Ecological environmental protection AI monitoring system based on computer vision

Abstract

The invention relates to the technical field of ecological environmental protection AI monitoring, and particularly discloses an ecological environmental protection AI monitoring system based on computer vision. Identifying shoreline area environment parameters, water body area pollution parameters and water body area water quality parameters in the target ecological research water area to obtain a shoreline area environment evaluation coefficient, a water body area pollution evaluation coefficient and a water body area water quality evaluation coefficient, and further evaluating the ecological function of the target ecological research water area. Therefore, accurate identification and quantitative analysis of the ecological function of the target ecological research water area are realized, whether the ecological function reaches the standard or not is judged and fed back, intelligent monitoring and decision basis is provided for ecological protection, planning and management of the target ecological research water area, ecological monitoring efficiency and accuracy are improved, and scientific management of the ecological environment is assisted.

Description

Technical Field

[0001] The present invention relates to the field of ecological and environmental protection AI monitoring technology, and specifically to an ecological and environmental protection AI monitoring system based on computer vision. Background Art

[0002] Traditional water ecological environment monitoring methods mainly rely on manual sampling and single-point sensor detection, which has the following limitations: manual sampling can only cover local areas, making it difficult to achieve comprehensive monitoring of large-scale water ecology; the detection cycle is long and cannot reflect the dynamic changes of the ecological environment in a timely manner;

[0003] Traditional methods usually only test a single parameter (such as water temperature and pH value) and lack a comprehensive analysis of multi-dimensional ecosystem parameters, resulting in unreliable and one-sided results of water ecological assessments.

[0004] In order to solve the above-mentioned defects, a technical solution is now provided. Summary of the Invention

[0005] The purpose of the present invention is to provide an ecological and environmental protection AI monitoring system based on computer vision to solve the problems raised in the above background.

[0006] The purpose of the present invention can be achieved through the following technical solutions: an ecological and environmental protection AI monitoring system based on computer vision, comprising:

[0007] A remote sensing image acquisition module is used to acquire remote sensing images of the target ecological research waters, preprocess the remote sensing images of the target ecological research waters, obtain preprocessed remote sensing images of the target ecological research waters, and identify the types of land cover in the preprocessed remote sensing images of the target ecological research waters, where the land cover types include water bodies and shorelines, to obtain a classified image of land cover of the target ecological research waters;

[0008] The shoreline area environmental identification module is used to identify shoreline area environmental parameters based on the remote sensing image of the shoreline area of ​​the target ecological research waters. The shoreline area ecological parameters include vegetation coverage and shoreline erosion index;

[0009] The shoreline area environmental analysis module is used to obtain the shoreline area environmental assessment coefficient of the target ecological research water area based on the shoreline area environmental parameters of the target ecological research water area;

[0010] A water body area pollution identification module is used to identify water body area pollution parameters based on the remote sensing image of the water body area of ​​the target ecological research water area. The water body area pollution parameters include a first water body pollution degree and a second water body pollution degree;

[0011] A water body area pollution analysis module is used to obtain a water body area pollution assessment index for the target ecological research water area based on water body area pollution parameters of the target ecological research water area;

[0012] The water quality identification module is used to identify water quality parameters of the water body area based on the remote sensing image of the water body area of ​​the target ecological research water area. The water quality parameters of the water body area include water temperature, pH, dissolved oxygen content and turbidity;

[0013] A water body area water quality analysis module is used to obtain a water body area water quality assessment index for the target ecological research water area based on the water body area water quality parameters of the target ecological research water area;

[0014] The ecological function assessment module is used to analyze the ecological function assessment coefficient of the target ecological research water area based on the shoreline area environmental assessment coefficient, the water area pollution assessment coefficient and the water quality assessment coefficient of the water area of ​​the target ecological research water area, and provide corresponding feedback.

[0015] Beneficial effects of the present invention:

[0016] The present invention uses remote sensing images of the shoreline area of ​​the target ecological research waters, combined with spectral feature extraction of near-infrared and red light bands, calculates vegetation coverage through the normalized difference vegetation index (NDVI) algorithm, and introduces time series comparison (initial and current vegetation coverage change rate) to quantify the rate of shoreline vegetation degradation. At the same time, by calculating the Euclidean distance of the three-dimensional coordinates of the monitoring points near the waterline and the landline, combined with the comparison of initial shoreline data, the degree of erosion is converted into a quantifiable index, overcoming the subjectivity of traditional visual interpretation. The vegetation degradation index and the erosion degree index are coupled through a cone volume model to form a shoreline environmental assessment coefficient, realizing the visual expression of the "vegetation-landform" two-dimensional ecological impact. Compared with single parameter evaluation, the early warning accuracy of shoreline ecological risks is improved to a certain extent.

[0017] The present invention analyzes color anomalies through the difference of RGB channel chromaticity values ​​in the remote sensing image of the water area of ​​the target ecological research water area, and then identifies the first pollution degree of the water area. The YOLO algorithm is used to detect floating objects, and the second pollution degree of the water area is identified according to the hazard impact factors, number and total area of ​​floating objects of each hazard level. The automatic acquisition, processing and analysis of remote sensing images are realized, which greatly improves the monitoring efficiency and accuracy, reduces human intervention and subjective errors, and comprehensively evaluates the water pollution status based on the first pollution degree and the second pollution degree, providing strong support for water pollution control.

[0018] The present invention inverts water temperature based on the split-window algorithm, combines color space conversion, spectral feature extraction, texture feature analysis and cosine similarity algorithm to accurately identify water quality parameters from different angles, providing reliable data for water quality monitoring and management, and improving the accuracy, reliability and intelligence of water quality identification.

[0019] The present invention comprehensively evaluates the ecological function of the target ecological research waters by comprehensively considering the shoreline area environmental assessment coefficient, the water area pollution assessment coefficient and the water quality assessment coefficient of the water area, provides a holistic decision-making basis for ecological protection and restoration, compares the ecological function assessment coefficient with the preset ecological function assessment index threshold, and promptly provides feedback on whether the ecological function of the water area meets the standard, which helps relevant departments to take timely measures for ecological protection and restoration, and improve the efficiency and effectiveness of ecological protection work. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The present invention will be further described below with reference to the accompanying drawings.

[0021] Figure 1 It is a system block diagram of the present invention.

[0022] Figure 2 This is a workflow diagram for inverting water temperature using the split-window algorithm of the present invention.

[0023] Figure 3 It is a logical schematic diagram of the water area pollution identification module of the present invention. DETAILED DESCRIPTION

[0024] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0025] See also Figure 1 As shown, the present invention is an ecological and environmental protection AI monitoring system based on computer vision, comprising:

[0026] The remote sensing image acquisition module is used to obtain remote sensing images of the target ecological research waters, and preprocess the remote sensing images of the target ecological research waters to obtain the preprocessed remote sensing images of the target ecological research waters, and identify the land feature coverage types in the preprocessed remote sensing images of the target ecological research waters, where the land feature coverage types include water bodies and coastlines, to obtain the land feature coverage classification images of the target ecological research waters.

[0027] It should be noted that the preprocessing of remote sensing images of the target ecological research waters includes three types of correction: geometric correction, atmospheric correction and radiation correction. Specifically, the remote sensing images of the target ecological research waters can be imported into ENVI software and the correction tools of the software can be used to complete the operation; then image enhancement processing is performed to highlight the characteristics of the ground objects by enhancing the grayscale, contrast, edges, colors, etc. of the image, making the ground object information easier to identify and analyze.

[0028] Furthermore, the specific process of identifying the coverage types of various features in the remote sensing image of the target ecological research water area is as follows:

[0029] The grayscale values ​​in the preprocessed remote sensing image of the target ecological research waters are obtained through image processing technology, and are matched with the grayscale value interval range of each land use type in the preset remote sensing image. When a grayscale value in the remote sensing image of the target ecological research waters belongs to the grayscale value interval range of a certain land use type, the area where the grayscale value in the remote sensing image of the target ecological research waters is located is recorded as the area corresponding to the land use type in the remote sensing image of the target ecological research waters. In this way, the areas corresponding to each land use type in the remote sensing image of the target ecological research waters are obtained statistically.

[0030] The shoreline area environment identification module is used to identify shoreline area environmental parameters based on the remote sensing image of the shoreline area of ​​the target ecological research waters. The shoreline area environmental parameters include vegetation degradation index and shoreline erosion degree index;

[0031] Furthermore, the process of identifying the environmental parameters of the shoreline area is as follows:

[0032] The pre-processed shoreline area remote sensing image of the target ecological research waters is divided according to the preset grid division method to obtain the remote sensing image of each shoreline sub-area. The near-infrared band and the red light band are selected to extract the spectral features of the remote sensing image of each shoreline sub-area, and the near-infrared band reflectance and the red light band reflectance of each pixel in the remote sensing image of each shoreline sub-area are obtained, which are recorded as i1 represents the number of the shoreline sub-area, i1 = 1, 2, ...., a1, a1 represents the maximum number of the shoreline sub-area number; j1 represents the number of the pixel, j1 = 1, 2, ...., b1, b1 represents the maximum number of the pixel number; through the normalized vegetation index algorithm Calculate the normalized vegetation index of each pixel in the remote sensing image of each coastline sub-area The normalized vegetation index of each pixel in the remote sensing image of each shoreline sub-area is accumulated and calculated to obtain the vegetation coverage of the shoreline area;

[0033] Obtain the initial coastline area remote sensing image of the target ecological research water area from the system repository, and analyze the vegetation coverage of the initial coastline area at the same time, and extract the acquisition time point of the initial coastline area remote sensing image of the target ecological research water area and the acquisition time point corresponding to the acquisition interval length of the current coastline area remote sensing image of the target ecological research water area, and calculate the vegetation coverage change rate of the coastline area;

[0034] It should be noted that the vegetation coverage change rate of the coastline area is the difference between the current coastline area vegetation coverage and the initial coastline area vegetation coverage, and the ratio of the difference value to the acquisition interval length;

[0035] Match the acquisition interval length with the preset reference vegetation coverage change rate corresponding to each acquisition interval length to obtain the reference vegetation coverage change rate of the coastline area, and calculate the vegetation degradation index VDI of the coastline area by the preset formula , wherein η and η' represent the vegetation coverage change rate and the reference vegetation coverage change rate of the coastline area, and e represents the natural constant;

[0036] Identify the waterline and the landline in the coastline area remote sensing image of the target ecological research water area, and arrange each monitoring point on the waterline and the landline respectively, and the monitoring points arranged on the waterline and the landline follow the one-to-one correspondence rule, and obtain the three-dimensional coordinates of each monitoring point on the waterline And the three-dimensional coordinates of each monitoring point on the landline h represents the number of each monitoring point, h = 1, 2,..., m, and m is the maximum value of the monitoring point number; and the total distance d of the coastline area monitoring points on the waterline and the landline is calculated by the Euclidean distance algorithm

[0037] It should be noted that the one-to-one correspondence rule is that the first monitoring point on the waterline corresponds to the first monitoring point on the landline, and the first monitoring point on the waterline corresponds to the first monitoring point on the landline.

[0038] Similarly, the initial waterline and the initial landline in the initial coastline area remote sensing image of the target ecological research water area are identified, and each monitoring point is arranged, and the total distance d' of the initial coastline area monitoring points on the waterline and the landline is obtained by analysis;

[0039] It should be noted that the number of monitoring points arranged on the initial waterline and the initial landline in the initial coastline area remote sensing image of the target ecological research water area is consistent with the number of monitoring points arranged on the waterline and the landline in the current coastline area remote sensing image of the target ecological research water area, and also follows the one-to-one correspondence rule.

[0040] The total distance d of the coastline area monitoring points on the waterline and the landline is calculated by the preset formula ​The shoreline erosion degree index SEDI of the shoreline area is calculated, where tanh is the hyperbolic tangent function operator, and Δd represents the preset allowable total distance difference;

[0041] The shoreline area environmental analysis module is used to obtain the shoreline area environmental assessment coefficient of the target ecological research water area based on the shoreline area environmental parameters of the target ecological research water area;

[0042] Furthermore, the analysis process of the shoreline area environmental assessment index of the target ecological research waters is as follows:

[0043] The vegetation degradation index and shoreline erosion index of the shoreline area of ​​the target ecological research waters were extracted and converted into lengths according to a preset ratio. A cone was constructed with the lengths of the vegetation degradation index and shoreline erosion index as the radius and height respectively, and the value of the cone volume was extracted as the environmental assessment coefficient of the shoreline area of ​​the target ecological research waters.

[0044] The water body area pollution identification module is used to identify water body area pollution parameters based on the water body area remote sensing image of the target ecological research water area. The water body area pollution parameters include the first water body pollution degree and the second water body pollution degree.

[0045] Furthermore, if Figure 3 As shown, the specific process of identifying the pollution parameters of the water area is as follows:

[0046] The pre-processed remote sensing images of the target ecological research waters are divided according to the preset grid division method to obtain remote sensing images of each water body sub-area. The chromaticity values ​​of each pixel point of the remote sensing image of each water body sub-area on the three color channels of red (R), green (G), and blue (B) are detected to obtain the chromaticity values ​​of each pixel point of the remote sensing image of each water body sub-area on the three color channels of red (R), green (G), and blue (B). The chromaticity values ​​of the preset normal water body on the three color channels of red (R), green (G), and blue (B) are compared, and the difference between the two is obtained to obtain the chromaticity value difference of each pixel point of the remote sensing image of each water body sub-area on the R, G, and B channels. The difference is compared with the preset chromaticity value difference threshold to screen out the abnormal chromaticity pixels and count the number of the abnormal chromaticity pixels of the remote sensing image of each water body sub-area, which is recorded as i2 represents the number of the water body sub-region, i2=1, 2, ...., a2, reads the total number of pixels in each water body sub-region Calculated by formula Get the first water pollution level CPI1 of the research water area;

[0047] It should be noted that the specific analysis method for screening out abnormal chromaticity pixels is:

[0048] Extract the chromaticity value of each pixel in the remote sensing image of each water body sub-region on the three color channels of red (R), green (G), and blue (B), and record it as At the same time, the standard chromaticity values ​​of normal water in the three color channels of red (R), green (G), and blue (B) are obtained and recorded as j2 represents the pixel number, j2=1, 2, ...., b2, b2 represents the maximum value of the pixel number; through the formula Calculate the color value difference between each pixel point of the remote sensing image of each water body sub-region and the normal water body in the R channel Denote the R channel chromaticity value difference of each pixel point of the remote sensing image of each water body sub-region. Similarly, the G channel chromaticity value difference and B channel chromaticity value difference of each pixel point of the remote sensing image of each water body sub-region are analyzed and compared with the preset chromaticity value difference threshold. If the chromaticity value difference of a pixel point on any channel is greater than the preset chromaticity value difference threshold, the pixel point is judged to be an abnormal chromaticity pixel point. If the chromaticity value difference of a pixel point on the R, G, and B channels is less than or equal to the preset chromaticity value difference threshold, the pixel point is judged to be a qualified chromaticity pixel point.

[0049] It should be noted that, in a specific embodiment, the preset chromaticity value difference thresholds for the red (R), green (G), and blue (B) color channels are 5, 3, and 4, respectively. If the chromaticity value difference of a pixel on the R channel is 6, the chromaticity value difference on the G channel is 4, and the chromaticity value difference on the G channel is 3, since the chromaticity value differences of the pixel on the R channel and the G channel all exceed their corresponding chromaticity value difference thresholds, the pixel is determined to be an abnormal chromaticity pixel;

[0050] The YOLO target detection algorithm is used to identify floating objects of different hazard levels in the remote sensing images of each water body sub-region. The number and area of ​​floating objects of each hazard level are extracted, and the areas corresponding to all the numbers of floating objects of each hazard level are accumulated to obtain the total area corresponding to the floating objects of each hazard level.

[0051] It should be noted that the floating objects of different hazard levels include light hazard level floating objects, medium hazard level floating objects, severe hazard level floating objects and extremely severe hazard level floating objects. Among them, light hazard level floating objects include but are not limited to: fallen leaves and grass clippings; medium hazard level floating objects include but are not limited to: aquatic organism debris and branches; severe hazard level floating objects include but are not limited to: domestic garbage (kitchen waste, fabrics); extremely severe hazard level floating objects include but are not limited to: plastics and algae (such as blue algae and green algae) forming a "water bloom" floating layer;

[0052] The number and total area of ​​floating objects of each hazard level in the remote sensing image of each water body sub-region are extracted and normalized and the values ​​are taken as At the same time, the hazard impact factors corresponding to floating objects of each hazard level are extracted from the system repository and recorded as x represents the number of the hazardous floating object, x=1, 2, ...., p, p represents the maximum value of the hazardous floating object number; according to the formula The second water pollution level CPI2 of the research waters was calculated. Indicates the maximum number of floating objects of each hazard level in the remote sensing image of each water body sub-area, Represents the area of ​​each water body sub-region.

[0053] The water body regional pollution analysis module obtains the water body regional pollution assessment index of the target ecological research water area based on the water body regional pollution parameters of the target ecological research water area;

[0054] Furthermore, the analysis process of the water body regional pollution assessment index of the target ecological research water area is as follows:

[0055] Extract the first water pollution degree and the second water pollution degree of the target ecological research water area respectively, and substitute them into the preset formula The water pollution assessment coefficient PAC of the target ecological research water area is calculated, where α and β represent the weight factors of the preset first water pollution degree and second water pollution degree respectively;

[0056] For example, α=β=0.5, which can be specifically set by using existing determination methods such as subjective assignment method or Delphi method.

[0057] The water quality identification module identifies water quality parameters of the water body area based on the remote sensing image of the water body area of ​​the target ecological research water area. The water quality parameters of the water body area include water temperature, pH, dissolved oxygen content and turbidity.

[0058] Furthermore, the specific process of identifying the water quality parameters of the water body area is as follows:

[0059] The water temperature of the target ecological research waters was obtained by inversion based on the split-window algorithm;

[0060] It should be noted that if Figure 2 As shown in Figure 2, the specific process of inverting the water temperature of each sub-area of ​​the target ecological research waters based on the split-window algorithm includes:

[0061] Obtain satellite data with thermal infrared split window bands for each water body sub-region, such as Landsat 8 (TIRS bands 10 and 11) remote sensing data;

[0062] The Landsat 8 remote sensing data was radiometrically calibrated and atmospheric correction was performed on its OLI band using the FLAASII tool.

[0063] Bands 3, 4, 5, and 6 of the OLI band are used to calculate indexes related to water body identification, such as the Modified Normalized Difference Water Index (MNDWI). The formula is: Green and SWIR1 are the reflectances of the green band and the shortwave infrared 1 band, respectively, to highlight water body information. It can also be combined with NDVI (Normalized Difference Vegetation Index, which distinguishes water bodies from vegetation) and NDBI (Normalized Building Index, which distinguishes water bodies from buildings) to accurately extract water body areas and clarify the inversion range through decision tree classification.

[0064] For the Landsat 8TIRS thermal infrared bands (bands 10 and 11), the radiation values ​​received by the sensor are converted into water brightness temperature according to the Planck formula. The formula is: B(λ,T) is the spectral radiance, λ is the central wavelength of the band, T is the brightness temperature, C1 and C2 are Planck constants, and the brightness temperature data of the water body is obtained by inversion. At the same time, the water body emissivity is determined as the key parameter for subsequent temperature inversion;

[0065] Combined with the atmospheric transmittance data provided by NASA (or obtained through atmospheric radiation transfer model simulation), the split-window algorithm formula is used to input the brightness temperature, emissivity, atmospheric transmittance and other parameters of the two thermal infrared bands of TIRS to invert and calculate the water temperature. The formula is: T = a0 + a1T 10 +a2T 11 +a3(ε 10 -ε 11 )+a4(τ 10 -τ 11 ), where T is the water temperature, T 10 、T 11 are the brightness temperatures of the two thermal infrared bands, ε 10 , ε 11 is the corresponding band emissivity, τ 10 , τ 11 is the atmospheric transmittance, a0, a1, a2, a3, and a4 are algorithm coefficients, which are determined based on the atmospheric conditions of the ecological research waters;

[0066] Use measured water temperature data (such as field temperature measurement and hydrological station data) to verify the accuracy of the inverted water temperature through statistical analysis (such as calculating the root mean square error and correlation coefficient) and evaluate the reliability of the process. If the error is large, retrospectively adjust the parameters (such as the emissivity value and algorithm coefficient).

[0067] The remote sensing images of each water body sub-region are converted from RGB color space to HSV color space, and the hue, saturation and brightness of each pixel point of the remote sensing images of each water body sub-region are extracted and recorded as g represents the number of each pixel, g = 1, 2, ...., q, q represents the maximum value of the pixel number; through the formula Obtain the hue mean value of each water body sub-region remote sensing image

[0068] Similarly, the saturation mean value of each water body sub-region remote sensing image is obtained by analysis And the brightness mean value

[0069] The hue mean value of each water body sub-region remote sensing image is averaged to obtain the hue mean value of the water body region remote sensing image, denoted as

[0070] Similarly, the saturation mean value of the water body region remote sensing image is obtained by analysis And the brightness mean value

[0071] The hue mean value, saturation mean value and brightness mean value of the water body region remote sensing image form the color feature vector of the water body region remote sensing image And the color feature vector of the water body region remote sensing image is calculated by the cosine similarity algorithm formula The matching degree of the color feature vector of the water body region remote sensing image and the color feature vector corresponding to each pH value is obtained The color feature vector corresponding to each pH value is denoted as The standard hue mean value, standard saturation mean value and standard brightness mean value in the color feature vector corresponding to each pH value are denoted as

[0072] The maximum matching degree of the color feature vector of the water body region remote sensing image and the color feature vector corresponding to each pH value is extracted, and the pH value corresponding to the maximum matching degree is taken as the pH value of the water body region.

[0073] It should be noted that the pH value of the water body affects the color of the water body. Under different pH values, the spectral absorption and reflection characteristics of certain substances in the water body (such as substances based on the principle of acid-base indicator, or the optical properties of dissolved substances in natural water bodies) are different. For example, acidic water bodies may have yellow or red tones, and alkaline water bodies may have blue or green tones (due to the optical performance of dissolved minerals, microorganisms, etc. in the water body and the effect of acid-base).

[0074] Select each band to extract the spectral features of the remote sensing images of each water body sub-region, obtain the spectral feature information of each band corresponding to the remote sensing images of each sub-region, extract the blue light band reflectance, green light band reflectance, red light band reflectance and near infrared band reflectance of each pixel in the remote sensing images of each water body sub-region, and calculate the ratio of the blue light band reflectance to the green light band reflectance to obtain the blue-green band ratio of each pixel in the remote sensing images of each water body sub-region, calculate the ratio of the red light band reflectance to the near infrared band reflectance to obtain the red-near infrared ratio of each pixel in the remote sensing images of each water body sub-region, and perform average calculation to obtain the average blue-green band ratio and the average red-near infrared ratio of the remote sensing images of the water body region;

[0075] The spectral feature vector of the studied water image is constructed by the mean of the blue-green band ratio and the mean of the red-near infrared ratio. The spectral feature vectors corresponding to the preset dissolved oxygen contents are calculated by the cosine similarity algorithm formula. Calculate the matching degree between the spectral feature vector of the remote sensing image of the water area and the spectral feature vector corresponding to each dissolved oxygen content Represents the color feature vector corresponding to each dissolved oxygen content, They are respectively expressed as the mean of the standard blue-green band ratio and the mean of the red-near infrared ratio in the color feature vector corresponding to each pH value;

[0076] Extract the maximum matching degree between the spectral feature vector of the remote sensing image of the water area and the spectral feature vector corresponding to each dissolved oxygen content, and use the dissolved oxygen content corresponding to the maximum matching degree as the dissolved oxygen content of the study water area;

[0077] It should be noted that dissolved oxygen levels can affect the activity of plankton and microorganisms in water, thereby altering the water's optical properties. For example, in water with high dissolved oxygen levels, phytoplankton photosynthesis is strong, which can cause changes in the chlorophyll content and affect spectral reflectance.

[0078] Gray-level co-occurrence matrix (GLCM) was used to extract the texture feature parameters of the remote sensing images of each water body sub-region, including contrast, entropy and angular second-order moment. The contrast, entropy and angular second-order moment of the remote sensing images of each water body sub-region were averaged to obtain the mean contrast, entropy and angular second-order moment of the remote sensing images of the water body region.

[0079] It should be noted that the calculation formulas for contrast, entropy, and angular second moment are as follows:

[0080]

[0081] Where N is the number of gray levels, P(s,t) is the element in the gray-level co-occurrence matrix, s represents the s-th row in the gray-level co-occurrence matrix, and t represents the t-th column in the gray-level co-occurrence matrix;

[0082] The texture feature vector of the water area remote sensing image is constructed by the contrast mean, entropy mean and angular matrix mean of the water area remote sensing image And compare it with the texture feature vector corresponding to each preset turbidity through the cosine similarity algorithm formula

[0083] Calculate the matching degree between the texture feature vector of the studied water area image and the texture feature vector corresponding to each turbidity Represents the texture feature vector corresponding to each turbidity, They are respectively expressed as the standard contrast mean, standard entropy mean, and standard angular matrix mean in the color feature vector corresponding to each turbidity;

[0084] Extract the maximum matching degree between the texture feature vector of the studied water area image and the texture feature vector corresponding to each turbidity, and use the turbidity corresponding to the maximum matching degree as the turbidity of the studied water area;

[0085] It should be noted that suspended particulate matter will affect the transparency and turbidity of the water body, changing the texture and grayscale distribution of the water image. The more particulate matter there is, the darker the water image may be and the rougher the texture.

[0086] A water body area water quality analysis module is used to obtain a water body area water quality assessment index for the target ecological research water area based on the water body area water quality parameters of the target ecological research water area;

[0087] Furthermore, the analysis process of the water quality assessment index of the target ecological research water area is as follows:

[0088] Extract the water temperature range, pH range, dissolved oxygen content range and turbidity threshold corresponding to the normal water body stored in the system repository, obtain the upper and lower limits of the water temperature range, pH range and dissolved oxygen content range and calculate the mean to obtain the ideal reference water temperature, ideal reference pH and ideal reference dissolved oxygen content, and subtract them from the lower limit of the water temperature range, pH range and dissolved oxygen content range to obtain the ideal deviation water temperature, ideal deviation pH and ideal deviation dissolved oxygen content.

[0089] Extract the water temperature, ideal reference water temperature, and ideal deviation water temperature of the target ecological research water area, and use the formula Calculate the water temperature deviation value of the target ecological research water area;

[0090] The pH deviation and dissolved oxygen content deviation of the target ecological research waters were obtained by similar analysis;

[0091] extracting the turbidity of the target ecological research water area, the turbidity threshold value, and calculating the turbidity deviation value of the target ecological research water area by the formula

[0092] adding up the water temperature deviation value, the pH deviation value, the dissolved oxygen content deviation value and the turbidity deviation value of the target ecological research water area to obtain the water body area water quality evaluation index of the target ecological research water area.

[0093] The ecological function evaluation module is configured to analyze the ecological function evaluation coefficient of the target ecological research water area according to the shoreline area environment evaluation coefficient, the water body area pollution evaluation coefficient and the water body area water quality evaluation coefficient of the target ecological research water area, and to perform corresponding feedback.

[0094] Further, the analysis of the ecological function evaluation coefficient of the target ecological research water area and the corresponding feedback include:

[0095] extracting the shoreline area environment evaluation coefficient, the water body area pollution evaluation coefficient and the water body area water quality evaluation coefficient of the target ecological research water area, and substituting them into the preset Softplus function formula to calculate the ecological function evaluation coefficient EFC of the target ecological research water area, wherein τ1, τ2 and τ3 represent the preset shoreline area environment impact factor, the water body area pollution impact factor and the water body area water quality impact factor, respectively.

[0096] For example, τ1=0.4, τ2=τ3=0.3.

[0097] Comparing the ecological function evaluation coefficient of the target ecological research water area with the preset ecological function evaluation index threshold value, if the ecological function evaluation coefficient of the target ecological research water area is greater than or equal to the preset ecological function evaluation index threshold value, it is determined that the ecological function of the target ecological research water area is not up to standard, and if the ecological function evaluation coefficient of the target ecological research water area is less than the preset ecological function evaluation index threshold value, it is determined that the ecological function of the target ecological research water area is up to standard, and feedback is performed to the system.

[0098] The above is only an example and description of the structure of the present application. Those skilled in the art can make various modifications or supplements to the described specific embodiments or use similar ways to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined by the present claims, and they should belong to the protection scope of the present application.​

Claims

1. The ecological and environmental protection AI monitoring system based on computer vision is characterized by: include: A remote sensing image acquisition module is used to acquire remote sensing images of the target ecological research waters and perform preprocessing, identify the types of land cover in the remote sensing images of the target ecological research waters, where land cover types include water bodies and shorelines, and obtain a classified image of land cover of the target ecological research waters; The shoreline area environmental identification module is used to identify shoreline area environmental parameters based on the remote sensing image of the shoreline area of ​​the target ecological research waters. The shoreline area ecological parameters include vegetation coverage and shoreline erosion index; The shoreline area environmental analysis module is used to obtain the shoreline area environmental assessment coefficient of the target ecological research water area based on the shoreline area environmental parameters of the target ecological research water area; A water body area pollution identification module is used to identify water body area pollution parameters based on the remote sensing image of the water body area of ​​the target ecological research water area. The water body area pollution parameters include a first water body pollution degree and a second water body pollution degree; A water body area pollution analysis module is used to obtain a water body area pollution assessment coefficient for the target ecological research water area based on the water body area pollution parameters of the target ecological research water area; The water quality identification module is used to identify water quality parameters of the water body area based on the remote sensing image of the water body area of ​​the target ecological research water area. The water quality parameters of the water body area include water temperature, pH, dissolved oxygen content and turbidity; A water body area water quality analysis module is used to obtain a water body area water quality assessment coefficient for the target ecological research water area based on the water body area water quality parameters of the target ecological research water area; The ecological function assessment module is used to analyze the ecological function assessment coefficient of the target ecological research water area based on the shoreline area environmental assessment coefficient, the water area pollution assessment coefficient and the water quality assessment coefficient of the water area of ​​the target ecological research water area, and provide corresponding feedback.

2. The computer vision-based ecological and environmental protection AI monitoring system according to claim 1 is characterized in that: The specific process of identifying the types of land cover in the remote sensing images of the target ecological research waters is as follows: The grayscale values ​​in the preprocessed remote sensing image of the target ecological research waters are obtained through image processing technology, and are matched with the grayscale value interval range of each land use type in the preset remote sensing image. When a grayscale value in the remote sensing image of the target ecological research waters belongs to the grayscale value interval range of a certain land use type, the area where the grayscale value in the remote sensing image of the target ecological research waters is located is recorded as the area corresponding to the land use type in the remote sensing image of the target ecological research waters. In this way, the areas corresponding to each land use type in the remote sensing image of the target ecological research waters are obtained statistically.

3. The computer vision-based ecological and environmental protection AI monitoring system according to claim 1 is characterized in that: The process of identifying the environmental parameters of the shoreline area is as follows: The pre-processed remote sensing image of the shoreline area of ​​the target ecological research waters is divided according to a preset grid division method to obtain remote sensing images of each shoreline sub-area. The near-infrared band and the red light band are selected to extract spectral features of the remote sensing images of each shoreline sub-area to obtain the near-infrared band reflectance and the red light band reflectance of each pixel in the remote sensing image of each shoreline sub-area. The normalized vegetation index of each pixel in the remote sensing image of each shoreline sub-area is calculated by the normalized vegetation index algorithm. The normalized vegetation index of each pixel in the remote sensing image of each shoreline sub-area is accumulated and calculated to obtain the vegetation coverage of the shoreline area. The initial shoreline area remote sensing image of the target ecological research water area is obtained from the system repository, and the vegetation coverage of the initial shoreline area is obtained by similar analysis. At the same time, the acquisition time point of the initial shoreline area remote sensing image of the target ecological research water area and the acquisition time point of the shoreline area remote sensing image of the current target ecological research water area are extracted, and the vegetation coverage change rate of the shoreline area is calculated; The acquisition interval is matched with the reference vegetation coverage change rate corresponding to each preset acquisition interval to obtain the reference vegetation coverage change rate of the shoreline area, and the vegetation degradation index of the shoreline area is calculated using a preset formula; The waterline and landline are identified in the remote sensing image of the shoreline area of ​​the target ecological research waters. Monitoring points are arranged on the waterline and landline respectively, and the monitoring points arranged on the waterline and landline follow a one-to-one correspondence rule. The three-dimensional coordinates of each monitoring point on the waterline and the three-dimensional coordinates of each monitoring point on the landline of the shoreline area are obtained. The total distance between the monitoring points on the waterline and the monitoring points on the landline of the shoreline area is calculated using the Euclidean distance algorithm; Similarly, the initial waterline and initial landline were identified from the remote sensing image of the initial shoreline area of ​​the target ecological research waters, and monitoring points were arranged. The total distance between the waterline monitoring points and the landline monitoring points in the initial shoreline area was analyzed. The shoreline erosion degree index of the shoreline area is calculated using a preset formula.

4. The computer vision-based ecological and environmental protection AI monitoring system according to claim 1, characterized in that: The analysis process of the shoreline area environmental assessment index of the target ecological research waters is as follows: The vegetation degradation index and shoreline erosion index of the shoreline area of ​​the target ecological research waters were extracted and converted into lengths according to a preset ratio. A cone was constructed with the lengths of the vegetation degradation index and shoreline erosion index as the radius and height respectively, and the value of the cone volume was extracted as the environmental assessment coefficient of the shoreline area of ​​the target ecological research waters.

5. The computer vision-based ecological and environmental protection AI monitoring system according to claim 1 is characterized in that: The specific process of identifying the pollution parameters of the water area is as follows: The pre-processed remote sensing image of the target ecological research water area is divided according to a preset grid division method to obtain remote sensing images of each water body sub-area, and the chromaticity value of each pixel point of the remote sensing image of each water body sub-area on the three color channels of red, green and blue is detected to obtain the chromaticity value of each pixel point of the remote sensing image of each water body sub-area on the three color channels of red, green and blue, and compared with the preset standard chromaticity value of normal water body on the three color channels of red, green and blue, and the difference between the two is obtained to obtain the chromaticity value difference of each pixel point of the remote sensing image of each water body sub-area on the R, G and B channels, and compared with the preset chromaticity value difference threshold, and each abnormal chromaticity pixel point is screened out, and the number of each abnormal chromaticity pixel point of the remote sensing image of each water body sub-area is counted, and the total number of pixels of each water body sub-area is read, and the first water pollution degree of the research water area is calculated by the formula; The YOLO target detection algorithm is used to identify floating objects of different hazard levels in the remote sensing images of each water body sub-region. The number and area of ​​floating objects of each hazard level are extracted, and the areas corresponding to all the numbers of floating objects of each hazard level are accumulated to obtain the total area corresponding to the floating objects of each hazard level. The number and total area of ​​floating objects of each hazard level in the remote sensing images of each water body sub-area are extracted, normalized and their values ​​are taken. At the same time, the hazard impact factors corresponding to floating objects of each hazard level are extracted from the system storage, and the pollution degree of the second water body in the study water area is calculated according to the formula.

6. The computer vision-based ecological and environmental protection AI monitoring system according to claim 1, characterized in that: The analysis process of the water area pollution assessment index of the target ecological research water area is as follows: The first water body pollution degree and the second water body pollution degree of the target ecological research water area are extracted respectively, and substituted into the preset formula to calculate the water body pollution assessment coefficient of the target ecological research water area.

7. The ecological and environmental protection AI monitoring system based on computer vision according to claim 1 is characterized in that: The specific process of identifying the water quality parameters of the water body area is as follows: The water temperature of the target ecological research waters was obtained by inversion based on the split-window algorithm; The remote sensing images of each water body sub-region are converted from RGB color space to HSV color space, the hue, saturation and brightness of each pixel of the remote sensing images of each water body sub-region are extracted, and the hue mean, saturation mean and brightness mean of the remote sensing images of each water body sub-region are calculated by averaging; The hue mean, saturation mean and brightness mean of the remote sensing image of each water body sub-region are averaged to obtain the hue mean, saturation mean and brightness mean of the remote sensing image of the water body region; The hue mean, saturation mean and brightness mean of the remote sensing image of the water body area are used to form the color feature vector of the remote sensing image of the water body area, and the color feature vector corresponding to each preset pH value is calculated by the cosine similarity algorithm formula to obtain the matching degree between the color feature vector of the remote sensing image of the water body area and the color feature vector corresponding to each pH value; Select each band to extract the spectral features of the remote sensing images of each water body sub-region, obtain the spectral feature information of each band corresponding to the remote sensing images of each sub-region, extract the blue light band reflectance, green light band reflectance, red light band reflectance and near-infrared band reflectance of each pixel in the remote sensing images of each water body sub-region, and obtain the average blue-green band ratio and the average red-near-infrared ratio of the remote sensing images of the water body region; The mean of the blue-green band ratio and the mean of the red-near infrared ratio constitute the spectral feature vector of the study water area image, and the spectral feature vector corresponding to each preset dissolved oxygen content is calculated by the cosine similarity algorithm formula to obtain the matching degree between the spectral feature vector of the water area remote sensing image and the spectral feature vector corresponding to each dissolved oxygen content, and the maximum matching degree among the matching degrees between the spectral feature vector of the water area remote sensing image and the spectral feature vector corresponding to each dissolved oxygen content is extracted, and the dissolved oxygen content corresponding to the maximum matching degree is used as the dissolved oxygen content of the study water area; The gray-level co-occurrence matrix is ​​used to extract the texture feature parameters of the remote sensing images of each water body sub-region, including contrast, entropy and angular second-order moment, and the contrast mean, entropy mean and angular second-order moment mean of the remote sensing images of the water body region are obtained; The contrast mean, entropy mean and angular matrix mean of the remote sensing image of the water area are used to form the texture feature vector of the remote sensing image of the water area, and the texture feature vector corresponding to each preset turbidity is calculated through the cosine similarity algorithm formula to obtain the matching degree between the texture feature vector of the study water area image and the texture feature vector corresponding to each turbidity. The maximum matching degree among the matching degrees of the texture feature vector of the study water area image and the texture feature vector corresponding to each turbidity is extracted, and the turbidity corresponding to the maximum matching degree is used as the turbidity of the study water area.

8. The computer vision-based ecological and environmental protection AI monitoring system according to claim 1, characterized in that: The analysis process of the water quality assessment index of the target ecological research water area is as follows: Extract the water temperature range, pH range, dissolved oxygen content range and turbidity threshold corresponding to the normal water body stored in the system repository, obtain the upper and lower limits of the water temperature range, pH range and dissolved oxygen content range and calculate the mean to obtain the ideal reference water temperature, ideal reference pH and ideal reference dissolved oxygen content, and subtract them from the lower limit of the water temperature range, pH range and dissolved oxygen content range to obtain the ideal deviation water temperature, ideal deviation pH and ideal deviation dissolved oxygen content. Extract the water temperature, ideal reference water temperature, and ideal deviation water temperature of the target ecological research water area, and calculate the water temperature deviation value of the target ecological research water area through the formula; The pH deviation and dissolved oxygen content deviation of the target ecological research waters were obtained by similar analysis; Extract the turbidity and turbidity threshold of the target ecological research waters, and calculate the turbidity deviation value of the target ecological research waters through the formula; The water temperature deviation values, pH deviation values, dissolved oxygen content deviation values ​​and turbidity deviation values ​​of the target ecological research waters are accumulated and calculated to obtain the water quality assessment index of the target ecological research waters.

9. The computer vision-based ecological and environmental protection AI monitoring system according to claim 1, characterized in that: The analysis of the ecological function assessment coefficient of the target ecological research waters and the corresponding feedback include: Extract the shoreline area environmental assessment coefficient, water area pollution assessment coefficient and water area water quality assessment coefficient of the target ecological research waters, and substitute them into the preset Softplus function formula to calculate the ecological function assessment coefficient of the target ecological research waters; The ecological function assessment coefficient of the target ecological research water area is compared with the preset ecological function assessment index threshold. If the ecological function assessment coefficient of the target ecological research water area is greater than or equal to the preset ecological function assessment index threshold, the ecological function of the target ecological research water area is judged to be substandard; if the ecological function assessment coefficient of the target ecological research water area is less than the preset ecological function assessment index threshold, the ecological function of the target ecological research water area is judged to be up to standard, and feedback is given to the system.

Images