Brinell vole harm monitoring method using computer vision

Abstract

The invention discloses a Brinell vole harm monitoring method using computer vision, and relates to the technical field of computer vision, and the method comprises the steps: obtaining a Brinell vole harm monitoring region, and carrying out the management region division; management area division and marking are carried out on the monitoring area, the unmanned aerial vehicle and the sensing equipment are used for obtaining data in the monitoring area, the rodent damage and vegetation state data of the monitoring area are analyzed, the damage comprehensive evaluation index is extracted according to the damage monitoring mechanism, and damage evaluation is carried out through the comprehensive evaluation model. According to the method, a remote sensing ecological index star-based library specially used for vegetation real-time state analysis is constructed, correlation coefficient analysis is performed by using a rodent damage assessment degree value, an objective function model between an ecological value and predicted rodent damage assessment is obtained, the mutual relation between vegetation and rodent damage in an ecological system is fully considered, and the rodent damage condition is predicted by using ecological data. The accuracy of rat damage assessment is improved, and a regional rat damage spatial distribution diagram is drawn.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer vision, and in particular to a method for monitoring Microtus brandti hazards using computer vision. BACKGROUND

[0002] Microtus brandti is a small rodent widely distributed in the Inner Mongolia grassland and surrounding areas of China. In the past few decades, due to changes in the grassland ecological environment and the influence of human activities, the population of Microtus brandti has fluctuated significantly, often leading to large-scale rodent damage, which poses a serious threat to the grassland ecosystem and agricultural and pastoral production. Microtus brandti feeds on a variety of fine forage grasses, and during the outbreak of rodent damage, large amounts of forage grasses are eaten, reducing the vegetation coverage, and significantly reducing the primary productivity of the grassland ecosystem. With the rapid development of information technology, computer vision technology has been widely applied in many fields. Computer vision uses image or video information collected by cameras and other image acquisition devices, and uses algorithms to process and analyze this information, thereby achieving the functions of recognizing, classifying, and positioning target objects.

[0003] Traditional methods for monitoring Microtus brandti hazards mainly rely on manual field surveys, which have many limitations. Manual surveys are inefficient and require a large amount of manpower, material resources, and time. It is difficult to achieve comprehensive and timely monitoring of the distribution and damage of Microtus brandti in vast grasslands and farmland areas. Manual surveys are highly subjective, and differences in judgment standards and experience among different survey personnel can lead to biased survey results, making it impossible to monitor the occurrence and development trend of rodent damage in real time.

[0004] In view of the above technical defects, a solution is proposed. SUMMARY

[0005] The purpose of the present application is to use large-scale remote sensing image data, cooperate with unmanned aerial remote sensing and high-time-efficiency ground surveys, build a comprehensive evaluation and early warning model for Microtus brandti hazards, and achieve multi-scale and all-around monitoring in the form of rodent damage area coverage, piece monitoring, and point estimation.

[0006] To achieve the above purpose, the present application adopts the following technical solution: a method for monitoring Microtus brandti hazards using computer vision, comprising the following steps:

[0007] Step 1: Divide the monitoring area of Microtus brandti hazards into management areas, build a divided monitoring area map, and mark each area, obtaining a divided marking network map of the monitoring area, and obtaining data in the monitoring area through unmanned aerial vehicles and sensing devices, forming a data set of the monitoring area;

[0008] Step two: obtain the dataset of the monitoring area, analyze the rodent and vegetation state data of the monitoring area, extract the comprehensive evaluation index of the hazard according to the pre-established hazard monitoring mechanism, generate the rodent comprehensive dataset and the vegetation ecological dataset, and based on the rodent comprehensive dataset, establish a hazard evaluation comprehensive evaluation model to dynamically evaluate the divided monitoring area, output the hazard analysis index of the monitoring area, and construct a remote sensing ecological index star-based library for real-time analysis of the vegetation state;

[0009] Step three: obtain the real-time monitoring area data and evaluate according to the hazard evaluation comprehensive evaluation model to generate the rodent hazard evaluation degree value of the real-time monitoring area;

[0010] Step four: based on the remote sensing ecological index star-based library, judge the real-time obtained ecological data of the monitoring area, generate the star-based vegetation ecological value of the regional point, and combine the rodent hazard evaluation degree value to perform correlation coefficient analysis to obtain the target function model between the ecological value and the predicted rodent hazard evaluation;

[0011] Step five: obtain the target function model between the ecological value and the predicted rodent hazard evaluation of the monitoring area, generate the total monitoring area data set, and draw the regional rodent hazard spatial distribution map according to the total monitoring area data set.

[0012] Further, the dataset of the monitoring area is formed, which specifically includes the following:

[0013] S100, obtain the initial area of the monitoring area, and divide the monitoring area into N sub-regions, N={1, 2, …, N}, record the initial division result, form a basic division document containing region boundary coordinates, area, and main ecological characteristics, and collect visible light, infrared, multispectral images and environmental data through multi-modal sensors, and pre-process and align the data;

[0014] S101, design regional markers: assign a unique code to each sub-region, and label key management information including the area manager and the latest rodent investigation time;

[0015] Overlay the marker network layer on the monitoring area map to form a "differential marker network map";

[0016] S102, based on the initial division result, integrate high-precision remote sensing images, geographic information data, and adjust the sub-region boundary;

[0017] Label the preset point of the unmanned aerial vehicle monitoring device, the deployment position of the sensor device, and the cruise path of the unmanned aerial vehicle cruise shooting, and input them into the internal of the monitoring area map;

[0018] S103, when the unmanned aerial vehicle regularly cruises, high-resolution images including mouse holes and activity tracks are acquired, and night activity data is detected by combining thermal infrared technology;

[0019] The sensor device acquires ground data through a soil moisture sensor and an infrared camera;

[0020] The ground data and the data acquired by the unmanned aerial vehicle are integrated and preprocessed to obtain a data set of the monitoring area, and are structured processed according to the area, time and data type, the data of the unmanned aerial vehicle and the sensor device are synchronized according to the set update time, and a visual report is generated.

[0021] Further, according to the pre-established hazard monitoring mechanism, the following is included:

[0022] S200, based on the ecological hazard characteristics of Microtus fortis, a four-dimensional index system is constructed, including population quantity index, cave related index and vegetation damage index;

[0023] S201, population quantity index data: ground active individuals are automatically counted by intelligent recognition of unmanned aerial vehicle aerial image; 50 to 100 individuals are marked in a typical sample area by marking recapture method, and the total number is calculated after recapture after 10 days;

[0024] Cave related index data: orthographic image is generated by low-altitude shooting of the unmanned aerial vehicle, cave entrances are identified by image segmentation, and cave distribution pattern is calculated based on spatial analysis tools;

[0025] Vegetation damage index data: vegetation index change is extracted from multispectral satellite image acquired by the unmanned aerial vehicle, and cover reduction amount is calculated;

[0026] S202, the population quantity index data, the cave related index data, the vegetation damage index data and the soil physical index data are standardized and unified to the interval [0, 1], and weights are given.

[0027] Further, a hazard evaluation comprehensive evaluation model is established, including the following:

[0028] S300, the hazard evaluation comprehensive evaluation model includes an input layer, a processing layer and an output layer;

[0029] Input layer: after standardization processing of the population quantity index data, the cave related index data, the vegetation damage index data and the soil physical index data, the sample data are taken as input;

[0030] S301, processing layer: the input sample data are acquired, the sample data are fused and evaluated, and multiple linear regression analysis is used to calculate the mouse damage evaluation degree value P, the calculation formula is:

[0031]

[0032] In the above formula, represents the standardized value of the i-th index, C is a constant term set, W I is the regression weight of the i-th index, I={1, 2, …, N};

[0033] S302, output layer: the rodent index calculated by the processing layer is obtained as the output value of the output layer.

[0034] Further, a remote sensing ecological index satellite-based library for real-time analysis of the state of vegetation is set, specifically including the following:

[0035] S400, by acquiring high-resolution satellite image data and infrared data of the vegetation, and pre-processing the image data;

[0036] S401, based on the satellite image data and the infrared data, extracting the band reflectivity reflecting the ecological quality, and obtaining the vegetation index, the enhanced vegetation index and the dryness index according to different bands;

[0037] S402, after standardizing the vegetation index, the enhanced vegetation index and the dryness index, multiplying the standardized components by the comprehensive weight and summing, to obtain the remote sensing ecological index value.

[0038] Further, the rodent damage evaluation degree value of the real-time monitoring area is generated, specifically including the following:

[0039] S500, acquiring the data of the real-time monitoring area, generating the monitoring area data value of each sub-area according to the divided sub-area, to input the value into the hazard evaluation comprehensive evaluation model for evaluation;

[0040] S501, the hazard evaluation comprehensive evaluation model generates the rodent damage evaluation degree value of the real-time monitoring area after evaluating and analyzing the input monitoring area data value of each sub-area;

[0041] S502, the rodent damage evaluation degree value of the real-time monitoring area of the sub-area is counted to obtain the total rodent damage evaluation degree value of the real-time monitoring area of the sub-area.

[0042] Further, the regional rodent damage spatial distribution map is drawn, specifically including the following:

[0043] S600, acquiring the real-time rodent damage evaluation value and the corresponding longitude and latitude information, and extracting the vegetation information in the remote sensing image, obtaining the remote sensing ecological index in the satellite-based library according to the remote sensing ecological index satellite-based library;

[0044] S600, according to the linear relationship between the remote sensing ecological index and the rodent damage evaluation degree value, a linear regression model is established, the linear regression model is applied to the entire monitoring area, and a predicted rodent damage index estimation value of each sub-area is generated,

[0045] P=K*R+B Y ;

[0046] In the above formula, K and B represent the slope and intercept of the linear regression, respectively representing the average change amount of the rodent damage index when the RSEI increases by 1 unit, and the rodent damage index reference value when the RSEI is 0;

[0047] S601, the predicted rodent damage index estimation value is associated with the corresponding coordinate point in the remote sensing image, a total monitoring area data set is generated, and is converted into an intuitive spatial distribution map.

[0048] Further, it also includes a regional rodent damage risk grade division mechanism, specifically including the following:

[0049] Each sub-area is divided into a regular grid, and each grid center point represents the ecological characteristics of the area, and a corresponding grid center point coordinate is generated;

[0050] The distance between a grid center point and the nearest point in the actual sampling point is calculated, the inverse model of the sample point corresponding position is extended to each grid by using the method of minimum distance spatial interpolation, the rodent damage degree of the nearest distance grid is calculated, and the rodent damage of the entire monitoring area is estimated, and the calculation process is as follows:

[0051]

[0052] In the above formula: (X W , Y W ) is a rodent damage evaluation model coordinate position variable, (X S , Y S ) is a position variable of the target pixel, when d is the smallest, the rodent damage evaluation degree value predicted by the adjacent rodent damage evaluation model of the target sampling point is used as the target point value;

[0053] The obtained target point value is used to divide the monitoring area into a high incidence area, a normal area and a low incidence area.

[0054] As described above, due to the adoption of the above technical scheme, the present application has the following beneficial effects:

[0055] This computer vision-based method for monitoring Brandt's vole damage involves dividing and marking the monitoring area into managed zones, and using drones and sensors to acquire data from these zones. It gathers rich information from different angles and levels, analyzes data on rodent and vegetation status, extracts comprehensive damage assessment indicators based on the damage monitoring mechanism, and uses a comprehensive damage assessment model to evaluate the damage. This allows for timely understanding of the actual situation of Brandt's vole damage within the monitoring area. A remote sensing ecological index database is established specifically for real-time vegetation status analysis. Correlation analysis of rodent damage assessment values ​​yields an objective function model between ecological values ​​and predicted rodent damage assessments. This method fully considers the interrelationship between vegetation and rodents in the ecosystem, uses ecological data to predict rodent damage, increases the accuracy of rodent damage assessment, and generates a spatial distribution map of rodent damage in the region, visually displaying the distribution of rodent damage in different areas and improving the efficiency and targeting of control efforts. Attached Figure Description

[0056] Figure 1 A schematic diagram of the method flow structure of the present invention is shown. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] Example 1: As Figure 1 As shown, a method for monitoring Brandt's vole damage using computer vision includes the following steps:

[0059] Step 1: Obtain the monitoring area for Brandt's vole damage and divide it into management areas. Construct a map of the divided monitoring areas and label each area to obtain a network map of distinguishing monitoring areas. Use drones and sensing devices to acquire data within the monitoring areas and form a dataset of the monitoring areas.

[0060] Step 2: Obtain the dataset of the monitoring area, analyze the data on rodent damage and vegetation status in the monitoring area, extract the comprehensive evaluation index of the damage according to the pre-established hazard monitoring mechanism, generate the comprehensive rodent damage dataset and the vegetation ecology dataset, establish a comprehensive hazard assessment model based on the comprehensive rodent damage dataset, conduct dynamic evaluation of the divided monitoring area, output the hazard analysis index of the monitoring area, and construct a remote sensing ecological index satellite-based library for real-time vegetation status analysis.

[0061] Step three: obtain the data of the real-time monitoring area, and evaluate according to the hazard evaluation comprehensive evaluation model to generate the mouse damage evaluation degree value of the real-time monitoring area;

[0062] Step four: based on the remote sensing ecological index star-based library, the ecological data of the monitoring area obtained in real time is judged to generate the star-based vegetation ecological value of the regional point, and the correlation coefficient analysis is carried out combined with the mouse damage evaluation degree value, and the target function model between the ecological value and the predicted mouse damage evaluation is obtained;

[0063] Step five: obtain the target function model between the ecological value and the predicted mouse damage evaluation of the monitoring area, generate the total monitoring area data set, and draw the regional mouse damage spatial distribution map according to the total monitoring area data set.

[0064] The data set of the monitoring area is formed, which specifically includes the following:

[0065] S100, obtain the initial area of the monitoring area, and divide the monitoring area into N sub-regions, N={1, 2, …, N}, record the initial division result, form a basic division document containing regional boundary coordinates, area, and main ecological characteristics, and collect visible light, infrared, multispectral images and environmental data through multi-modal sensors, and pre-process and align the data;

[0066] S101, design regional marker: assign a unique code to each sub-region, and mark key management information including regional responsible person and the latest mouse damage investigation time;

[0067] Overlay the marker network layer on the monitoring area map to form a "differential marker network map";

[0068] S102, based on the initial division result, integrate high-precision remote sensing images and geographic information data of unmanned aerial vehicle shooting, and adjust the sub-region boundary;

[0069] Mark the preset point of the unmanned aerial vehicle monitoring device, the deployment position of the sensor device, and the cruise path of the unmanned aerial vehicle shooting, and input them into the internal of the monitoring area map;

[0070] S103, when the unmanned aerial vehicle cruises regularly, obtain high-resolution images including mouse holes and activity tracks, and combine thermal infrared technology to detect night activity data;

[0071] The sensor device will obtain ground data through the soil moisture sensor and infrared camera;

[0072] Integrate and pre-process the ground data and the data obtained by the unmanned aerial vehicle to obtain the data set of the monitoring area, and structure the data according to the region, time and data type, synchronize the data of the unmanned aerial vehicle and the sensor device according to the set update time, and generate a visual report.

[0073] According to the pre-established hazard monitoring mechanism, specifically including the following:

[0074] S200, based on the ecological hazard characteristics of Brandt's vole, a four-dimensional index system is constructed, including population quantity index, cave related index and vegetation damage index;

[0075] S201, population quantity index data: through unmanned aerial vehicle aerial image intelligent recognition automatic counting ground active individual; Marked recapture method marks 50 to 100 individuals in typical sample area, recaptures the total number after 10 days;

[0076] Cave related index data: unmanned aerial vehicle low-altitude photography generates orthophoto map, identifies cave entrance through image segmentation, and calculates cave distribution pattern based on spatial analysis tool;

[0077] Vegetation damage index data: vegetation index change is extracted from multispectral satellite image obtained by unmanned aerial vehicle, and cover reduction amount is calculated;

[0078] S202, the population quantity index data, the cave related index data, the vegetation damage index data and the soil physical index data are standardized, unified to the interval [0, 1], and the weight is given.

[0079] Establish a hazard evaluation comprehensive evaluation model, specifically including the following:

[0080] S300, the hazard evaluation comprehensive evaluation model includes input layer, processing layer and output layer;

[0081] Input layer: obtain population quantity index data, cave related index data, vegetation damage index data and soil physical index data, and standardize the data as input sample data after standardization;

[0082] S301, processing layer: obtain the input sample data, fuse the sample data, and calculate the mouse damage evaluation degree value P by using multiple linear regression analysis, the calculation formula is:

[0083]

[0084] In the above formula, The standardized value of the ith index is represented, C is a constant term, W I The regression weight of the first index is represented, I={1, 2, …, N};

[0085] S302, output layer: obtain the mouse damage index calculated by the processing layer as the output value of the output layer.

[0086] Set the remote sensing ecological index star-based library for real-time analysis of vegetation state, specifically including the following:

[0087] S400, high-resolution satellite image data and infrared data of the vegetation are acquired, and the image data is preprocessed;

[0088] S401, based on the satellite image data and the infrared data, band reflectivity reflecting ecological quality is extracted, and vegetation index, enhanced vegetation index and aridity index are obtained according to different bands;

[0089] S402, after the vegetation index, the enhanced vegetation index and the aridity index are standardized, the standardized components are multiplied by the comprehensive weight and summed up to obtain a remote sensing ecological index value.

[0090] The mouse damage evaluation degree value of the real-time monitoring area is generated, specifically including the following:

[0091] S500, the data of the real-time monitoring area is acquired, the monitoring area data value of each sub-area is generated according to the divided sub-area, and the input value is evaluated by a hazard evaluation comprehensive evaluation model;

[0092] S501, the hazard evaluation comprehensive evaluation model evaluates and analyzes the input monitoring area data value of each sub-area, and generates a mouse damage evaluation degree value of the real-time monitoring area;

[0093] S502, the mouse damage evaluation degree value of the real-time monitoring area of the sub-area is counted to obtain the total mouse damage evaluation degree value of the real-time monitoring area of the sub-area.

[0094] Draw a regional mouse hazard spatial distribution map, specifically including the following:

[0095] S600, the real-time mouse damage evaluation value and the corresponding latitude and longitude information are acquired, the vegetation information in the remote sensing image is extracted, and the remote sensing ecological index in the star-based library is obtained according to the remote sensing ecological index star-based library;

[0096] S600, a linear regression model is established according to the linear relationship between the remote sensing ecological index and the mouse damage evaluation degree value, the linear regression model is applied to the entire monitoring area, and a predicted mouse damage index estimation value of each sub-area is generated,

[0097] P = K·R Y +B;

[0098] In the above formula, K and B represent the slope and intercept of linear regression, respectively representing the average change amount of mouse index when RSEI increases by 1 unit, and the mouse index reference value when RSEI is 0;

[0099] S601, the predicted mouse damage index estimation value is associated with the corresponding coordinate point in the remote sensing image to generate a total monitoring area data set, and is converted into an intuitive spatial distribution map.

[0100] Further comprising a regional rodent risk level classification mechanism, specifically comprising the following:

[0101] Divide each sub-region into a regular grid, and each grid center point represents the ecological characteristics of the region, and the corresponding grid center point coordinates are generated;

[0102] Calculate the distance between a grid center point and the nearest point in the actual sampling point, and use the minimum distance spatial interpolation method to extend the inversion model of the sample point corresponding position to each grid, calculate the rodent degree of the nearest distance grid, and then estimate the rodent of the whole monitoring area, the calculation process is as follows:

[0103]

[0104] In the above formula: (X W , Y W ) is the rodent evaluation model coordinate position variable, (X S , Y S ) is the position variable of the target pixel, when dmin, use the rodent evaluation degree value predicted by the adjacent rodent evaluation model of the target sampling point as the target point value;

[0105] Divide the monitoring area into high incidence area, ordinary area and low incidence area through the obtained target point value.

[0106] The setting of the size of the interval and the threshold value is for easy comparison, and the size of the threshold value depends on the number of sample data and the base number set by the person skilled in the art for each group of sample data; as long as it does not affect the proportional relationship of the parameters and the quantized values.

[0107] The above formulas are all dimensionless values, and the formulas are obtained by collecting a large amount of data to simulate the nearest real situation, and the preset parameters in the formula are set by the person skilled in the art according to the actual situation;

[0108] In the two embodiments provided in the present application, it should be understood that the disclosed device and system can be implemented in other ways; for example, the device embodiments described above are only schematic, for example, the division of the modules is only a logical function division, and actual implementation can have another division manner, for example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed; another point, the coupling or direct coupling or communication connection between the displayed or discussed each other can be through some interface, indirect coupling or communication connection between the devices or modules, which can be electrical, mechanical or other forms;

[0109] The above merely describes preferred specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art, according to the technical solution and inventive concept of the present application, makes equivalent replacement or change within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A method for monitoring damage by Peromyscus maniculatus using computer vision, the method comprising: Comprising the following steps: Step one: obtain the monitoring area of prairie vole damage to carry out management area division, construct the divided monitoring area map, and mark each area, obtain the division mark net map of the monitoring area, obtain the data in the monitoring area through the unmanned aerial vehicle and the sensing device, form the data set of the monitoring area; Step two: obtain the data set of the monitoring area, analyze the data of the monitoring area, and generate the comprehensive data set of the monitoring area and the ecological data set of the vegetation, based on the comprehensive evaluation model of the hazard evaluation, the dynamic evaluation of the divided monitoring area is carried out, and the hazard analysis index of the monitoring area is output, and the remote sensing ecological index star-based library for real-time analysis of the state of the vegetation is constructed; Step three: obtain the real-time monitoring area data, and evaluate according to the hazard evaluation comprehensive evaluation model to generate the real-time monitoring area mouse damage evaluation degree value; Step four: based on the remote sensing ecological index star-based library, the ecological data of the monitoring area is obtained in real time, the star-based vegetation ecological value of the region point is generated, and the correlation coefficient analysis is carried out combined with the mouse damage evaluation degree value, to obtain the target function model between the ecological value and the predicted mouse damage evaluation; Step five: obtain the target function model between the ecological value and the predicted mouse damage evaluation of the monitoring area, generate the total monitoring area data set, and draw the regional mouse damage spatial distribution map according to the total monitoring area data set.

2. The method for monitoring damage by Orymolops daurica using computer vision according to claim 1, characterized in that, The data set of the monitoring area is formed, which specifically includes the following: S100, obtain the initial area of the monitoring area, and divide the monitoring area into N sub-regions, N={1, 2, …, N}, record the initial division result, form a basic division document containing region boundary coordinates, area, and main ecological characteristics, and collect visible light, infrared, multispectral images and environmental data through multi-modal sensors, and pre-process and align the data; S101, design region marking: assign a unique code to each sub-region, and mark key management information including region responsible person and the latest mouse damage investigation time; Overlay the marking net layer on the monitoring area map to form a "division mark net map"; S102, based on the initial division result, integrate high-precision remote sensing images shot by unmanned aerial vehicles and geographic information data to adjust the sub-region boundary; Mark the preset point of the unmanned aerial vehicle monitoring device, the deployment position of the sensor device, and the cruise path of the unmanned aerial vehicle shooting, and input to the inside of the monitoring area map; S103, when the unmanned aerial vehicle cruises regularly, high-resolution images including mouse holes and activity tracks are obtained, combined with thermal infrared technology to detect night activity data; The sensor device will obtain ground data through soil moisture sensor and infrared camera; Integrate and pre-process the ground data and the data obtained by the unmanned aerial vehicle to obtain the data set of the monitoring area, and structure the data according to the region, time and data type, synchronize the data of the unmanned aerial vehicle and the sensing device according to the set update time, and generate a visual report.

3. The method of prairie vole infestation monitoring using computer vision of claim 1, wherein, According to the pre-established hazard monitoring mechanism, specifically including the following: S200, based on the ecological damage characteristics of Brandt's vole, a four-dimensional index system is constructed, including population quantity index, cave related index, vegetation damage index; S201, population quantity index data: through unmanned aerial vehicle aerial image intelligent recognition automatic counting ground active individuals; mark recapture method marks 50 to 100 individuals in a typical sample area, and recaptures the total number after 10 days; Cave related index data: unmanned aerial vehicle low-altitude photography generates orthophoto map, identifies cave entrance through image segmentation, and calculates cave distribution pattern based on spatial analysis tool; Vegetation damage index data: through the multispectral sensor loaded on the unmanned aerial vehicle, multispectral satellite image is obtained, and vegetation index change is extracted to calculate the reduction amount of coverage; S202, standardize population quantity index data, cave related index data, vegetation damage index data and soil physical index data, unify them to [0, 1] interval, and give weight.

4. The method for monitoring damage by Orymolops daurica using computer vision according to claim 1, characterized in that, A hazard evaluation comprehensive evaluation model is established, which specifically includes the following: S300, the hazard evaluation comprehensive evaluation model includes input layer, processing layer and output layer; Input layer: obtain population quantity index data, cave related index data, vegetation damage index data and soil physical index data, and standardize them as sample data for input; S301, processing layer: obtain the sample data, fuse and evaluate the sample data, and calculate the mouse damage evaluation degree value P by using multiple linear regression analysis, the calculation formula is: In the above formula, represents the standardized value of the i-th index, C is a constant term set, W I is the regression weight of the i-th index, I = {1, 2, …, N}; S302, output layer: obtain the mouse damage index calculated by the processing layer as the output value of the output layer.

5. The method for monitoring damage by Orymolops daurica using computer vision according to claim 1, characterized in that, A remote sensing ecological index satellite-based library for real-time analysis of vegetation state is set, which specifically includes the following: S400, obtain high-resolution satellite image data and infrared data of the vegetation, and pretreat the image data; S401, based on satellite image data and infrared data, extract band reflectivity reflecting ecological quality, and obtain vegetation index, enhanced vegetation index and dryness index according to different bands; S402, after standardizing the vegetation index, enhanced vegetation index and dryness index, multiply the standardized components by the comprehensive weight and sum them up to obtain the remote sensing ecological index value.

6. The method for monitoring damage by Orymolops daurica using computer vision according to claim 1, characterized in that, Generate the mouse damage evaluation degree value of the real-time monitoring area, which specifically includes the following: S500, obtain the data of the real-time monitoring area, generate the monitoring area data value of each subarea according to the divided subareas, and input the value into the hazard evaluation comprehensive evaluation model for evaluation; S501, the hazard evaluation comprehensive evaluation model evaluates and analyzes the input monitoring area data value of each subarea, and generates the mouse damage evaluation degree value of the real-time monitoring area; S502, statistics the mouse damage evaluation degree value of the real-time monitoring area of the subarea to obtain the total mouse damage evaluation degree value of the real-time monitoring area of the subarea.

7. The method of prairie vole infestation monitoring using computer vision of claim 1, wherein, Draw the regional mouse damage spatial distribution map, which specifically includes the following: S600, obtain the real-time mouse damage evaluation value and the corresponding latitude and longitude information, extract the vegetation information in the remote sensing image, and obtain the remote sensing ecological index in the satellite-based library according to the remote sensing ecological index satellite-based library; S600, according to the linear relationship between the remote sensing ecological index and the mouse damage evaluation degree value, a linear regression model is established, and the linear regression model is applied to the entire monitoring area to generate a predicted mouse damage index estimation value for each sub-region, P = K - R Y + B; In the above formula, K and B represent the slope and intercept of the linear regression, respectively representing the average change in the mouse damage index when the RSEI increases by 1 unit, and the baseline value of the mouse damage index when the RSEI is 0; S601, associate the predicted mouse damage index estimation value with the corresponding coordinate point in the remote sensing image, generate a total monitoring area data set, and convert it into an intuitive spatial distribution map.

8. The method for monitoring damage by Ormylinus torquatus using computer vision as claimed in claim 1, wherein, It also includes a regional mouse damage risk level classification mechanism, which specifically includes the following: Divide each sub-region into regular grids, and each grid center point represents the ecological characteristics of the region, and generate the corresponding grid center point coordinates; Calculate the distance between a grid center point and the nearest point in the actual sampling points, and use the minimum distance spatial interpolation method to extend the inversion model of the sample point corresponding position to each grid, calculate the mouse damage degree of the nearest distance grid, and then estimate the mouse damage of the entire monitoring area, the calculation process is as follows: In the above formula: (X W , Y W ) is a mouse evaluation model coordinate position variable, (X S , Y S ) is a position variable of the target pixel, and when dmin is the smallest, the mouse evaluation degree value predicted by the mouse evaluation model adjacent to the target sampling point is used as the target point value. Divide the monitoring area into high-risk areas, ordinary areas and low-risk areas through the obtained target point values.

Images