Forest grass ecological environment monitoring system and method

Abstract

The invention provides a forest grass ecological environment monitoring system and method, and the method comprises the steps: extracting dynamic significance features from a multi-temporal remote sensing image, and generating a spatial-temporal change diagram of forest grass ecological elements according to the dynamic significance features; screening out a plurality of ecological response characteristics reflecting the target forest grass ecological region according to the spatial-temporal change diagram, and further converting all the ecological response characteristics into image quantitative indexes of the forest grass ecological state; determining the dynamic influence quantity of the environmental factors on the forest grass ecological state in the forest grass ecological environment based on the incidence relation between the environmental factors and the vegetation characteristics in the multi-modal environmental data; and carrying out evolution prediction on the forest grass ecological state in the target forest grass ecological region through the image quantitative index and the dynamic influence quantity to obtain an ecological environment evolution monitoring result of the target forest grass ecological region. By adopting the scheme of the invention, the evolution prediction of the forest grass ecological state can be realized in a complex and changeable forest grass ecological environment, so that the monitoring capability of the forest grass ecological state in a large range is improved.

Description

Technical Field

[0001] This application relates to the field of ecological environment monitoring technology, and more specifically, to a forest and grassland ecological environment monitoring system and method. Background Technology

[0002] Forest and grassland ecological environment monitoring refers to the process of continuously observing and evaluating the structure, function and dynamic changes of ecosystems such as forests and grasslands using technologies such as remote sensing and ground observation. Its purpose is to accurately grasp information such as vegetation cover, growth status and degradation trends, so as to provide a scientific basis for ecological restoration, disaster early warning and the formulation of sustainable development policies.

[0003] However, existing methods for monitoring forest and grassland ecological environments mainly rely on periodic manual ground surveys or single remote sensing vegetation index analyses. While these methods can provide some information on ecological status, they have significant limitations in practical applications. On the one hand, traditional ground surveys are time-consuming, labor-intensive, and costly, and it is difficult to achieve large-scale, high-frequency continuous monitoring, resulting in data discontinuity in the spatiotemporal dimensions and an inability to effectively capture gradual or sudden changes in the ecosystem. On the other hand, existing remote sensing-based technologies often focus on static or short-term vegetation cover analysis, frequently viewing remote sensing image features in isolation and failing to deeply integrate and quantify the dynamic trends of vegetation changes with key environmental drivers (such as surface temperature and soil moisture). This makes it difficult for existing methods to accurately distinguish between natural fluctuations in vegetation change and degradation caused by human activities or climate, and to quantify the specific impact of environmental pressures on the ecological status. Consequently, the diagnosis of forest and grassland ecological status is one-sided, and the accuracy of evolution prediction is insufficient, ultimately affecting the timeliness and scientific rigor of ecological protection and restoration decisions. Therefore, how to predict the evolution of forest and grassland ecological status in a complex and ever-changing forest and grassland ecological environment, so as to improve the monitoring capability of large-scale forest and grassland ecological status, has become a challenge for the industry. Summary of the Invention

[0004] This application provides a forest and grassland ecological environment monitoring system and method, which can predict the evolution of forest and grassland ecological status in complex and ever-changing forest and grassland ecological environments.

[0005] Firstly, this application provides a method for analyzing the evolution of forest and grassland ecological environment based on image recognition, which is applied to a forest and grassland ecological environment monitoring system. The method includes the following steps: Acquire multi-temporal remote sensing images of the target forest and grassland ecological area; Dynamic saliency features characterizing forest and grassland vegetation degradation and growth are extracted from the multi-temporal remote sensing images, and spatiotemporal variation maps of forest and grassland ecological elements are generated based on the dynamic saliency features. Based on the trend changes in the spatiotemporal change map, multiple ecological response features reflecting the local degradation, rapid growth and thinning trends of the target forest and grassland ecological area are selected, and then all ecological response features are converted into image quantitative indicators of forest and grassland ecological status. Multimodal environmental data including surface temperature, soil moisture and regional vegetation coverage are collected, and the dynamic impact of environmental factors on the forest and grassland ecological status is determined based on the correlation between environmental factors and vegetation characteristics in the multimodal environmental data. The evolution of the forest and grassland ecological status in the target forest and grassland ecological area is predicted by the image quantification index and the dynamic influence quantity, and the monitoring results of the ecological environment evolution of the target forest and grassland ecological area are obtained.

[0006] In conjunction with the first aspect, in one possible implementation, extracting dynamic saliency features characterizing forest and grassland vegetation degradation and growth from the multi-temporal remote sensing images specifically includes: Determine the temporal vegetation index of each pixel in the multi-temporal remote sensing image; Trend analysis and significance tests were performed on the temporal vegetation index of each pixel to obtain the slope value and significance level value of the change for each pixel. Dynamic significance features characterizing forest and grassland vegetation degradation and growth were obtained by screening based on the slope values ​​and significance levels of all pixels.

[0007] In conjunction with the first aspect, in one possible implementation, generating a spatiotemporal variation map of forest and grassland ecological elements based on the aforementioned dynamic saliency characteristics specifically includes: The significant degradation features and significant growth features in the dynamic saliency features are spatially fused to obtain a fused layer; The fused layer is color-coded and visualized to generate a spatiotemporal variation map of forest and grassland ecological elements.

[0008] In conjunction with the first aspect, in one possible implementation, the selection of multiple ecological response features reflecting the local degradation, rapid growth, and thinning trends of the target forest and grassland ecological area based on trend changes in the spatiotemporal change map specifically includes: Obtain the fused layer and the original temporal vegetation index corresponding to the spatiotemporal change map; Based on the trend changes in the fused layer and the original time-series vegetation index, multi-condition threshold filtering is performed to output multiple ecological response features that reflect the local degradation, rapid growth and thinning trends of the target forest and grassland ecological area.

[0009] In conjunction with the first aspect, one possible implementation method for converting all ecological response features into image-based quantitative indicators of forest and grassland ecological status specifically includes: Determine the pixel area corresponding to each ecological response characteristic; Image quantification indicators of forest and grassland ecological status are determined based on the proportion of each pixel area in the target forest and grassland ecological region.

[0010] In conjunction with the first aspect, in one possible implementation, determining the dynamic impact of environmental factors on the forest and grassland ecological state based on the correlation between environmental factors and vegetation characteristics in the multimodal environmental data specifically includes: Spatially register the multimodal environmental data with the vegetation index change slope value of each pixel to construct an environment-change associated sample set; Using the aforementioned environment-change associated sample set as input features and the corresponding vegetation index change slope as the target variable, a random forest regression model is trained. The dynamic impact of environmental factors on the ecological state of forests and grasslands is determined based on the characteristic importance scores of each environmental factor in the random forest regression model.

[0011] In conjunction with the first aspect, in one possible implementation, the evolution prediction of the forest and grassland ecological state in the target forest and grassland ecological area is obtained by using the image quantification index and the dynamic influence quantity, specifically including: Construct a prediction dataset that includes a sequence of historical image quantification indicators and dynamic impact quantities; A prediction model for forest and grassland ecological status based on time series analysis was established based on the aforementioned prediction dataset. The ecological status of forests and grasslands in the target forest and grassland ecological region is predicted by the prediction model, the image quantification index, and the dynamic impact quantity, thereby obtaining the monitoring results of the ecological environment evolution of the target forest and grassland ecological region.

[0012] Secondly, this application provides a forest and grassland ecological environment monitoring system, including a forest and grassland ecological environment evolution analysis unit, wherein the forest and grassland ecological environment evolution analysis unit includes: The acquisition module is used to acquire multi-temporal remote sensing images of the target forest and grassland ecological area; The processing module is used to extract dynamic saliency features that characterize the degradation and growth of forest and grassland vegetation from the multi-temporal remote sensing images, and generate a spatiotemporal variation map of forest and grassland ecological elements based on the dynamic saliency features. The processing module is also used to filter out multiple ecological response features that reflect the local degradation, rapid growth and thinning trend of the target forest and grassland ecological area based on the trend changes in the spatiotemporal change map, and then convert all ecological response features into image quantitative indicators of forest and grassland ecological status. The processing module is also used to collect multimodal environmental data including surface temperature, soil moisture and regional vegetation coverage, and to determine the dynamic impact of environmental factors on the forest and grassland ecological status based on the correlation between environmental factors and vegetation characteristics in the multimodal environmental data. The execution module is used to predict the evolution of the forest and grassland ecological status in the target forest and grassland ecological area through the image quantification index and the dynamic influence quantity, and obtain the monitoring results of the ecological environment evolution of the target forest and grassland ecological area.

[0013] Thirdly, this application provides a computer device, which includes a memory and a processor. The memory is used to store a computer program, and the processor is used to call and run the computer program from the memory, so that the computer device performs the above-described method for analyzing the evolution of forest and grassland ecological environment based on image recognition.

[0014] Fourthly, this application provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the above-mentioned image recognition-based forest and grassland ecological environment evolution analysis method.

[0015] The technical solution provided in this application has the following beneficial effects: In this application, multi-temporal remote sensing images of a target forest and grassland ecological region are acquired; dynamic saliency features characterizing forest and grassland vegetation degradation and growth are extracted from the multi-temporal remote sensing images, and a spatiotemporal variation map of forest and grassland ecological elements is generated based on the dynamic saliency features; multiple ecological response features reflecting the trends of local degradation, rapid growth, and thinning in the target forest and grassland ecological region are selected based on the trend changes in the spatiotemporal variation map, and then all ecological response features are converted into image quantification indicators of forest and grassland ecological status; multimodal environmental data including surface temperature, soil moisture, and regional vegetation coverage are collected, and the dynamic influence of environmental factors on forest and grassland ecological status is determined based on the correlation between environmental factors and vegetation characteristics in the multimodal environmental data; the evolution prediction of forest and grassland ecological status in the target forest and grassland ecological region is performed through the image quantification indicators and the dynamic influence, and the monitoring results of ecological environment evolution of the target forest and grassland ecological region are obtained.

[0016] Therefore, in this application, firstly, dynamic saliency features characterizing forest and grassland vegetation degradation and growth are extracted from the multi-temporal remote sensing images. Based on these dynamic saliency features, a spatiotemporal variation map of forest and grassland ecological elements is generated. This allows for the differentiation between rapid fluctuations in short periods and long-term trends, improving the ability to capture dynamic changes in forest and grassland ecological regions and effectively enhancing the precision of large-scale ecological state perception. Secondly, converting all ecological response features into image-quantitative indicators of forest and grassland ecological state can identify key drivers of vegetation change that represent the internal vegetation state of a region. This effectively compresses complex spatiotemporal information, enabling standardized presentation of state changes in different regions and time periods, thereby improving the comparability of monitoring and automated processing capabilities. Thirdly, based on the correlation between environmental factors and vegetation characteristics in the multimodal environmental data, the dynamic impact of environmental factors on the forest and grassland ecological state is determined, revealing... The external pressures of ecological change provide a basis for determining whether vegetation degradation is caused by extreme environmental changes, enabling a deeper understanding of the causes of ecological state changes in complex environments and improving the interpretability of monitoring results. Finally, by using the image quantification indicators and dynamic impact quantities to predict the evolution of forest and grassland ecological states in the target forest and grassland ecological areas, the monitoring results of the ecological environment evolution of the target forest and grassland ecological areas can be obtained. This can upgrade static monitoring to proactive prediction to generate prediction results that are more consistent with the actual ecological succession laws, enhance the foresight and responsiveness of the monitoring system, reduce monitoring blind spots and lags, and improve the dynamic control capability of large-scale ecological monitoring. This makes the forest and grassland ecological states in complex environments not only visible but also predictable, ultimately promoting a comprehensive improvement in monitoring capabilities. In summary, this scheme can achieve the prediction of the evolution of forest and grassland ecological states in complex and ever-changing forest and grassland ecological environments, thereby improving the monitoring capability of large-scale forest and grassland ecological states. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is an exemplary flowchart of a forest and grassland ecological environment evolution analysis method based on image recognition, according to some embodiments of this application; Figure 2 This is an exemplary flowchart illustrating the extraction of dynamic saliency features according to some embodiments of this application; Figure 3 This is an exemplary flowchart illustrating the generation of a spatiotemporal variation map according to some embodiments of this application; Figure 4This is a schematic diagram of the structure of a forest and grassland ecological environment evolution analysis unit according to some embodiments of this application; Figure 5 This is a schematic diagram of the structure of a computer device for implementing an image recognition-based forest and grassland ecological environment evolution analysis method, according to some embodiments of this application. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] refer to Figure 1 The figure is an exemplary flowchart of an image recognition-based forest and grassland ecological environment evolution analysis method according to some embodiments of this application. The image recognition-based forest and grassland ecological environment evolution analysis method mainly includes the following steps: In step 101, multi-temporal remote sensing images of the target forest and grassland ecological area are acquired.

[0021] In practice, acquiring multi-temporal remote sensing images of the target forest and grassland ecological area can be achieved in the following way: Multiple Earth observation platforms, including my country's High-Resolution Earth Observation System (HREOS) series satellites, resource satellites, and Fengyun meteorological satellites, can be used to periodically image the target forest and grassland ecological area, acquiring a multispectral remote sensing image set with a time span of no less than five years, a temporal resolution of no less than monthly, and a spatial resolution better than 16 meters. Then, the raw image data undergoes standardized preprocessing. First, the original digital quantization values ​​from the satellite payload can be converted into top-atmosphere radiance using appropriate radiometric calibration coefficients. Subsequently, 6S or other methods suitable for the characteristics of domestic satellite sensors are applied. Atmospheric radiative transfer model is used for atmospheric correction to eliminate interference from aerosol scattering and water vapor absorption, and the true surface reflectance data is obtained by inversion. On this basis, the images of each time phase are unified to the same geographic coordinate system by a precise geometric correction algorithm based on control points, and spatial resampling is performed by pixel bilinear interpolation to ensure that all image pixels are aligned. Finally, cloud detection algorithm is used to identify and mark invalid pixels covered by clouds and cloud shadows in each image, generating multi-temporal remote sensing images with clear quality identifiers, unified spatiotemporal references, and consistent radiative physical meaning. Other methods can also be used in other embodiments for determination, which are not limited here.

[0022] It should be noted that the multi-temporal remote sensing images in this application refer to a standardized set of remote sensing images that record the reflectance spectral characteristics of surface vegetation and environmental elements within the target forest and grassland ecological area at different points in time.

[0023] In step 102, dynamic saliency features characterizing the degradation and growth of forest and grassland vegetation are extracted from the multi-temporal remote sensing images, and a spatiotemporal variation map of forest and grassland ecological elements is generated based on the dynamic saliency features.

[0024] In some embodiments, reference Figure 2 As shown, this figure is an exemplary flowchart of extracting dynamic saliency features in some embodiments of this application. In this embodiment, the extraction of dynamic saliency features characterizing forest and grassland vegetation degradation and growth from the multi-temporal remote sensing images can be achieved by the following steps: First, in step 1021, the temporal vegetation index of each pixel in the multi-temporal remote sensing image is determined; Secondly, in step 1022, trend analysis and significance test are performed on the temporal vegetation index of each pixel to obtain the slope value and significance level value of the change corresponding to each pixel; Finally, in step 1023, dynamic significance features characterizing forest and grassland vegetation degradation and growth are obtained by screening based on the slope values ​​and significance levels of all pixels.

[0025] In specific implementation, the temporal vegetation index of each pixel in the multi-temporal remote sensing image can be determined in the following way: based on the multi-temporal remote sensing image, the normalized vegetation index and the enhanced vegetation index can be calculated separately through pixel operations using the surface reflectance data of its near-infrared band and red band. For each scene in the multi-temporal remote sensing image, this calculation is performed pixel by pixel, thereby generating a numerical sequence of equal time intervals for each pixel location, consisting of normalized vegetation index values ​​and enhanced vegetation index values ​​from multiple temporal phases. This sequence is the temporal vegetation index of each pixel. Other methods can also be used to determine it in other embodiments, which are not limited here.

[0026] In specific implementation, trend analysis and significance testing are performed on the temporal vegetation index of each pixel to obtain the slope value and significance level value corresponding to each pixel. This can be achieved in the following way: For the temporal vegetation index of each pixel, the Theil-Sen Median trend estimation method can be used to calculate the interannual slope of the temporal vegetation index over time. This slope value is the slope value corresponding to the pixel. At the same time, the Mann-Kendall nonparametric statistical test method can be used to perform hypothesis testing on whether there is a monotonic trend in the temporal vegetation index. The calculated p value is the significance level value corresponding to the pixel. This process is performed independently on each pixel in the target forest and grassland ecological area. Finally, a pair of quantitative values ​​representing the direction and magnitude of the vegetation index change and the statistical significance of the change trend are output for each pixel: the slope value and the significance level value. Other methods can also be used in other embodiments, which are not limited here.

[0027] In practice, the dynamic significance characteristics representing forest and grassland vegetation degradation and growth can be obtained by screening the slope values ​​and significance levels of all pixels. This can be achieved as follows: First, the slope values ​​and significance levels of all pixels can be used as input. Second, an objective threshold for screening dynamic significance characteristics is determined. This objective threshold can be set, for example, based on historical remote sensing images of the target forest and grassland ecological area and concurrent ecological survey data, selecting areas known to have experienced severe degradation as degradation sample areas and areas known to have good vegetation recovery as growth sample areas. Then, the statistical distribution of the slope values ​​and significance levels of all pixels in the degradation sample areas and growth sample areas is calculated respectively. The 95th percentile of the slope values ​​in the degradation sample areas is used as the degradation slope threshold, and the 5th percentile of the slope values ​​in the growth sample areas is used as the threshold. The quantile is used as the growth slope threshold. Simultaneously, considering the standard significance requirements for statistical hypothesis testing, the significance level threshold can be uniformly set to 0.05. Next, logical judgment and pixel selection are performed based on the objective threshold. For example, pixels that simultaneously meet the conditions of a change slope value less than the degradation slope threshold and a significance level value less than 0.05 are categorized and output as dynamic significance features representing significant degradation of forest and grassland vegetation. Pixels that simultaneously meet the conditions of a change slope value greater than the growth slope threshold and a significance level value less than 0.05 are categorized and output as dynamic significance features representing significant growth of forest and grassland vegetation. Finally, the dynamic significance features are represented at the data level as two binary raster layers composed of corresponding pixel positions. Other methods can also be used in other embodiments, which are not limited here.

[0028] It should be noted that the temporal vegetation index in this application refers to a data indicator showing the continuous changes in the growth status and coverage of forest and grassland vegetation at each pixel location within the target forest and grassland ecological area over time; the vegetation index slope value in this application refers to an indicator that quantifies the overall direction and intensity of the growth status of forest and grassland vegetation at each pixel location within its observation period, and is used to determine whether the vegetation in that pixel area is in a state of degradation, stability, or recovery; the significance level value in this application refers to assessing the vegetation change trend shown at each pixel location, which is not caused by random fluctuations but has statistical significance. Confidence index; The dynamic significance feature in this application refers to the set of pixels in the target forest and grassland ecological area that is statistically significant and clearly defined in intent, including significant degradation features and significant growth features. It is used to extract massive pixel-level data into a spatial distribution map that can directly and clearly indicate the core ecological problems (degradation) and positive processes (growth) in the region. The significant degradation feature refers to the set of pixels in the target forest and grassland ecological area that indicates a substantial decline in vegetation cover and vitality, and the significant growth feature refers to the set of pixels in the target forest and grassland ecological area that indicates a substantial improvement in vegetation cover and vitality.

[0029] In some embodiments, reference Figure 3 As shown, this figure is an exemplary flowchart of generating a spatiotemporal variation map in some embodiments of this application. The generation of a spatiotemporal variation map of forest and grassland ecological elements based on the dynamic saliency features can be achieved by the following steps: The significant degradation features and significant growth features in the dynamic saliency features are spatially fused to obtain a fused layer; The fused layer is color-coded and visualized to generate a spatiotemporal variation map of forest and grassland ecological elements.

[0030] In specific implementation, the significant degradation feature and significant growth feature in the dynamic saliency features are spatially fused to obtain the fused layer. This can be achieved in the following way: two binary raster layers representing the degradation and growth of forest and grassland vegetation respectively in the dynamic saliency features can be used as input data. Using a raster calculator or the layer algebra operation function in a Geographic Information System (GIS) platform, each pixel is assigned a classification code value. Pixels with true values ​​in the significant degradation feature layer are assigned a value of 1, pixels with true values ​​in the significant growth feature layer are assigned a value of 2, and pixels with false values ​​in both layers are assigned a value of 0. By using this coding rule, the two independent binary layers are fused into a three-value raster data layer with a unified classification system, thus completing the spatial layer fusion and obtaining the fused layer. Other methods can also be used in other embodiments, which are not limited here.

[0031] In specific implementation, color encoding and visualization rendering of the fused layer to generate a spatiotemporal variation map of forest and grassland ecological elements can be achieved in the following way: Based on the fused three-value raster data layer, a color mapping table is established for it in the GIS mapping module or programming visualization library. For example, significantly degraded pixels coded as 1 are rendered using red colors, significantly increased pixels coded as 2 are rendered using green colors, and pixels with no significant change coded as 0 are rendered using light gray or transparent colors. At the same time, geographic coordinate system information, scale bar, and legend can be added to the raster layer, and the rendering result is overlaid with the base map data. Finally, a spatiotemporal variation map that can intuitively show the spatial distribution pattern of different significant change types in the target forest and grassland ecological area is output, that is, a spatiotemporal variation map of forest and grassland ecological elements. Other methods can also be used in other embodiments, which are not limited here.

[0032] It should be noted that the fusion layer in this application refers to a spatial data layer with a unified classification code formed by integrating the two types of features: degradation and growth. It is used to provide a data foundation for generating a thematic map that can comprehensively reflect the overall ecological change pattern of the target forest and grassland ecological area. The forest and grassland ecological elements in this application refer to the set of core attributes that can be quantitatively represented by remote sensing imagery and used to describe and indicate the basic status and dynamic changes of the forest and grassland ecosystem. The spatiotemporal change map in this application refers to a thematic geographic map that shows the types of changes and spatial distribution of forest and grassland vegetation in different geographical locations within the target forest and grassland ecological area. It is used to transform the data analysis results into a graphical tool that can be directly used by managers for visual interpretation and spatial decision support.

[0033] In step 103, based on the trend changes in the spatiotemporal change map, multiple ecological response features reflecting the local degradation, rapid growth and thinning trends of the target forest and grassland ecological area are selected, and then all ecological response features are converted into image quantification indicators of forest and grassland ecological status.

[0034] In some embodiments, the following steps can be used to screen out multiple ecological response features reflecting the trends of local degradation, rapid growth, and thinning of the target forest and grassland ecological area based on the trend changes in the spatiotemporal variation map: Obtain the fused layer and the original temporal vegetation index corresponding to the spatiotemporal change map; Based on the trend changes in the fused layer and the original time-series vegetation index, multi-condition threshold filtering is performed to output multiple ecological response features that reflect the local degradation, rapid growth and thinning trends of the target forest and grassland ecological area.

[0035] In specific implementation, obtaining the fused layer corresponding to the spatiotemporal change map and the original time-series vegetation index can be achieved in the following way: calling the fused layer data corresponding to the spatiotemporal change map, which is a three-value raster data layer containing classification coding values; simultaneously, calling the original time-series vegetation index dataset that completely corresponds to the time range of the multi-temporal remote sensing image from the result database output by the time-series vegetation index calculation step, which contains the vegetation index observation values ​​of each pixel in all time phases; in addition, the spatial coordinate system can be used to ensure that the fused layer and the original time-series vegetation index dataset are completely registered in space. Other methods can also be used in other embodiments, which are not limited here.

[0036] In specific implementation, based on the trend changes in the fused layer and the original time-series vegetation index, multi-condition threshold filtering is performed to output multiple ecological response features reflecting the local degradation, rapid growth, and thinning trends of the target forest and grassland ecological area. This can be achieved in the following way: a raster calculator or programming method can be used to spatially overlay the fused layer and the original time-series vegetation index dataset. Pixel-level filtering is achieved by setting multiple sets of logical judgment conditions. For example, pixels with a classification code (i.e., trend change) of 1 in the fused layer are directly identified as ecological response features of local degradation; pixels with a classification code of 2 are directly identified as ecological response features of rapid growth. At the same time, the average vegetation index of each pixel in the complete time series is calculated, and pixels whose average value is lower than the preset thinning trend vegetation index threshold and whose classification code in the fused layer is 0 are identified as ecological response features of thinning trend. Finally, the three screening results are saved as three independent binary raster layers. The set of pixels with true values ​​in each layer is the spatial distribution data of the corresponding type of ecological response characteristics. The vegetation index threshold for the sparsity trend can be set based on the correspondence between remote sensing vegetation index and land cover type, combined with the typical forest and grassland vegetation spectral characteristics of the target forest and grassland ecological area. It is necessary to take into account both domain consensus and regional specificity to ensure that the identification of forest and grassland vegetation sparsity status is both physically meaningful and adaptable to actual application scenarios. For example, based on the land use / cover classification map of the target forest and grassland ecological area, long-term stable bare land and low-cover grassland type areas can be extracted, and the multi-year average value of the normalized vegetation index of pixels in these areas can be calculated. The 75th percentile value of the average value sequence is set as the vegetation index threshold for sparsity judgment. Other methods can also be used to determine it in other embodiments, which are not limited here.

[0037] It should be noted that the ecological response features in this application refer to the set of pixel spaces that can represent a specified ecological process or state type within the target forest and grassland ecological region; among them, the ecological response features of local degradation refer to the set of pixel spaces specifically used to indicate a substantial and continuous decline in vegetation cover and vitality within the target forest and grassland ecological region, which is used to accurately locate and identify the core areas of deteriorating health status in the ecosystem; among them, the ecological response features of rapid growth refer to the set of pixel spaces specifically used to indicate a substantial and continuous improvement in vegetation cover and vitality within the target forest and grassland ecological region, which is used to objectively identify and display areas showing positive trends in ecological restoration or natural succession; among them, the ecological response features of sparseness trend refer to the set of pixel spaces specifically used to indicate that the vegetation cover level within the target forest and grassland ecological region has been low for a long time and has not shown a significant recovery trend, which is used to identify potential risk areas that, although they have not experienced severe degradation, are fragile ecosystems with persistently low cover.

[0038] In some embodiments, converting all ecological response features into image-based quantitative indicators of forest and grassland ecological status can be achieved using the following steps: Determine the pixel area corresponding to each ecological response characteristic; Image quantification indicators of forest and grassland ecological status are determined based on the proportion of each pixel area in the target forest and grassland ecological region.

[0039] In specific implementation, the pixel area corresponding to each ecological response feature can be determined in the following way: obtain binary raster layers corresponding to the three ecological response features representing local degradation, rapid growth, and thinning trends, respectively; then, use the raster attribute statistics function of the GIS platform or the raster statistics function in the programming environment to calculate the total number of pixels with true values ​​in each binary raster layer; then multiply the total number of pixels by the actual surface area of ​​a single pixel, where the actual surface area can be determined according to the spatial resolution of the multi-temporal remote sensing image of the target forest and grassland ecological area, for example, a 30-meter resolution pixel corresponds to 900 square meters. Finally, the pixel area corresponding to each ecological response feature is obtained, namely: the absolute value data of the local degradation area, the rapid growth area, and the thinning area; other methods can also be used in other embodiments, which are not limited here.

[0040] In specific implementation, the image quantification index for determining the forest and grassland ecological status based on the proportion of each pixel area in the target forest and grassland ecological region can be achieved in the following way: First, calculate the total area of ​​the target forest and grassland ecological region. This total area can be obtained by multiplying the total number of effective pixels corresponding to multi-temporal remote sensing images by the actual surface area of ​​a single pixel. Then, divide the pixel area corresponding to each ecological response feature by the total area respectively. Divide the locally degraded area by the total area to obtain the degradation rate, divide the rapidly growing area by the total area to obtain the growth rate, and divide the sparse area by the total area to obtain the sparseness rate. Finally, output these three percentage values ​​as image quantification indicators representing the degree of degradation, improvement, and sparseness of the forest and grassland ecological status. Other methods can also be used in other embodiments, which are not limited here.

[0041] It should be noted that the pixel area in this application refers to converting the spatially distributed set of pixels representing different ecological processes into a physical spatial measure that can be used for regional statistics and quantitative comparison; the image quantification index of forest and grassland ecological status in this application refers to a standardized measure that reflects the health status and changing trend of the forest and grassland ecosystem in the target forest and grassland ecological area.

[0042] In step 104, multimodal environmental data including surface temperature, soil moisture and regional vegetation coverage are collected, and the dynamic impact of environmental factors on the forest and grassland ecological status is determined based on the correlation between environmental factors and vegetation characteristics in the multimodal environmental data.

[0043] In specific implementation, the collection of multimodal environmental data including surface temperature, soil moisture, and regional vegetation coverage can be achieved in the following way: the multimodal environmental data of the target forest and grassland ecological area can be obtained through multi-source remote sensing data products and inversion algorithms. Surface temperature data can be obtained by inversion using split-window algorithm on satellite imagery equipped with thermal infrared sensors, such as Landsat TIRS; soil moisture data can be obtained by surface soil volumetric water content data from microwave remote sensing satellite products such as SMAP / Sentinel-1 joint inversion data or reanalysis data such as ERA5-Land; regional vegetation coverage is calculated using normalized vegetation index based on multi-temporal remote sensing images of the target forest and grassland ecological area through a pixel-based bisection model; then all environmental data are uniformly resampled to the same spatial resolution and geographic coordinate system as the multi-temporal remote sensing images and cropped to the target forest and grassland ecological area to form a spatiotemporally aligned multimodal environmental dataset. Other methods can also be used for collection in other embodiments, and no specific limitation is made here.

[0044] It should be noted that the multimodal environmental data in this application refers to a spatial dataset that characterizes the key physical environmental conditions affecting the growth and evolution of forest and grassland vegetation in the target forest and grassland ecological area; it is used as objective evidence to analyze the coupling relationship between environmental driving forces and vegetation changes.

[0045] In some embodiments, determining the dynamic impact of environmental factors on the ecological state of forest and grassland based on the correlation between environmental factors and vegetation characteristics in the multimodal environmental data can be achieved through the following steps: Spatially register the multimodal environmental data with the vegetation index change slope value of each pixel to construct an environment-change associated sample set; Using the aforementioned environment-change associated sample set as input features and the corresponding vegetation index change slope as the target variable, a random forest regression model is trained. The dynamic impact of environmental factors on the ecological state of forests and grasslands is determined based on the characteristic importance scores of each environmental factor in the random forest regression model.

[0046] In specific implementation, the multimodal environmental data is spatially registered with the vegetation index change slope value of each pixel to construct an environment-change associated sample set. This can be achieved in the following way: First, the vegetation index change slope value of each pixel in the multi-temporal remote sensing image of the target forest and grassland ecological area is obtained; then, the surface temperature, soil moisture, and vegetation cover layers in the multimodal environmental data are uniformly resampled to the same spatial resolution as the vegetation index change slope value; then, the accurate spatial registration of all layers is achieved through GIS spatial analysis tools or the GDAL / rasterio library in Python, ensuring that each pixel location contains complete environmental factor data and the corresponding vegetation index change slope value; finally, all registered layer data are converted into tabular data with pixels as the recording unit, where each row represents the complete environmental factor characteristics of a pixel and the corresponding vegetation change trend, forming an environment-change associated sample set; other methods can also be used in other embodiments, which are not limited here.

[0047] In specific implementation, using the environment-change associated sample set as input features and the corresponding vegetation index change slope value as the target variable, the random forest regression model can be trained in the following way: First, the environment-change associated sample set can be randomly divided into a training subset and a test subset, where the training subset contains 70% of the total sample size and the test subset contains 30%; then, using Python's Scikit-learn machine learning library, using the data from the training subset as input, and the corresponding vegetation index change slope value as the target variable, the search range for the number of decision trees in the random forest regression model is set to [50, 1]. [00, 200], 5-fold cross-validation is used to evaluate the performance of different parameter combinations on the training subset, and the optimal hyperparameter combination is determined based on the minimum mean square error. Then, the optimal parameters are used to train the final model on the complete training subset. Through the collaborative learning of multiple decision trees within the model, the complex nonlinear relationship between environmental factors and vegetation change slope is established, and a reliable prediction model from environmental characteristics to vegetation change trends is established. The model performance is verified on the test subset, requiring that the coefficient of determination between the predicted values ​​and the true values ​​of the test subset is not less than 0.6 to ensure the reliability of the model. Other methods can also be used to determine the parameters in other embodiments, which are not limited here.

[0048] In specific implementation, the dynamic impact of environmental factors on the ecological state of forest and grassland ecosystems based on the feature importance scores of each environmental factor in the random forest regression model can be achieved in the following way: The feature_importance_ attribute is called on the trained random forest regression model to obtain the original feature importance scores of each environmental factor. This score is calculated based on the sum of the impurity reduction brought about by the splitting of all decision tree nodes during model training. Then, all original feature importance scores are transformed to the 0-1 interval using the Min-Max normalization method. Finally, the normalized feature importance scores are used as the dynamic impact of each environmental factor on the ecological state of forest and grassland ecosystems. The closer the score is to 1, the greater the influence of the environmental factor on vegetation change. Other methods can also be used in other embodiments, which are not limited here.

[0049] It should be noted that the environment-change association sample set in this application refers to the basic dataset used to establish the quantitative relationship between environmental driving factors and vegetation dynamics; the random forest regression model in this application refers to a machine learning model used to learn and extract the relative contribution weights of each environmental factor to the vegetation change trend from multi-dimensional environmental data. It can fit the complex nonlinear relationship between environmental factors and vegetation change, and internally calculate the importance of each environmental factor in explaining vegetation change; the dynamic impact quantity in this application refers to the relative degree of influence of environmental factors on the trend of forest and grassland ecological status changes. It is used to reveal the dominant and secondary environmental factors affecting the dynamics of forest and grassland vegetation in the target forest and grassland ecological area.

[0050] In step 105, the evolution prediction of the forest and grassland ecological status in the target forest and grassland ecological area is carried out by the image quantification index and the dynamic influence quantity, so as to obtain the monitoring results of the ecological environment evolution of the target forest and grassland ecological area.

[0051] In some embodiments, the following steps can be used to predict the evolution of the forest and grassland ecological status in the target forest and grassland ecological area using the image quantification index and the dynamic influence quantity to obtain the monitoring results of the ecological environment evolution of the target forest and grassland ecological area: Construct a prediction dataset that includes a sequence of historical image quantification indicators and dynamic impact quantities; A prediction model for forest and grassland ecological status based on time series analysis was established based on the aforementioned prediction dataset. The ecological status of forests and grasslands in the target forest and grassland ecological region is predicted by the prediction model, the image quantification index, and the dynamic impact quantity, thereby obtaining the monitoring results of the ecological environment evolution of the target forest and grassland ecological region.

[0052] In specific implementation, the prediction dataset containing historical image quantitative index sequences and dynamic impact quantities can be constructed in the following way: annual image quantitative indicators of the target forest and grassland ecological area for multiple consecutive years, including degradation rate, growth rate and sparsity rate, can be collected to form time series data. At the same time, the dynamic impact quantities of each environmental factor calculated are used as fixed weight parameters. These data are integrated into a structured dataset with year as time index, containing the values ​​of each quantitative index and the impact weights of environmental factors, which is the prediction dataset. Other methods can also be used in other embodiments, which are not limited here.

[0053] In specific implementation, the prediction model of forest and grassland ecological status based on time series analysis based on the prediction dataset can be implemented in the following way: First, construct a time series prediction dataset with exogenous variables. For example, the dynamic impact of each environmental factor can be multiplied by the observed environmental factor values ​​of its corresponding year to obtain a weighted annual environmental impact sequence, which is then used as an exogenous variable. Next, the Python statsmodels library can be used to model the degradation rate, growth rate, and sparsity rate of the prediction dataset. An augmented Dickey-Fuller test is performed on each time series to determine stationarity, and non-stationary series are differencingd until stationarity is achieved. Then, using the weighted annual environmental impact sequence as an exogenous variable, the optimal (p, d, q) parameter combination is determined through an automatic order determination function to establish a comprehensive prediction model considering the impact of environmental factors, i.e., a prediction model of forest and grassland ecological status. Other methods can also be used in other embodiments, which are not limited here.

[0054] In specific implementation, the ecological status of forest and grassland in the target forest and grassland ecological area is predicted by the prediction model, the image quantification index, and the dynamic impact quantity. The monitoring results of the ecological environment evolution of the target forest and grassland ecological area can be obtained in the following way: the trained prediction model can be used to predict the degradation rate, growth rate, and thinning rate of the target forest and grassland ecological area in the next 1-3 years. At the same time, the environmental factors are ranked according to the magnitude of the dynamic impact quantity, and the two environmental factors with the largest dynamic impact quantity are determined as the dominant impact factors. Finally, an environmental monitoring report containing the predicted value of the future ecological status of the target forest and grassland ecological area, the trend analysis of change, and the description of the dominant impact factors is generated. This is the monitoring result of the ecological environment evolution of the target forest and grassland ecological area. Based on the monitoring results of the ecological environment evolution, ecological protection and restoration strategies can be formulated. For example, key governance can be implemented in areas predicted to be degraded, and targeted control measures can be taken according to the dominant impact factors. Other methods can also be used in other embodiments, which are not limited here.

[0055] It should be noted that the prediction dataset in this application refers to a standardized dataset that supports the prediction and analysis of the evolution of forest and grassland ecological status and integrates the weights of historical status indicators and environmental driving factors; the prediction model of forest and grassland ecological status in this application refers to a computational model used to extrapolate the future development trend of forest and grassland ecosystems in the target forest and grassland ecological region. It generates a quantitative prediction of the future ecological status by quantifying the dynamic relationship between status indicators and environmental factors in the historical sequence.

[0056] In another aspect, in some embodiments, this application provides a forest and grassland ecological environment monitoring system, which includes a forest and grassland ecological environment evolution analysis unit, with reference to... Figure 4The figure is a schematic diagram of the structure of a forest and grassland ecological environment evolution analysis unit according to some embodiments of this application. The forest and grassland ecological environment evolution analysis unit 400 includes: an acquisition module 401, a processing module 402, and an execution module 403, which are described below: The acquisition module 401 in this application is mainly used to acquire multi-temporal remote sensing images of the target forest and grassland ecological area; Processing module 402, in this application, is mainly used to extract dynamic saliency features that characterize the degradation and growth of forest and grassland vegetation from the multi-temporal remote sensing images, and generate a spatiotemporal change map of forest and grassland ecological elements based on the dynamic saliency features. The processing module 402 described in this application is also used to filter out multiple ecological response features that reflect the local degradation, rapid growth and thinning trend of the target forest and grassland ecological area based on the trend changes in the spatiotemporal change map, and then convert all ecological response features into image quantification indicators of forest and grassland ecological status. The processing module 402 described in this application is also used to collect multimodal environmental data including surface temperature, soil moisture and regional vegetation coverage, and to determine the dynamic impact of environmental factors on the forest and grassland ecological status based on the correlation between environmental factors and vegetation characteristics in the multimodal environmental data. The execution module 403 in this application is mainly used to predict the evolution of the forest and grassland ecological status in the target forest and grassland ecological area through the image quantification index and the dynamic influence quantity, so as to obtain the monitoring results of the ecological environment evolution of the target forest and grassland ecological area.

[0057] The foregoing has detailed examples of the forest and grassland ecological environment monitoring system and method provided in the embodiments of this application. It is understood that the corresponding apparatus, in order to achieve the above functions, includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed by hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0058] In some embodiments, this application also provides a computer device, the computer device including a memory and a processor, the memory for storing a computer program, and the processor for calling and running the computer program from the memory, so that the computer device performs the above-described image recognition-based forest and grassland ecological environment evolution analysis method.

[0059] In some embodiments, reference Figure 5 The dashed lines in the figure indicate that the unit or module is optional. This figure is a schematic diagram of the computer device implementing the image recognition-based forestry and grassland ecological environment evolution analysis method of this application. The image recognition-based forestry and grassland ecological environment evolution analysis method in the above embodiments can be... Figure 5 The computer device 500 shown is used to implement this, and the computer device 500 includes at least one processor 501, a memory 502 and at least one communication unit 505. The computer device 500 may be a terminal device, a server or a chip.

[0060] The processor 501 can be a general-purpose processor or a special-purpose processor. For example, the processor 501 can be a central processing unit (CPU). The CPU can be used to control the computer device 500, execute software programs, and process data from the software programs. The computer device 500 may also include a communication unit 505 for inputting (receiving) and outputting (transmitting) signals.

[0061] For example, computer device 500 may be a chip, communication unit 505 may be the input and / or output circuit of the chip, or communication unit 505 may be the communication interface of the chip, and the chip may be a component of terminal device, network device or other device.

[0062] For example, computer device 500 may be a terminal device or a server, and communication unit 505 may be a transceiver of the terminal device or the server, or communication unit 505 may be a transceiver circuit of the terminal device or the server.

[0063] The computer device 500 may include one or more memories 502 storing a program 504. The program 504 can be executed by a processor 501 to generate instructions 503, causing the processor 501 to perform the methods described in the above method embodiments according to the instructions 503. Optionally, the memory 502 may also store data (such as a target audit model). Optionally, the processor 501 may also read data stored in the memory 502, which may be stored at the same storage address as the program 504, or the data may be stored at a different storage address than the program 504.

[0064] The processor 501 and memory 502 can be configured separately or integrated together, for example, integrated on the system-on-chip (SOC) of the terminal device.

[0065] It should be understood that each step of the above method embodiment can be completed by hardware logic circuits or software instructions in the processor 501. The processor 501 can be a CPU, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, such as discrete gates, transistor logic devices, or discrete hardware components.

[0066] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0067] For example, in some embodiments, this application also provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the above-described image recognition-based forest and grassland ecological environment evolution analysis method.

[0068] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0069] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for analyzing the evolution of forest and grassland ecological environment based on image recognition, applied to a forest and grassland ecological environment monitoring system, characterized in that, The method includes the following steps: Acquire multi-temporal remote sensing images of the target forest and grassland ecological area; Dynamic saliency features characterizing forest and grassland vegetation degradation and growth are extracted from the multi-temporal remote sensing images, and spatiotemporal variation maps of forest and grassland ecological elements are generated based on the dynamic saliency features. Based on the trend changes in the spatiotemporal change map, multiple ecological response features reflecting the local degradation, rapid growth and thinning trends of the target forest and grassland ecological area are selected, and then all ecological response features are converted into image quantitative indicators of forest and grassland ecological status. Multimodal environmental data including surface temperature, soil moisture and regional vegetation coverage are collected, and the dynamic impact of environmental factors on the forest and grassland ecological status is determined based on the correlation between environmental factors and vegetation characteristics in the multimodal environmental data. The evolution of the forest and grassland ecological status in the target forest and grassland ecological area is predicted by the image quantification index and the dynamic influence quantity, and the monitoring results of the ecological environment evolution of the target forest and grassland ecological area are obtained.

2. The method as described in claim 1, characterized in that, Extracting dynamic saliency features characterizing forest and grassland vegetation degradation and growth from the multi-temporal remote sensing images specifically includes: Determine the temporal vegetation index of each pixel in the multi-temporal remote sensing image; Trend analysis and significance tests were performed on the temporal vegetation index of each pixel to obtain the slope value and significance level value of the change for each pixel. Dynamic significance features characterizing forest and grassland vegetation degradation and growth were obtained by screening based on the slope values ​​and significance levels of all pixels.

3. The method as described in claim 1, characterized in that, The generation of spatiotemporal variation maps of forest and grassland ecological elements based on the aforementioned dynamic saliency characteristics specifically includes: The significant degradation features and significant growth features in the dynamic saliency features are spatially fused to obtain a fused layer; The fused layer is color-coded and visualized to generate a spatiotemporal variation map of forest and grassland ecological elements.

4. The method as described in claim 1, characterized in that, Based on the trend changes in the spatiotemporal variation map, several ecological response characteristics reflecting the local degradation, rapid growth, and thinning trends of the target forest and grassland ecological area were selected, specifically including: Obtain the fused layer and the original temporal vegetation index corresponding to the spatiotemporal change map; Based on the trend changes in the fused layer and the original time-series vegetation index, multi-condition threshold filtering is performed to output multiple ecological response features that reflect the local degradation, rapid growth and thinning trends of the target forest and grassland ecological area.

5. The method as described in claim 1, characterized in that, The specific image-based quantitative indicators that convert all ecological response characteristics into forest and grassland ecological status include: Determine the pixel area corresponding to each ecological response characteristic; Image quantification indicators of forest and grassland ecological status are determined based on the proportion of each pixel area in the target forest and grassland ecological region.

6. The method as described in claim 1, characterized in that, The dynamic impact of environmental factors on the ecological state of forests and grasslands, determined based on the correlation between environmental factors and vegetation characteristics in the multimodal environmental data, specifically includes: Spatially register the multimodal environmental data with the vegetation index change slope value of each pixel to construct an environment-change associated sample set; Using the aforementioned environment-change associated sample set as input features and the corresponding vegetation index change slope as the target variable, a random forest regression model is trained. The dynamic impact of environmental factors on the ecological state of forests and grasslands is determined based on the characteristic importance scores of each environmental factor in the random forest regression model.

7. The method as described in claim 1, characterized in that, The evolution of the forest and grassland ecological status in the target forest and grassland ecological area is predicted by the image quantification index and the dynamic influence quantity. The specific monitoring results of the ecological environment evolution of the target forest and grassland ecological area include: Construct a prediction dataset that includes a sequence of historical image quantification indicators and dynamic impact quantities; A prediction model for forest and grassland ecological status based on time series analysis was established based on the aforementioned prediction dataset. The ecological status of forests and grasslands in the target forest and grassland ecological region is predicted by the prediction model, the image quantification index, and the dynamic impact quantity, thereby obtaining the monitoring results of the ecological environment evolution of the target forest and grassland ecological region.

8. A forest and grassland ecological environment monitoring system, comprising a forest and grassland ecological environment evolution analysis unit, characterized in that, The forest and grassland ecological environment evolution analysis unit includes: The acquisition module is used to acquire multi-temporal remote sensing images of the target forest and grassland ecological area; The processing module is used to extract dynamic saliency features that characterize the degradation and growth of forest and grassland vegetation from the multi-temporal remote sensing images, and generate a spatiotemporal variation map of forest and grassland ecological elements based on the dynamic saliency features. The processing module is also used to filter out multiple ecological response features that reflect the local degradation, rapid growth and thinning trend of the target forest and grassland ecological area based on the trend changes in the spatiotemporal change map, and then convert all ecological response features into image quantitative indicators of forest and grassland ecological status. The processing module is also used to collect multimodal environmental data including surface temperature, soil moisture and regional vegetation coverage, and to determine the dynamic impact of environmental factors on the forest and grassland ecological status based on the correlation between environmental factors and vegetation characteristics in the multimodal environmental data. The execution module is used to predict the evolution of the forest and grassland ecological status in the target forest and grassland ecological area through the image quantification index and the dynamic influence quantity, and obtain the monitoring results of the ecological environment evolution of the target forest and grassland ecological area.

9. A computer device, characterized in that, The computer device includes a memory and a processor. The memory is used to store computer programs, and the processor is used to call and run the computer programs from the memory, so that the computer device performs the image recognition-based forest and grassland ecological environment evolution analysis method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions or code that, when executed on a computer, cause the computer to implement the image recognition-based forestry and grassland ecological environment evolution analysis method as described in any one of claims 1 to 7.

Images