AI analysis method and system applied to land resource survey

Abstract

This application provides an AI analysis method and system for land resource surveys. This method acquires multi-source remote sensing imagery data of a target area, analyzes its spectral and textural characteristics, generates a comprehensive land characteristic map, utilizes a land classification model to generate an initial land classification map, and combines this with historical data for time-series comparative analysis to identify dynamic areas of land cover change and generate a land resource change report. This system also includes functions such as soil quality assessment, visualization of change areas, real-time monitoring of abnormal areas, land classification model training, and adjustment of the classification model based on user feedback, improving the efficiency and accuracy of land resource surveys.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and more specifically, to an AI analysis method and system for land resource surveys. Background Art

[0002] Land resources are a crucial material foundation for human survival and development. Accurate surveying and monitoring of land resources is crucial for rational land use planning, ecological environmental protection, and sustainable development. Traditional land resource survey methods often rely on manual field surveys, which are not only inefficient but also limited by human, material, and time resources, making it difficult to obtain comprehensive, timely, and accurate land information.

[0003] With the development of remote sensing technology, multi-source remote sensing image data has provided a rich source of information for land resource surveys, but how to efficiently process and analyze these massive data has become a new challenge. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide an AI analysis method and system for land resource survey.

[0005] In conjunction with the first aspect of the present application, there is provided an AI analysis method for land resource survey, which is applied to an AI analysis system for land resource survey. The method includes:

[0006] Acquiring multi-source remote sensing image data of the target area, wherein the multi-source remote sensing image data includes visible light imagery, infrared imagery, and radar imagery, and contains spectral characteristics and texture characteristics related to land cover types;

[0007] Performing spectral feature analysis and texture feature extraction on the multi-source remote sensing image data to generate a comprehensive land feature map of the target area, wherein the comprehensive land feature map is used to characterize the spatial distribution and physical properties of land cover types;

[0008] Perform pixel-by-pixel classification processing on the comprehensive land feature map based on a preset land classification model to obtain an initial land classification map of the target area, wherein each pixel in the initial land classification map is marked with a corresponding land type label;

[0009] Performing a time series comparative analysis on the land type labels in the initial land classification map based on historical land survey data, identifying dynamic areas of land cover change in the target area, and extracting change trajectory information of the dynamic areas;

[0010] A land resource change report is generated based on the position coordinates, area data and change trajectory information of the dynamic area, and the land resource change report includes boundary vector data of the change area and change trend prediction results.

[0011] In a possible implementation of the first aspect, performing spectral feature analysis and texture feature extraction on the multi-source remote sensing image data to generate a comprehensive land feature map of the target area includes:

[0012] Dividing the visible light image into a plurality of spectral bands, calculating a spectral reflectance curve for each spectral band, and extracting peak characteristics and slope characteristics of the spectral reflectance curve;

[0013] Performing thermal radiation intensity analysis on the infrared image to determine thermal radiation difference parameters of different land cover types in the target area;

[0014] Extracting polarization scattering features from the radar image to obtain surface roughness and vegetation height parameters;

[0015] The peak characteristics of the spectral reflectance curve, the thermal radiation difference parameter and the surface roughness parameter are integrated to generate a spatial resolution enhanced version of the comprehensive land feature map.

[0016] In a possible implementation of the first aspect, the method further includes:

[0017] Determining moisture parameters, organic matter content parameters, and pH parameters related to soil quality based on the physical properties of different land types in the comprehensive land characteristic map;

[0018] Matching the humidity parameter, organic matter content parameter, and pH parameter with a preset soil quality assessment standard to generate a soil quality grade distribution map for the target area;

[0019] The soil quality grade distribution map is spatially overlaid with the change areas in the land resource change report to determine the priority areas for land degradation or improvement.

[0020] In a possible implementation of the first aspect, performing a time-series comparative analysis on land type labels in the initial land classification map based on historical land survey data to identify dynamic areas of land cover change in the target area includes:

[0021] Obtaining land classification historical maps of the target area at at least three historical time points, and extracting the land type distribution pattern corresponding to each historical time point;

[0022] Calculating spatial differences between the initial land classification map and the land type distribution patterns at each historical time point to determine transition areas where land cover types undergo transformation;

[0023] The conversion types of the conversion areas are classified to generate a set of change types including forest to farmland, farmland to construction land, and water body to wetland.

[0024] In a possible implementation of the first aspect, after generating the land resource change report based on the location coordinates, area data, and change trajectory information of the dynamic area, the method further includes:

[0025] Marking the geometric shape and center coordinates of the changed area on the digital map of the target area according to the boundary vector data of the changed area;

[0026] Based on the change trend prediction results, calculate the area expansion rate and type conversion probability of the change area within a future preset time period;

[0027] The geometric shape, center coordinates, area expansion rate and type conversion probability are integrated into a visualization layer and superimposed on the interactive interface of the digital map.

[0028] In a possible implementation of the first aspect, the method further includes:

[0029] Receive the latest image data transmitted by the satellite remote sensing system in real time, perform spectral feature analysis and texture feature extraction on the latest image data, and generate a real-time land feature map;

[0030] Performing difference detection between the real-time land feature map and the comprehensive land feature map to identify abnormal areas of sudden land cover changes within the target area;

[0031] According to the geographical location and change type of the abnormal area, a land resource abnormality warning signal is sent to the monitoring terminal, and the warning signal includes the coordinate range of the abnormal area and recommended verification measures.

[0032] In a possible implementation of the first aspect, the land classification model is trained by the following steps:

[0033] Obtain a sample image dataset containing multiple land cover types, and label each sample in the sample image dataset with a true land type label;

[0034] Constructing a deep convolutional neural network, inputting the sample image dataset into the deep convolutional neural network for feature learning, and outputting a predicted land type probability distribution;

[0035] Calculating a cross entropy loss value based on the difference between the predicted land type probability distribution and the true land type label, and optimizing the weight parameters of the deep convolutional neural network using a back propagation algorithm;

[0036] When the cross entropy loss value is lower than a preset threshold, the training is stopped and the optimized deep convolutional neural network is used as the land classification model.

[0037] In a possible implementation of the first aspect, the method further includes:

[0038] receiving land survey feedback information submitted by a user through an interactive terminal, wherein the land survey feedback information includes a suggestion for correcting a type labeling of a specific area in the land resource change report;

[0039] Adjusting the classification weight parameters of the land classification model according to the type labeling correction suggestion, and reclassifying the comprehensive land characteristic map of the target area to generate an updated land classification map;

[0040] The updated land classification map is checked for consistency with historical data to determine the degree to which the correction suggestion optimizes the classification result, and the check result is fed back to the interactive terminal.

[0041] In combination with the second aspect of the present application, an AI analysis system for land resource survey is provided, wherein the AI ​​analysis system for land resource survey includes a machine-readable storage medium and a processor, wherein the machine-readable storage medium stores machine-executable instructions. When the processor executes the machine-executable instructions, the AI ​​analysis system for land resource survey implements the aforementioned AI analysis method for land resource survey.

[0042] In combination with the third aspect of the present application, a computer-readable storage medium is provided, in which computer-executable instructions are stored. When the computer-executable instructions are executed, the aforementioned AI analysis method applied to land resource survey is implemented.

[0043] In combination with any of the above aspects, embodiments of the present application acquire multi-source remote sensing imagery data of a target area, analyze its spectral and textural characteristics to generate a comprehensive land characteristic map, utilize a land classification model to generate an initial land classification map, and then perform a time-series comparative analysis combined with historical data to identify dynamic areas of land cover change and generate a land resource change report. This system also includes functions such as soil quality assessment, visualization of change areas, real-time monitoring of abnormal areas, land classification model training, and adjustment of the classification model based on user feedback, thereby improving the efficiency and accuracy of land resource surveys. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained by combining these drawings without paying any creative work.

[0045] Figure 1 A flowchart of an AI analysis method for land resource survey provided in an embodiment of the present application. DETAILED DESCRIPTION

[0046] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0047] The terms "first," "second," and so on, in the specification and claims of the present invention and the accompanying drawings are used to distinguish between different items, not to describe a specific order. Furthermore, the terms "including," "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, apparatus, product, or end comprising a series of steps or elements is not limited to the listed steps or elements but may optionally include steps or elements not listed therein, or may optionally include other steps or elements inherent to such process, method, product, or end.

[0048] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present invention. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute a separate or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0049] Figure 1 The flowchart of the AI ​​analysis method for land resource survey provided by the embodiment of the present application is shown. It should be understood that in other embodiments, the order of some steps in the AI ​​analysis method for land resource survey of this embodiment can be shared based on actual needs, or some steps can be omitted or maintained. The details of the AI ​​analysis method for land resource survey include:

[0050] Step S110: Acquire multi-source remote sensing image data of the target area, wherein the multi-source remote sensing image data includes visible light images, infrared images, and radar images, and contains spectral features and texture features related to land cover types.

[0051] In a real-world land resource survey scenario, let's assume the target area is a large city and its surrounding areas. This target area encompasses a variety of land cover types, such as urban construction land, surrounding cultivated land, forests, water bodies, and wetlands. To comprehensively capture land information within the area, satellite remote sensing technology can be used to generate multi-source remote sensing imagery data. Visible light imagery can reveal the reflectivity characteristics of surface objects in different visible light bands. For example, green vegetation has a higher reflectivity in the green band, while buildings exhibit distinct reflectivity characteristics compared to vegetation and water bodies in various visible light bands. Infrared imagery primarily reflects surface thermal radiation. Different land cover types exhibit varying thermal radiation intensities due to factors such as their material composition and moisture content. For example, water bodies, with their high heat capacity, appear relatively low in infrared imagery; whereas concrete buildings in cities absorb heat during the day, resulting in higher thermal radiation intensities. Radar imagery acquires surface information by transmitting and receiving radar waves. Its polarization scattering characteristics are closely related to factors such as surface roughness and vegetation height. For example, the polarization scattering characteristics of forested areas, due to the high vegetation height and rough surface, are significantly different from those of flat cultivated areas. These different types of remote sensing imagery collectively contain a rich set of spectral and textural features related to land cover types, providing a sufficient data foundation for subsequent analysis.

[0052] Step S120: performing spectral feature analysis and texture feature extraction on the multi-source remote sensing image data to generate a comprehensive land feature map of the target area, wherein the comprehensive land feature map is used to characterize the spatial distribution and physical properties of land cover types.

[0053] First, the visible light image is divided into multiple spectral bands, the spectral reflectance curve of each spectral band is calculated, and the peak characteristics and slope characteristics of the spectral reflectance curve are extracted. In the target area mentioned above in this embodiment, taking the vegetation-covered area as an example, the spectral reflectance curve has a specific morphology in the blue light band and the red light band of visible light. The reflectance in the blue light band is relatively low, and as the wavelength increases to the green light band, the reflectance reaches a peak, and then decreases again in the red light band. By calculating and analyzing the peak characteristics and slope characteristics of these spectral reflectance curves, different types of vegetation can be distinguished. For example, coniferous forests and broad-leaved forests may have slight differences in the peak and slope of the spectral reflectance curve. For non-vegetated areas, such as buildings and roads, their spectral reflectance curves show different characteristics from vegetation in each band, and their peak and slope can reflect the characteristics of building materials.

[0054] Next, the infrared image is analyzed for thermal radiation intensity to determine the thermal radiation difference parameters for different land cover types within the target area. Taking water bodies and urban areas in the target area as an example, due to their high specific heat capacity, water bodies have relatively stable thermal radiation intensity at different times of the day and relatively low overall thermal radiation intensity. In contrast, urban areas have higher thermal radiation intensity due to heat generated by buildings, roads, and human activities, and experience significant variations between day and night. By analyzing the thermal radiation intensity of infrared images, the thermal radiation difference parameters for different land cover types can be accurately determined, thereby helping to distinguish between different types of land.

[0055] Polarimetric scattering features are then extracted from the radar image to obtain surface roughness and vegetation height parameters. In forested areas of the target region, tall trees increase surface roughness, and the polarimetric scattering features of the radar waves exhibit strong scattering characteristics. In cultivated areas, the surface is relatively flat, and the polarimetric scattering features are weaker. By analyzing the polarimetric scattering features of the radar image, surface roughness and vegetation height parameters can be accurately obtained, which are also important for distinguishing different land cover types.

[0056] Finally, the peak characteristics of the spectral reflectance curve, the thermal radiation difference parameters, and the surface roughness parameters are fused to generate a spatial resolution-enhanced version of the comprehensive land feature map. In this process, the various features extracted from visible light images, infrared images, and radar images are fused so that the comprehensive land feature map can more comprehensively and accurately reflect the spatial distribution and physical properties of land cover types in the target area. For example, in the generated comprehensive land feature map, the boundaries of different land cover types such as urban areas, forest areas, cultivated land areas, and water areas, as well as their respective characteristic distributions, can be clearly seen.

[0057] Step S130: performing pixel-by-pixel classification processing on the comprehensive land feature map based on a preset land classification model to obtain an initial land classification map of the target area, wherein each pixel point in the initial land classification map is marked with a corresponding land type label.

[0058] For example, a pre-built and trained land classification model can be put to use. This land classification model is trained using a large dataset of sample images containing a variety of land cover types. The land classification model classifies each pixel on the comprehensive land feature map of the target area. Taking a single pixel as an example, the land classification model can determine whether the corresponding land type is construction land, cultivated land, forest, water body, or another type based on various characteristic information in the comprehensive land feature map, such as spectral reflectance, thermal radiation, and surface roughness. For example, if a pixel's characteristics include a high peak in spectral reflectance in the green band, low thermal radiation intensity, and low surface roughness, the model may label the pixel as cultivated land. After pixel-by-pixel classification of the entire comprehensive land feature map, an initial land classification map of the target area is obtained, in which each pixel is labeled with the corresponding land type.

[0059] Step S140: performing a time series comparative analysis on the land type labels in the initial land classification map based on historical land survey data, identifying dynamic areas of land cover change in the target area, and extracting change trajectory information of the dynamic areas.

[0060] First, obtain the land classification history map of the target area at least three historical time points, and extract the land type distribution pattern corresponding to each historical time point. Assume that this embodiment obtains the land classification history map of the target area 10 years ago, 5 years ago and 1 year ago. 10 years ago, the city in the target area was relatively small, with large tracts of cultivated land and forests around it, and the area of ​​water bodies was relatively stable. 5 years ago, the city began to expand outward, part of the cultivated land was converted into construction land, and the forest area also decreased slightly. 1 year ago, the city's expansion speed accelerated, more cultivated land and part of the forest were converted into construction land, and at the same time, some wetlands around the water bodies were also affected to a certain extent.

[0061] Then, spatial differences are calculated between the initial land classification map and the land type distribution patterns at each historical point in time to identify transition areas where land cover types have changed. For example, by comparing the current initial land classification map with the historical land classification map from 10 years ago, it is found that the areas extending outward from the city center are transition areas where land cover types have changed, where original farmland or forest has now become construction land. Similarly, by comparing the historical land classification maps from 5 years ago and 1 year ago, other areas where land type conversion has occurred can also be identified.

[0062] Finally, the conversion types of the conversion areas were categorized, generating a set of change types including forest-to-cultivated land, cultivated land-to-construction land, and water-to-wetland. Within the target area, comparative analysis revealed numerous areas where forests were cleared and converted to cultivated land; these areas were categorized as forest-to-cultivated land; urban expansion resulted in the occupation of large amounts of cultivated land for housing and road construction; these areas were categorized as cultivated land-to-construction land; and human activities around some water bodies resulted in a decrease in wetland area and an increase in water area; these cases were categorized as water-to-wetland. By categorizing these conversion types, we can more clearly understand the dynamics of land cover change within the target area and extract information about the change trajectories of these dynamic areas, such as the specific timescale and spatial extent of a particular area's transition from forest to cultivated land.

[0063] Step S150: generating a land resource change report based on the location coordinates, area data and change trajectory information of the dynamic area, wherein the land resource change report includes boundary vector data of the change area and change trend prediction results.

[0064] After identifying the dynamic areas of land cover change within the target region and obtaining the relevant location coordinates, area data, and change trajectory information, a land resource change report is generated. For example, if urban expansion in the target region leads to a reduction in cultivated land, the land resource change report will clearly indicate the boundary vector data of the urban expansion area (i.e., the change area). This data can accurately depict the scope of cultivated land occupied by urban expansion. Furthermore, based on historical trends in land cover change, such as the rate of urban expansion and the rate of cultivated land loss, future trends are predicted. For example, if the average annual urban expansion area over the past five years was 10 square kilometers, based on current urban development plans and population growth trends, it is predicted that the city may continue to expand at a rate of 12 square kilometers per year over the next five years, and this expansion will primarily occupy surrounding cultivated land. This prediction result is included in the land resource change report, providing a basis for the rational planning and management of land resources.

[0065] Next, step S120 can be applied for:

[0066] Step S121 : Divide the visible light image into a plurality of spectral bands, calculate a spectral reflectance curve of each spectral band, and extract a peak feature and a slope feature of the spectral reflectance curve.

[0067] In the target area, the processing of visible light images is very detailed. Assume that this embodiment divides the visible light band into several main bands, such as the blue light band (400-500nm), the green light band (500-600nm), and the red light band (600-700nm). For an area with a mixture of different vegetation types, the reflectivity of the vegetation in the blue light band is relatively low because blue light is easily absorbed by substances such as chlorophyll in the vegetation. When entering the green light band, the chlorophyll in the vegetation reflects the green light more strongly, causing the reflectivity to reach a peak. For example, the reflectivity peak of the grass in this area in the green light band may be around 550nm, while the reflectivity peak of the forest area in the green light band may be slightly higher, around 560nm, due to the more complex vegetation structure. Looking at the red light band, the reflectivity of the vegetation begins to decrease again. By accurately calculating the spectral reflectivity curve of each spectral band, its peak characteristics and slope characteristics can be further extracted. Taking grassland as an example, its slope gradually increases from the blue light band to the green light band, while the slope gradually decreases from the green light band to the red light band. These characteristics can help this embodiment distinguish different types of vegetation, and can also distinguish it from other non-vegetative land cover types (such as buildings, water bodies, etc.). The reflectance curves of buildings in various visible light bands are relatively smooth, without the obvious peaks and slope changes like vegetation.

[0068] Step S122: performing thermal radiation intensity analysis on the infrared image to determine thermal radiation difference parameters of different land cover types in the target area.

[0069] Continuing with the example of different land cover types within a target area, during sunny daytime, urban asphalt roads, due to their dark color, absorb significant solar heat, resulting in high thermal radiation intensity in infrared images. In contrast, adjacent parks and green spaces exhibit relatively low thermal radiation intensity due to transpiration from vegetation and its inherent heat dissipation properties. For water bodies, such as lakes, thermal radiation intensity varies relatively little between day and night, remaining generally low. Detailed thermal radiation intensity analysis of infrared images can accurately determine thermal radiation difference parameters for each pixel or small area. For example, at the border between a city and surrounding farmland, the average thermal radiation intensity in the urban area may reach a certain value, while the average thermal radiation intensity in the farmland area may be significantly lower than that in the urban area due to soil evaporation and vegetation. These thermal radiation difference parameters provide an important basis for distinguishing different land cover types.

[0070] Step S123: performing polarization scattering feature extraction on the radar image to obtain surface roughness and vegetation height parameters.

[0071] In forested areas of the target region, tree height and density are important factors affecting the polarimetric scattering characteristics of radar images. Tall and dense forests cause radar waves to scatter multiple times during propagation, resulting in a strong polarimetric scattering characteristic. This embodiment establishes a specific mathematical model to invert vegetation height in forested areas based on the polarimetric scattering data of radar images. For example, by analyzing the polarimetric scattering characteristics of radar images, the average vegetation height of a mature forest can be calculated to be between 20 and 30 meters. Regarding surface roughness, taking the comparison between cultivated land and wasteland as an example, cultivated land, due to cultivation and leveling, has a relatively smooth surface, and its polarimetric scattering characteristics in radar images exhibit weaker scattering characteristics. However, wasteland, due to its lack of artificial leveling, has more undulating surfaces and rocks, and its polarimetric scattering characteristics have a higher scattering intensity than cultivated land. By extracting polarimetric scattering characteristics from radar images, it is possible to accurately obtain surface roughness and vegetation height parameters. These parameters help to more accurately reflect the characteristics of land cover types in comprehensive land feature maps.

[0072] Step S124: fusing the peak characteristics of the spectral reflectance curve, the thermal radiation difference parameter, and the surface roughness parameter to generate a spatial resolution enhanced version of the comprehensive land feature map.

[0073] Fusion of features from these diverse sources is a crucial step in the data processing of a target area. For example, in a mixed area encompassing urban areas, forests, and cultivated land, the peak characteristics of the spectral reflectance curves obtained from visible light imagery (e.g., forest reflectance peak in the green band, cultivated land reflectance peak in the red band), thermal radiation difference parameters obtained from infrared imagery (e.g., high thermal radiation intensity in urban areas, relatively low thermal radiation intensity in forests), and surface roughness parameters obtained from radar imagery (e.g., high surface roughness in forests, low surface roughness in cultivated land) can be fused. Using a specific fusion algorithm, these features are combined according to specific weights and rules. The resulting spatially enhanced resolution version of the comprehensive land feature map can more clearly display the boundaries and internal structure between different land cover types. For example, at the urban-forest boundary, the boundary, which may have been blurred due to the limitations of a single data source, can be precisely displayed in the fused map, accurately reflecting the characteristic changes of land cover types on both sides of the boundary.

[0074] In a possible implementation, the method may further include:

[0075] Step S125: Determine the moisture parameter, organic matter content parameter, and pH parameter related to soil quality based on the physical properties of different land types in the comprehensive land characteristic map.

[0076] For cultivated land in the target area, soil quality is crucial. The comprehensive land characteristic map can be used to obtain information about the physical properties of cultivated land, allowing parameters related to soil quality to be inferred. For example, soil moisture can be determined based on the characteristics of the spectral reflectance curve in certain wavelengths. Low reflectance in the near-infrared band may indicate high soil moisture, as water absorbs light in the near-infrared band. Organic matter content can be inferred by analyzing the soil's texture and color. Soil with a high organic matter content is typically darker and has a finer texture. For example, cultivated land that has been cultivated for a long time will have a darker soil color than ordinary cultivated land, and its texture in the comprehensive land characteristic map will also show a finer texture, indicating a high organic matter content in the soil. pH can be determined based on spectral characteristics related to certain mineral components in the soil. For example, the absorption peaks of certain minerals in specific wavelengths can reflect the soil's pH.

[0077] Step S126: matching the humidity parameter, organic matter content parameter, and pH parameter with a preset soil quality assessment standard to generate a soil quality grade distribution map of the target area.

[0078] Assume that this embodiment has a set of preset soil quality assessment standards, and the soil quality is divided into four levels: excellent, good, medium, and poor based on parameters such as soil moisture, organic matter content, and pH. In the target area, for each cultivated land area, the previously determined humidity parameters, organic matter content parameters, and pH parameters can be matched with the assessment standards. For example, the humidity of a certain cultivated land is moderate, the organic matter content is high, and the pH is close to neutral. According to the assessment standards, the soil quality of this cultivated land is rated as "excellent". By performing such an assessment on all cultivated land and other land types related to soil quality in the entire target area, a soil quality grade distribution map can be generated. In this soil quality grade distribution map, it can be clearly seen which areas have good soil quality and which areas have poor soil quality, providing a basis for the rational use of land resources and soil improvement.

[0079] Step S127: performing spatial overlay analysis on the soil quality grade distribution map and the change areas in the land resource change report to determine priority areas for land degradation or improvement.

[0080] In a target area, when urban expansion leads to the occupation of some cultivated land, these occupied lands are marked as changed areas in the land resource change report. A soil quality grade distribution map can also be provided. By spatially overlaying these two data, it can be found that some occupied cultivated lands originally had excellent soil quality (e.g., "excellent" soil quality). If these areas were occupied by urban development, they would be high-priority areas for land degradation, as they originally had high agricultural production value. Conversely, areas with originally poor soil quality (e.g., "poor" soil quality) would be prioritized for land improvement if their soil quality was improved through measures such as land consolidation and soil improvement during land cover change. This spatial overlay analysis can help develop more scientific and rational strategies for land resource management and protection.

[0081] In a possible implementation, step S140 may include:

[0082] Step S141: Obtain land classification history maps of the target area at at least three historical time points, and extract the land type distribution pattern corresponding to each historical time point.

[0083] In the target area example, historical land classification maps have been identified for three historical time points: 10 years ago, 5 years ago, and 1 year ago. The land classification map from 10 years ago shows that the city was relatively small and concentrated in a few areas, surrounded by large tracts of farmland and forest. The distribution of water bodies was relatively stable, and wetlands also maintained a certain area. This is the land type distribution pattern corresponding to 10 years ago, where each land type has a specific spatial distribution range and shape. The land classification map from 5 years ago shows that the city began its initial expansion, extending toward the periphery along major transportation arteries. Some farmland on the city's fringes was incorporated into the city and converted into construction land. Small areas of forest around the city were also cleared to build small industrial parks or residential areas. At this time, the area of ​​water bodies and wetlands did not change much, but in some areas, the distribution and volume of water bodies underwent slight adjustments due to agricultural irrigation and urban water use. This is the land type distribution pattern corresponding to 5 years ago, reflecting the spatial layout and relationships of land cover types within the target area at that point in time.

[0084] A historical land classification map from one year ago shows that urban expansion has accelerated further, with more farmland being converted to construction and commercial and residential areas expanding outward. Forest area continues to decline, primarily due to urban expansion and illegal logging. Regarding water bodies, some small rivers and ponds have dried up or shrunk due to the construction of urban drainage systems and changes in land use. Wetlands are also threatened to some extent and have decreased in area. This is the land type distribution pattern from one year ago, reflecting the distribution characteristics of land cover types in the target area over a relatively recent period.

[0085] Step S142: performing spatial difference calculation between the initial land classification map and the land type distribution pattern at each historical time point to determine the conversion area where the land cover type changes.

[0086] For example, a specific area within the target region can be compared with the current initial land classification map and the land type distribution pattern from 10 years ago. Suppose that in the initial land classification map, a certain area is labeled as construction land, but in the historical land classification map from 10 years ago, this area is labeled as cultivated land. Through precise spatial difference calculation, it can be determined that this area is the transition zone where the land cover type has changed. This calculation involves comparing the location and land type label of each pixel in the two maps, calculating the set of pixels that have changed spatially, and the area formed by these pixels is where the land cover type change occurred.

[0087] Similarly, spatial differences in land type distribution patterns can be calculated between the initial land classification map and those from five years and one year ago. For example, a comparison with the map from five years ago revealed an area on the edge of a city that was previously forested but is now labeled as construction land and partially wasteland (due to its undeveloped state after deforestation) on the initial land classification map. This area represents a transition zone where land cover types shifted during these five years. A comparison with the map from one year ago also identifies transition zones such as newly occupied farmland due to urban expansion and areas where forests have been further destroyed.

[0088] Step S143: Classify the conversion types of the conversion area to generate a set of change types including forest to farmland, farmland to construction land, and water body to wetland.

[0089] After identifying multiple conversion zones within the target area through the previous steps, these conversion zones are then classified into specific types. For example, in mountainous areas surrounding cities, areas where forests are cut down and converted to farmland for agricultural development fall into the forest-to-farm conversion category. During urban expansion, large amounts of farmland are used to build urban infrastructure, residential buildings, and commercial facilities. These conversion zones fall into the farmland-to-construction conversion category.

[0090] Regarding the conversion of water bodies and wetlands, in some corners of the target region, due to human reclamation activities or water conservancy project construction, some areas that were originally water bodies have gradually dried up, and the surrounding wetland vegetation has also been destroyed, resulting in a decrease in the area of ​​water bodies and wetlands. In other places, due to increased awareness of wetland protection and the implementation of restoration measures, the land surrounding some water bodies has gradually transformed into wetlands. The change types in these areas are respectively water bodies to wetlands (negative conversion) and wetlands to water bodies (positive conversion). By classifying the conversion types of all conversion areas, a complete set of various change types can be generated, which can comprehensively reflect the dynamic characteristics of land cover change in the target region.

[0091] Furthermore, in a possible implementation, step S150 may include:

[0092] Step S151: marking the geometric shape and center coordinates of the changed area in the digital map of the target area according to the boundary vector data of the changed area.

[0093] On a digital map of the target area, for example, the area of ​​change due to urban expansion occupying arable land can be used. Based on the boundary vector data of the change area in the previously generated land resource change report, this data accurately describes the boundary location of the change area. The geometric shape of the change area can be accurately mapped on the digital map. For example, if it is an irregular polygon, the vertices and edges of the polygon are depicted according to the vector data. Simultaneously, the center coordinates of the change area are calculated. These center coordinates can be obtained by performing specific mathematical calculations on the coordinates of the polygon vertices, such as taking the center coordinates of the polygon as the center coordinates. This center coordinate point is marked on the digital map, which intuitively displays the location and shape characteristics of the change area on the map.

[0094] Step S152: Based on the change trend prediction result, the area expansion rate and type conversion probability of the change area within a future preset time period are calculated.

[0095] Continuing with the example of changes in the area of ​​arable land occupied by urban expansion, if previous land resource change reports predict that cities will continue to expand outward over the next five years, primarily occupying surrounding arable land, we can calculate the average annual rate of urban expansion based on historical data on urban expansion, such as the area of ​​urban expansion over the past five years. Assuming that urban expansion has occupied an average of 10 square kilometers of arable land annually over the past five years, then this 10 square kilometers / year represents the past rate of urban expansion. Based on factors such as existing urban development plans, population growth trends, and land use policies, we can forecast the rate of urban expansion over the next five years. If we predict that urban expansion will accelerate over the next five years, we might occupy an average of 12 square kilometers of arable land annually. This represents the rate of urban expansion over the predefined time period.

[0096] The calculation of land type conversion probabilities can consider the likelihood of conversion between different land types within the changing area. For example, during urban expansion, in addition to the primary conversion from cultivated land to construction land, there may also be a small amount of cultivated land converted to wasteland (due to idle or temporarily undeveloped land). Based on historical data and various current influencing factors, the probability of cultivated land converting to construction land is calculated to be 0.8 (indicating an 80% probability), and the probability of cultivated land converting to wasteland is calculated to be 0.2 (indicating a 20% probability). These calculations of area expansion rates and land type conversion probabilities provide a quantitative basis for land resource planning and management.

[0097] Step S153: Integrate the geometric shape, center coordinates, area expansion rate, and type conversion probability into a visualization layer, and overlay it on the interactive interface of the digital map.

[0098] Within the interactive interface of a digital map of the target area, information such as the geometric shape (e.g., polygon) of the previously marked change area, the calculated center coordinates, the area expansion rate, and the probability of type conversion can be integrated into a visualization layer. The geometric shape is clearly displayed on the map using a specific color and line; the center coordinates are displayed on the map using a special marker (e.g., a small circle or a cross). The area expansion rate can be represented by adding a data label or using different color shades to indicate the expansion rate (e.g., the greater the expansion rate, the darker the color); and the type conversion probability can also be represented by similar data labels or color grading to indicate the likelihood of different types of conversion. This visualization layer is then overlaid on the interactive interface of the digital map. This allows land resource managers and other relevant personnel to intuitively obtain detailed information about the change area when viewing the digital map, facilitating land resource planning, decision-making, and management.

[0099] Furthermore, the method may further include:

[0100] Step S161: receiving the latest image data transmitted by the satellite remote sensing system in real time, performing spectral feature analysis and texture feature extraction on the latest image data, and generating a real-time land feature map.

[0101] During real-time monitoring of land resources in a target area, satellite remote sensing systems continuously transmit the latest image data. This image data contains the current land information for the target area. The processing of this latest image data is similar to the processing of multi-source remote sensing image data. First, spectral feature analysis is performed. The visible light image is divided into multiple spectral bands. The spectral reflectance curve for each spectral band is calculated, and its peak and slope characteristics are extracted. Thermal radiation intensity analysis is performed on infrared images to determine the thermal radiation difference parameters for different land cover types. Polarimetric scattering feature extraction is performed on radar images to obtain surface roughness and vegetation height parameters. These features extracted from different image types are then integrated to generate a real-time land feature map. For example, at a certain moment, the satellite transmits the latest image data containing urban, forest, cultivated land, and water bodies in the target area. After the above processing, the real-time land feature map accurately reflects the characteristics of each land cover type at that moment, such as the spectral reflectance curve, thermal radiation intensity, and polarimetric scattering characteristics of urban buildings, and the vegetation height, thermal radiation, and spectral reflectance characteristics of forest areas.

[0102] Step S162: performing difference detection between the real-time land feature map and the comprehensive land feature map to identify abnormal areas of sudden land cover changes within the target area.

[0103] The newly generated real-time land feature map is compared with the previously generated comprehensive land feature map. For example, for a forested area within the target region, if the vegetation height and spectral reflectance of a particular forested area appear normal in the comprehensive land feature map, but the real-time land feature map reveals a sudden decrease in vegetation height and a significant shift in the spectral reflectance curve, this could be due to a forest fire or large-scale illegal logging. Such areas with significant differences between the two maps are anomalous areas of sudden land cover change. For another example, in cultivated land, if the real-time land feature map reveals large irregularities in previously flat farmland, and the spectral reflectance curve differs significantly from that in the comprehensive land feature map, this could be due to natural disasters such as floods and mudslides, or illegal land occupation. This difference detection allows for the rapid and accurate identification of anomalous areas of sudden land cover change within the target region.

[0104] Step S163: Sending a land resource anomaly warning signal to the monitoring terminal according to the geographical location and change type of the abnormal area, wherein the warning signal includes the coordinate range of the abnormal area and recommended verification measures.

[0105] Once an abnormal area is identified, its specific location is determined based on its geographic location within the target area, such as its latitude and longitude coordinates. If the abnormality is caused by a possible forest fire in a forest area, and the change type is vegetation destruction, the land resource abnormality warning signal sent to the monitoring terminal will include the precise coordinate range of the abnormal area and recommend verification measures, such as dispatching drones or ground personnel to the area to investigate fire hazards or forest damage. If the abnormality is caused by a possible natural disaster or illegal occupation in a cultivated land area, the warning signal will also include the coordinate range. Recommended verification measures may include further detailed analysis using satellite imagery or dispatching land inspectors for on-site inspections. This allows relevant personnel to be notified in a timely manner to address the abnormality in land resources, protecting the safety and rational use of land resources.

[0106] Furthermore, the method may further include:

[0107] Step S171: obtaining a sample image dataset containing multiple land cover types, and marking each sample in the sample image dataset with a real land type label.

[0108] When building a land classification model, it's first necessary to collect a large dataset of sample images. These sample images should cover all possible land cover types in the target area, such as different types of forest (coniferous, broadleaf, etc.), cultivated land with different cropping patterns (paddy fields, dry land, etc.), urban construction land of various sizes, water bodies of varying forms (rivers, lakes, ponds, etc.), and wetlands with different ecological environments. Each sample image should be annotated with the actual land type label through field surveys, historical data queries, or high-precision reference images. For example, if a sample image shows a forested area filled with pine trees, the sample image could be labeled with the land type label "coniferous forest"; if it shows a paddy field planted with rice, the sample image could be labeled with the land type label "paddy field." This annotation of a large number of sample images provides an accurate reference for subsequent model training.

[0109] Step S172: Construct a deep convolutional neural network, input the sample image dataset into the deep convolutional neural network for feature learning, and output the predicted land type probability distribution.

[0110] A deep convolutional neural network (DCNN) is constructed, consisting of multiple convolutional layers, pooling layers, and fully connected layers. A previously labeled sample image dataset is input into the DCNN. In the network's convolutional layers, the convolution kernel slides across the sample image, extracting local features such as edges and texture. The pooling layer downsamples the features extracted by the convolutional layer, reducing the data size while retaining important feature information. After processing through multiple convolutional and pooling layers, the features are integrated in the fully connected layer, ultimately outputting a predicted land type probability distribution. For example, for a sample image input, the DCNN might output a probability distribution of 0.3 for "coniferous forest," 0.2 for "broadleaved forest," 0.1 for "paddy field," 0.15 for "dry land," 0.15 for "urban construction land," 0.05 for "water body," and 0.1 for "wetland."

[0111] Step S173: Calculate the cross entropy loss value based on the difference between the predicted land type probability distribution and the true land type label, and use the back propagation algorithm to optimize the weight parameters of the deep convolutional neural network.

[0112] The cross-entropy loss is calculated based on the predicted land type probability distribution output by the network and the true land type label originally annotated in the sample image. The cross-entropy loss is a metric that measures the difference between the predicted result and the true result. For example, if the true label of a sample image is "coniferous forest," and the network's predicted probability of "coniferous forest" is 0.3, there is a certain difference between the predicted result and the true result. The cross-entropy loss corresponding to this difference is calculated using a specific formula. Then, using the backpropagation algorithm, the network's weight parameters are gradually adjusted, starting from the output layer and moving toward the input layer, based on the calculated cross-entropy loss. The weight parameter adjustment is intended to make the network's prediction closer to the true result and reduce the cross-entropy loss. For example, if adjusting the weight parameter of a convolution kernel in a convolutional layer brings the network's predicted probability of "coniferous forest" sample image closer to 1 (indicating a more accurate prediction of "coniferous forest"), then the weight parameter adjustment is effective.

[0113] Step S174: When the cross entropy loss value is lower than a preset threshold, stop training and use the optimized deep convolutional neural network as the land classification model.

[0114] During the training process, a preset threshold, such as 0.01, can be set. When the calculated cross-entropy loss falls below this threshold, the network's predictions are sufficiently close to the true results, and training is terminated. The optimized deep convolutional neural network is then used as the final land classification model. This land classification model can then be used to perform pixel-by-pixel classification on the comprehensive land feature map of the target area. In practical applications, for example, it can accurately predict the land type of each pixel in the target area, providing reliable classification results for land resource surveys.

[0115] Furthermore, the method may further include:

[0116] Step S181: receiving land survey feedback information submitted by a user through an interactive terminal, wherein the land survey feedback information includes correction suggestions for the type labeling of a specific area in the land resource change report.

[0117] During the land resource survey process, users (who may be land resource managers, field survey personnel, or other relevant experts) submit land survey feedback information through interactive terminals (such as computers, tablets, or specialized land survey equipment). For example, when reviewing a land resource change report, a user may find that a certain area is labeled "wasteland" in the report. However, based on the user's own field survey or other reliable information, the user believes that the area is actually "fallow farmland." The user then submits land survey feedback information containing the correction suggestion (correcting "wasteland" to "fallow farmland") through the interactive terminal.

[0118] Step S182: adjusting the classification weight parameters of the land classification model according to the type labeling correction suggestion, and reclassifying the comprehensive land characteristic map of the target area to generate an updated land classification map.

[0119] Upon receiving a correction suggestion from the user, the classification weight parameters of the land classification model are adjusted based on the correction suggestion. For example, if the user's correction suggestion is regarding the type labels of "wasteland" and "fallow cultivated land," then the feature extraction and classification weight parameters related to these two land types will be adjusted in the land classification model. This adjustment may be done by increasing the weight of features related to "fallow cultivated land" (such as specific spectral reflectance features, texture features, etc.) and reducing the weight of features related to "wasteland." The adjusted land classification model is then used to reclassify the comprehensive land feature map of the target area. During this process, the land classification model can reclassify each pixel based on the new weight parameters, ultimately generating an updated land classification map. In this updated land classification map, areas previously incorrectly labeled as "wasteland" may be correctly labeled as "fallow cultivated land."

[0120] Step S183: performing consistency check on the updated land classification map and historical data to determine the degree of optimization of the classification result by the correction suggestion, and feeding back the check result to the interactive terminal.

[0121] The updated land classification map is checked for consistency with previous historical data (such as historical land classification maps and historical land survey records). For example, the land type labels for the same area on the updated land classification map are compared with those on the historical land classification map from five years ago. If an area is correctly labeled as "fallow cultivated land" on the updated land classification map, while the same area is also "cultivated land" (just under a different cultivation status) in the historical map from five years ago, this indicates that the correction suggestion has a positive effect on the classification results. By comparing and analyzing multiple areas and multiple historical data, the overall degree of improvement of the classification results by the correction suggestion is determined. Finally, the verification results (such as qualitative results such as "high," "medium," or "low" optimization levels, or specific quantitative data) are fed back to the interactive terminal, allowing users to understand the impact of their feedback on the land classification results, enabling them to make more accurate judgments and decisions in subsequent land resource surveys and management.

[0122] In the above embodiments, the AI ​​analysis system applied to land resource survey for executing the above method embodiments has at least one processor, a control module (chip set) coupled to at least one of (at least one) processors, a memory coupled to the control module, a non-volatile memory (NVM) / storage device coupled to the control module, at least one load / output device coupled to the control module, and a network interface coupled to the control module.

[0123] The processor may include at least one single-core or multi-core processor, and the processor may include any combination of general-purpose processors or dedicated processors (such as graphics processors, application processors, baseband processors, etc.). For some alternative implementations, the AI ​​analysis system applied to land resource surveys can be based on electronic devices such as the gateway described in the embodiments of this application.

[0124] For some alternative embodiments, the AI ​​analysis system applied to land resource surveys may include at least one computer-readable medium (e.g., a memory or NVM / storage device) having instructions and at least one processor integrated with the at least one computer-readable medium and configured to execute instructions to implement a module to perform the actions described in the present disclosure.

[0125] For one embodiment, the control module may include any suitable interface controller to provide any suitable interface to at least one of the processor(s) and / or any suitable device or component in communication with the control module.

[0126] The control module may include a memory controller module to provide an interface to the memory. The memory controller module may be a hardware module, a software module, and / or a firmware module.

[0127] The memory may be used, for example, to load and store data and / or instructions for an AI analysis system applied to land resource surveys. For one embodiment, the memory may include any suitable volatile memory, for example, a suitable DRAM.

[0128] For one embodiment, the control module may include at least one load / output controller to provide an interface to the NVM / storage device and the (at least one) load / output device.

[0129] For example, NVM / storage devices may be used to store data and / or instructions. The NVM / storage devices may include any suitable non-volatile memory (e.g., flash memory) and / or may include any suitable (at least one) non-volatile storage device (e.g., at least one hard disk drive (HDD), at least one compact disc (CD) drive, and / or at least one digital versatile disc (DVD) drive).

[0130] The NVM / storage device may include a storage resource that is physically part of the device on which the AI ​​analysis system for land resource survey is installed, or it may be accessible to the device without being part of the device. For example, the NVM / storage device may be accessed via (at least one) load / output device over a network.

[0131] (At least one) load-in / output device may provide an interface for the AI ​​analysis system applied to land resource surveys to communicate with any other appropriate device. The load-in / output device may include a communication component, a pinyin component, a sensor component, etc. A network interface may provide an interface for the AI ​​analysis system applied to land resource surveys to communicate based on at least one network. The AI ​​analysis system applied to land resource surveys may wirelessly communicate with at least one component of a wireless network based on any of at least one wireless network priors and / or protocols, for example, accessing a wireless network based on a communication prior.

[0132] For one embodiment, at least one of the (at least one) processors may be loaded together with the logic of at least one controller of a control module (e.g., a memory controller module). For one embodiment, at least one of the (at least one) processors may be loaded together with the logic of at least one controller of a control module to form a system-level load. For one embodiment, at least one of the (at least one) processors may be fused on the same die with the logic of at least one controller of a control module. For one embodiment, at least one of the (at least one) processors may be fused on the same die with the logic of at least one controller of a control module to form a system-on-chip (SoC).

[0133] The above is a detailed introduction to the embodiments of the present application. Specific examples are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method and core idea of ​​the present application. At the same time, for those skilled in the art, based on the ideas of the present application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

[0134] An embodiment of the present invention discloses a computer-readable storage medium storing a computer program for electronic data exchange, wherein the computer program enables a computer to execute the steps of the AI ​​analysis method applied to land resource survey described in the aforementioned embodiment.

[0135] An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to enable a computer to execute the steps of the AI ​​analysis method applied to land resource surveys described in the aforementioned embodiment.

[0136] The device embodiments described above are merely illustrative, wherein the modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, i.e., they may be located in one place or distributed across multiple modules. Some or all of the modules may be selected based on actual needs to achieve the purpose of the present embodiment. Those skilled in the art can understand and implement the present invention without inventive effort.

[0137] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus the necessary general hardware platform, or of course, by hardware. Based on this understanding, the above technical solution, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product, which can be stored in a computer-readable storage medium, the storage medium including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electronically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium capable of storing or storing data.

[0138] Finally, it should be noted that what is disclosed above is only a preferred embodiment of the present invention, which is only used to illustrate the technical solution of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. An AI analysis method applied to land resource survey, characterized in that: The method comprises: Acquiring multi-source remote sensing image data of the target area, wherein the multi-source remote sensing image data includes visible light images, infrared images, and radar images, and contains spectral characteristics and texture characteristics related to land cover types; Performing spectral feature analysis and texture feature extraction on the multi-source remote sensing image data to generate a comprehensive land feature map of the target area, wherein the comprehensive land feature map is used to characterize the spatial distribution and physical properties of land cover types; Perform pixel-by-pixel classification processing on the comprehensive land feature map based on a preset land classification model to obtain an initial land classification map of the target area, wherein each pixel in the initial land classification map is marked with a corresponding land type label; Performing a time series comparative analysis on the land type labels in the initial land classification map based on historical land survey data, identifying dynamic areas of land cover change in the target area, and extracting change trajectory information of the dynamic areas; Generate a land resource change report based on the location coordinates, area data and change trajectory information of the dynamic area, wherein the land resource change report includes boundary vector data of the change area and change trend prediction results; Marking the geometric shape and center coordinates of the changed area on the digital map of the target area according to the boundary vector data of the changed area; Based on the change trend prediction results, calculate the area expansion rate and type conversion probability of the change area within a future preset time period; The geometric shape, center coordinates, area expansion rate and type conversion probability are integrated into a visualization layer and superimposed on the interactive interface of the digital map.

2. The AI ​​analysis method for land resource survey according to claim 1, characterized in that: The performing spectral feature analysis and texture feature extraction on the multi-source remote sensing image data to generate a comprehensive land feature map of the target area includes: Dividing the visible light image into a plurality of spectral bands, calculating a spectral reflectance curve for each spectral band, and extracting peak characteristics and slope characteristics of the spectral reflectance curve; Performing thermal radiation intensity analysis on the infrared image to determine thermal radiation difference parameters of different land cover types in the target area; Extracting polarization scattering features from the radar image to obtain surface roughness and vegetation height parameters; The peak characteristics of the spectral reflectance curve, the thermal radiation difference parameter and the surface roughness parameter are integrated to generate a spatial resolution enhanced version of the comprehensive land feature map.

3. The AI ​​analysis method for land resource survey according to claim 2, characterized in that: The method further comprises: Determining moisture parameters, organic matter content parameters, and pH parameters related to soil quality based on the physical properties of different land types in the comprehensive land characteristic map; Matching the humidity parameter, organic matter content parameter, and pH parameter with a preset soil quality assessment standard to generate a soil quality grade distribution map for the target area; The soil quality grade distribution map is spatially overlaid with the change areas in the land resource change report to determine the priority areas for land degradation or improvement.

4. The AI ​​analysis method for land resource survey according to claim 1, characterized in that: The performing of a time series comparative analysis on the land type labels in the initial land classification map based on historical land survey data to identify dynamic areas of land cover changes within the target area includes: Obtaining land classification historical maps of the target area at at least three historical time points, and extracting the land type distribution pattern corresponding to each historical time point; Calculating spatial differences between the initial land classification map and the land type distribution patterns at each historical time point to determine transition areas where land cover types undergo transformation; The conversion types of the conversion areas are classified to generate a set of change types including forest to farmland, farmland to construction land, and water body to wetland.

5. The AI ​​analysis method for land resource survey according to claim 1, characterized in that: The method further comprises: Receive the latest image data transmitted by the satellite remote sensing system in real time, perform spectral feature analysis and texture feature extraction on the latest image data, and generate a real-time land feature map; Performing difference detection between the real-time land feature map and the comprehensive land feature map to identify abnormal areas of sudden land cover changes within the target area; According to the geographical location and change type of the abnormal area, a land resource abnormality warning signal is sent to the monitoring terminal, and the warning signal includes the coordinate range of the abnormal area and recommended verification measures.

6. The AI ​​analysis method for land resource survey according to claim 1, characterized in that: The land classification model is trained by the following steps: Obtain a sample image dataset containing multiple land cover types, and label each sample in the sample image dataset with a true land type label; Constructing a deep convolutional neural network, inputting the sample image dataset into the deep convolutional neural network for feature learning, and outputting a predicted land type probability distribution; Calculating a cross entropy loss value based on the difference between the predicted land type probability distribution and the true land type label, and optimizing the weight parameters of the deep convolutional neural network using a back propagation algorithm; When the cross entropy loss value is lower than a preset threshold, the training is stopped and the optimized deep convolutional neural network is used as the land classification model.

7. The AI ​​analysis method for land resource survey according to claim 1, characterized in that: The method further comprises: receiving land survey feedback information submitted by a user through an interactive terminal, wherein the land survey feedback information includes a suggestion for correcting a type labeling of a specific area in the land resource change report; Adjusting the classification weight parameters of the land classification model according to the type labeling correction suggestion, and reclassifying the comprehensive land characteristic map of the target area to generate an updated land classification map; The updated land classification map is checked for consistency with historical data to determine the degree to which the correction suggestion optimizes the classification result, and the check result is fed back to the interactive terminal.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores machine-executable instructions, which, when executed by a computer, implement the AI ​​analysis method for land resource survey described in any one of claims 1 to 7.

9. An AI analysis system for land resource survey, characterized in that: It includes a processor and a computer-readable storage medium, wherein the computer-readable storage medium stores machine-executable instructions, and when the machine-executable instructions are executed by a computer, the AI ​​analysis method for land resource survey described in any one of claims 1 to 8 is implemented.

Images