodiment, the vegetation coverage calculation formula is as follows:

[0133]

[0134] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto, and various changes can be made within the knowledge scope of technicians in the relevant technical field without departing from the purpose of the present invention.

Claims

1. A comprehensive ecological assessment method for key areas based on natural resource survey and monitoring data, characterized by: include: Performing timed remote sensing image acquisition of a target key area according to a fixed camera position to obtain a plurality of remote sensing images at different times, and sorting the plurality of remote sensing images at different times according to a time series to obtain a remote sensing image sequence; Performing image enhancement on the remote sensing image sequence of the target key area to obtain a standard remote sensing image sequence, inputting the standard remote sensing image sequence into a trained image segmentation model, and outputting a target water body image sequence corresponding to the standard remote sensing image sequence through the trained image segmentation model; Performing water body pixel number statistics on the target water body image sequence to obtain a water body pixel number sequence corresponding to the target water body image sequence, and performing a water body ecological assessment on the target key area based on the water body pixel number to obtain a water body ecological coefficient of the target key area; Manually sampling multiple sampling points in the target key area to obtain soil pollution content at the multiple sampling points, and conducting soil ecological assessment on the target key area based on the soil pollution content at the multiple sampling points to obtain a soil ecological coefficient of the target key area; Based on the remote sensing sensor, spectral information of the target key area is collected to obtain spectral information of the target key area, and based on the spectral information of the target key area, a vegetation coverage ecological assessment is performed on the target key area to obtain a vegetation coverage ecological coefficient of the target key area; The comprehensive ecological coefficient of the target key area is obtained based on the water ecological coefficient, soil ecological coefficient and vegetation cover ecological coefficient of the target key area.

2. The method for comprehensive ecological assessment of key areas based on natural resource survey and monitoring data according to claim 1 is characterized by: Performing image enhancement on the remote sensing image sequence of the target key area to obtain a standard remote sensing image sequence, including: The remote sensing image is expressed as a color three-dimensional vector form, and the RGB space pixel ratio is obtained based on the color three-dimensional vector of the remote sensing image. The remote sensing image expression formula is as follows: S(x,y)={R i ,G i ,B i }; Among them, S(x,y) represents the remote sensing image, {R i ,G i ,B i } represents the color three-dimensional vector of the i-th pixel of the remote sensing image; The RGB space pixel ratio calculation formula is as follows: Among them, α represents the pixel ratio in RGB space; A brightness gain surface is determined based on the RGB spatial pixel ratio of the remote sensing image, and the brightness of the remote sensing image is separated based on the brightness gain surface. The brightness separation formula is as follows: B(x,y)=α(x,y)max(R(x,y),G(x,y),B(x,y)); Among them, B(x,y) represents the brightness of the remote sensing image, α(x,y) represents the brightness gain surface, and max(R(x,y),G(x,y),B(x,y)) represents the grayscale degree of the remote sensing image.

3. The method for comprehensive ecological assessment of key areas based on natural resource survey and monitoring data according to claim 2 is characterized by: Performing image enhancement on the remote sensing image sequence of the target key area to obtain a standard remote sensing image sequence, further comprising: The brightness of the separated remote sensing image is enhanced based on the Retinex algorithm, and the brightness enhancement formula is as follows: EB(x,y,ε)=lgB(x,y)-lg[F(x,y,ε)*B(x,y)]; Among them, EB(x,y,ε) represents the enhanced brightness, ε represents the pixel scale, and F(x,y,ε) represents the center surrounding function of the remote sensing image by the Retinex algorithm; The contrast of the remote sensing image is enhanced based on the enhanced brightness of the remote sensing image. The contrast enhancement formula is as follows: Among them, BL(x,y,ε) represents the enhanced contrast, EB avg (x, y, ε) represents the mean value of the enhanced brightness.

4. The method for comprehensive ecological assessment of key areas based on natural resource survey and monitoring data according to claim 1 is characterized by: Inputting the standard remote sensing image sequence into a trained image segmentation model, and outputting a target water body image sequence corresponding to the standard remote sensing image sequence through the trained image segmentation model, including: Inputting the standard remote sensing image into the image segmentation model; For the pixel k of the input standard remote sensing image, its pixel value is x k , the corresponding true category label value is The pixel value x of the input standard remote sensing image k , forward propagated through the ISUNet network to obtain the C-dimensional output feature vector Wherein, C represents the total number of categories in the standard remote sensing image, and c represents the cth category in the total number of categories in the standard remote sensing image; Based on the C-dimensional output feature vector Get the predicted category label of pixel k of the standard remote sensing image belonging to the cth category The probability calculation formula for the predicted category label of the pixel k of the standard remote sensing image belonging to the cth category is as follows: Among them, p c (x k ) represents the probability that pixel k of the standard remote sensing image belongs to the predicted category label of category c.

5. The method for comprehensive ecological assessment of key areas based on natural resource survey and monitoring data according to claim 4 is characterized by: Inputting the standard remote sensing image sequence into a trained image segmentation model, and outputting a target water body image sequence corresponding to the standard remote sensing image sequence through the trained image segmentation model, further comprising: Based on the probability that the pixel k of the standard remote sensing image belongs to the predicted category label of the cth category, the loss between the true value and the predicted value is calculated through the loss function, and it is determined whether the loss reaches the preset loss threshold. If it does not reach the preset loss threshold, the error is transmitted through the back propagation algorithm, and the Adam optimization algorithm is used to train the image segmentation model and update the network parameters. Otherwise, the image segmentation model training process is terminated. The loss function is as follows: Among them, Loss represents the loss between the true value and the predicted value, and N represents the total number of pixels in the standard remote sensing image; After the image segmentation model training is completed, the standard remote sensing image is segmented, and the category corresponding to the maximum probability is obtained using the argmax function. The corresponding pixels are segmented based on the category, and the above operation is repeated until each pixel in the standard remote sensing image is traversed to obtain the target water body image. The calculation formula for the category corresponding to the maximum probability using the argmax function is as follows: Among them, C k Represents the kth category among the total number of categories in the standard remote sensing image.

6. The method for comprehensive ecological assessment of key areas based on natural resource survey and monitoring data according to claim 1 is characterized by: Based on the number of water body pixels, a water body ecology assessment is performed on the target key area to obtain a water body ecology coefficient of the target key area, including: Acquire a water body pixel area sequence based on the water body pixel quantity sequence; Fitting the water body pixel area sequence into a water body pixel area curve; Comparing the water body pixel area curve with a preset standard water body pixel area curve to obtain an error between the two; The error is compared with a preset error threshold to obtain the water ecological coefficient of the target key area.

7. The method for comprehensive ecological assessment of key areas based on natural resource survey and monitoring data according to claim 1 is characterized by: Based on the soil pollution content of the multiple sampling points, the soil ecology of the target key area is evaluated to obtain the soil ecology coefficient of the target key area, including: The pollution index of each pollution factor in the target key area is calculated, and the pollution index calculation formula of the pollution factor is as follows: Among them, P i represents the pollution index of the ith pollution factor in the target key area, U i represents the measured value of the i-th pollution factor in the target key area, S i represents the evaluation standard value of the i-th pollution factor in the target key area; The comprehensive pollution index of each pollution factor in the target key area is calculated based on the maximum value and average value of the pollution index of each pollution factor in the target key area. The calculation formula of the comprehensive pollution index of the pollution factor is as follows: in, represents the comprehensive pollution index of the ith pollution factor at the nth sampling point in the target key area, It represents the maximum value of the pollution index of the ith pollution factor at the nth sampling point in the target key area. It represents the average value of the pollution index of the ith pollution factor at the nth sampling point in the target key area.

8. The method for comprehensive ecological assessment of key areas based on natural resource survey and monitoring data according to claim 7 is characterized by: Based on the soil pollution content of the plurality of sampling points, soil ecological assessment is performed on the target key area to obtain the soil ecological coefficient of the target key area, and further includes: The soil ecological coefficient in the target key area is calculated based on the toxicity coefficient of each pollution factor and the comprehensive pollution index of each pollution factor in the target key area. The soil ecological coefficient calculation formula is as follows: Among them, RI represents the soil ecological coefficient in the target key area, T i represents the toxicity coefficient of the i-th pollution factor, and M represents the total number of sampling points in the target key area.

9. The method for comprehensive ecological assessment of key areas based on natural resource survey and monitoring data according to claim 1 is characterized by: Based on the spectral information of the target key area, the vegetation coverage ecological assessment is performed on the target key area to obtain the vegetation coverage ecological coefficient of the target key area, including: A pixel binary model is constructed, and the pixel binary model is as follows: S=S V +S S ; Where S represents the spectral information observed by the remote sensing sensor, S V Represents the spectral information of the green vegetation part, S S represents the spectral information of the part without vegetation coverage, and S V =S veg *f c , S veg represents the spectral information of pure pixels covered by vegetation, f c represents vegetation coverage, and S S =S soil *(1-f c ), S soil Represents the spectral information obtained from pure pixels covered entirely by soil; The pixel binary model is solved based on the spectral information of the target key area to obtain the vegetation coverage.

10. The method for comprehensive ecological assessment of key areas based on natural resource survey and monitoring data according to claim 9, characterized in that: The vegetation coverage calculation formula is as follows:

Images