Automatic map drawing method and system

Abstract

The invention discloses an automatic map drawing method and system, and the method comprises the steps: obtaining farmland multi-source data containing multispectral sensor data and a high-resolution image, and generating a first farmland data set through time sequence fusion; according to a pre-established unified data model, performing standardization processing on the first farmland data set to obtain a second farmland data set; extracting farmland boundary and topographic features from the second farmland data set, and if the complexity of the topographic features exceeds a preset threshold, introducing a digital elevation model to supplement topographic data to generate a third farmland data set; obtaining dynamic change characteristics according to the third farmland data set, judging crop growth and soil state change trends by adopting a time sequence analysis algorithm, and generating a fourth farmland data set; and extracting update features from the fourth farmland data set, fusing preset high-precision map data by adopting an incremental update algorithm, and generating a dynamic farmland map. The map updating efficiency and precision are improved, and reliable data support is provided for agricultural management and decision making.

Description

Technical Field

[0001] The present invention relates to the technical field of map drawing, and in particular discloses an automatic map drawing method and system. Background Art

[0002] Automated farmland mapping is a core technology for the development of smart agriculture, crucial for improving agricultural production efficiency, achieving precise management, and ensuring food security. With the widespread adoption of smart agricultural machinery, high-precision farmland maps have become a crucial foundation for supporting machinery navigation, resource optimization, and scientific decision-making.

[0003] However, limitations of existing methods hinder their further development. Traditional farmland mapping relies on manual surveying or single remote sensing technologies, which are inefficient, costly, and difficult to achieve large-scale, high-frequency dynamic updates. Especially in complex terrain and remote areas, data accuracy and real-time performance struggle to meet the demands of smart agricultural machinery. These limitations stem from several interrelated core challenges.

[0004] Inefficient data collection is a primary issue. Traditional methods, limited by equipment coverage and operational complexity, cannot quickly acquire high-precision data over a wide area. This complicates data integration, as multi-source data (such as remote sensing and drone imagery) lack uniform formats and standards, making it difficult to efficiently integrate and generate unified maps. Furthermore, the high cost and low frequency of data updates create bottlenecks, hindering the timely reflection of dynamic changes in the farmland environment (such as crop growth and soil conditions), limiting the maps' practicality.

[0005] Therefore, how to achieve rapid collection, standardized integration and dynamic updating of multi-source data through low-cost and high-efficiency technical means has become a key issue in the automated drawing of smart agricultural machinery farmland maps. Summary of the Invention

[0006] The present invention provides a method and system for automatic map drawing, aiming to solve at least one of the above-mentioned defects of the prior art.

[0007] One aspect of the present invention relates to a method for automatically drawing a map, comprising the following steps: Acquire multi-source farmland data including multispectral sensor data and high-resolution images, and generate the first farmland dataset through time series fusion; According to a pre-established unified data model, the first farmland dataset is standardized to obtain a second farmland dataset; Extracting farmland boundaries and terrain features from the second farmland dataset. If the complexity of the terrain features exceeds a preset threshold, a digital elevation model is introduced to supplement the terrain data to generate a third farmland dataset. The dynamic change characteristics of the third farmland dataset are obtained, and the time series analysis algorithm is used to determine the trend of crop growth and soil state changes to generate the fourth farmland dataset; Update features are extracted from the fourth farmland dataset, and the preset high-precision map data are fused using an incremental update algorithm to generate a dynamic farmland map.

[0008] Furthermore, the steps of obtaining multi-source farmland data including multispectral sensor data and high-resolution images and generating a first farmland dataset through time series fusion include: The spectral data of farmland are collected by multispectral sensors and image data by high-resolution cameras. The acquisition time is recorded by timestamp to generate an initial dataset containing spectral characteristics and spatial information. The time series analysis method is used to align the spectral data and image data in the initial dataset according to the timestamps, and the multi-source features are fused through the weighted average method to generate the first farmland dataset.

[0009] Furthermore, the step of performing standardization processing on the first farmland dataset according to the pre-established unified data model to obtain the second farmland dataset includes: Obtain the first farmland dataset and use data cleaning tools to process missing values ​​and outliers. If a data point exceeds the preset threshold, replace it with the time series mean to obtain the cleaned dataset. According to the field definition of the unified data model, spectral features and spatial features are extracted from the cleaned dataset and aligned to a standard format using a format conversion tool to obtain a feature-aligned dataset. For the feature alignment dataset, a data verification tool is used to check the data integrity and consistency. If data is missing, it is filled by interpolation to obtain a verified dataset. The multi-source features in the verified dataset are fused through a weighted average algorithm, and the fused feature values ​​are determined to obtain the second farmland dataset.

[0010] Furthermore, the steps of extracting farmland boundaries and terrain features from the second farmland dataset and introducing a digital elevation model to supplement the terrain data if the complexity of the terrain features exceeds a preset threshold to generate a third farmland dataset include: Obtain farmland boundary data from the second farmland dataset and use edge detection tools to extract boundary lines. If the continuity of the boundary lines is lower than a preset threshold, interpolation is used to fill in the discontinuities to obtain a boundary extraction dataset. Terrain feature data is obtained from the boundary extraction dataset, and slope and elevation distribution are calculated using terrain analysis tools. If the complexity of terrain features exceeds a preset threshold, high-resolution terrain data is supplemented with a digital elevation model to obtain a terrain enhancement dataset. For the terrain enhancement dataset, the data fusion tool is used to weighted average the boundary features and terrain features, and the format conversion tool is used to align them into a standard format to obtain a feature fusion dataset; Verification data is obtained from the feature fusion dataset, and data verification tools are used to check the integrity and consistency. If verification data is found to be missing, it is filled in by linear interpolation to obtain the third farmland dataset.

[0011] Furthermore, the steps of obtaining dynamic change characteristics based on the third farmland dataset and using a time series analysis algorithm to determine the trend of crop growth and soil state changes to generate the fourth farmland dataset include: Crop growth data and soil status data were obtained from the third farmland dataset. Data extraction tools were used to separate crop height, leaf area index, soil moisture, and nutrient content data in the time dimension to obtain a dynamic feature dataset. For dynamic feature datasets, time series analysis tools are used to calculate the periodic change characteristics of crop growth and soil conditions. If the fluctuation amplitude of the change characteristics exceeds the preset threshold, the data is smoothed using the moving average method to obtain a smoothed feature dataset. Obtain time series data from the smoothed feature dataset, use the autoregressive moving average algorithm to calculate the changing trends of crop growth stages and soil conditions, and obtain a trend analysis dataset; For the trend analysis dataset, a data format conversion tool was used to align the crop growth trend and soil status trend into a standard format to obtain the fourth farmland dataset.

[0012] Furthermore, the steps of extracting update features from the fourth farmland dataset and fusing the preset high-precision map data with an incremental update algorithm to generate a dynamic farmland map include: Extract crop distribution data and soil property data from the fourth farmland dataset. Use data extraction tools to separate crop type, coverage, and soil nutrients and moisture data in the spatial dimension to obtain a spatial feature dataset. For the spatial feature dataset, an incremental update algorithm is used to match the preset high-precision map data. If the spatial coordinate deviation between the crop distribution data and the high-precision map exceeds a preset threshold, the coordinates are adjusted through spatial interpolation to obtain an aligned feature dataset. Based on the aligned feature dataset, a data fusion tool is used to merge crop distribution data, soil property data, and spatial grid data of high-precision maps to obtain a fused feature dataset. By fusing feature datasets, a map generation tool is used to map crop distribution data and soil property data onto a spatial grid to obtain a dynamic farmland map.

[0013] Another aspect of the present invention relates to an automated map drawing system for implementing the automated map drawing method described above. The automated map drawing system comprises: The first generation module acquires multi-source farmland data including multispectral sensor data and high-resolution images, and generates the first farmland dataset through time series fusion; The second generation module performs standardization processing on the first farmland dataset according to a pre-established unified data model to obtain a second farmland dataset; The third generation module extracts farmland boundaries and terrain features from the second farmland dataset. If the complexity of the terrain features exceeds a preset threshold, a digital elevation model is introduced to supplement the terrain data to generate a third farmland dataset. The fourth generation module obtains dynamic change characteristics based on the third farmland dataset, uses a time series analysis algorithm to determine the trend of crop growth and soil state changes, and generates a fourth farmland dataset; The fifth generation module extracts update features from the fourth farmland dataset, adopts an incremental update algorithm to fuse the preset high-precision map data, and generates a dynamic farmland map.

[0014] Furthermore, the first generation module includes: The first generation unit is used to collect farmland spectral data and image data using a multispectral sensor and a high-resolution camera, record the collection time using a timestamp, and generate an initial data set containing spectral characteristics and spatial information; The second generation unit is used to align the spectral data and image data in the initial data set according to timestamps using a time series analysis method, fuse multi-source features using a weighted average method, and generate a first farmland data set.

[0015] Furthermore, the second generation module includes: A first acquisition unit is used to acquire a first farmland dataset, process missing values ​​and outliers using a data cleaning tool, and replace a data point with a time series mean value if the data point exceeds a preset threshold to obtain a cleaned dataset; The second acquisition unit is used to extract spectral features and spatial features from the cleaned data set according to the field definition of the unified data model, and align them into a standard format through a format conversion tool to obtain a feature-aligned data set; The third acquisition unit is used to check the integrity and consistency of the feature alignment dataset using a data verification tool. If any data is missing, it is filled in by interpolation to obtain a verified dataset. The fourth acquisition unit is used to fuse the multi-source features in the verified data set through a weighted average algorithm, determine the fused feature value, and obtain the second farmland data set.

[0016] Furthermore, the third generation module includes: a fifth acquisition unit, configured to acquire farmland boundary data from the second farmland dataset, extract boundary lines using an edge detection tool, and fill in discontinuities using an interpolation method if the continuity of the boundary lines is lower than a preset threshold, thereby obtaining a boundary extraction dataset; a sixth acquisition unit, configured to obtain terrain feature data from the boundary extraction dataset, calculate slope and elevation distribution using terrain analysis tools, and, if the complexity of the terrain features exceeds a preset threshold, supplement the high-resolution terrain data with a digital elevation model to obtain a terrain enhancement dataset; a seventh acquisition unit, configured to perform weighted averaging of boundary features and terrain features using a data fusion tool for the terrain enhancement dataset, and align the features into a standard format using a format conversion tool to obtain a feature fusion dataset; The eighth acquisition unit is used to obtain verification data from the feature fusion dataset, and use data verification tools to check the integrity and consistency. If the verification data is found to be missing, it is filled by linear interpolation to obtain the third farmland dataset.

[0017] The beneficial effects achieved by the present invention are: The present invention provides an automated mapping method and system that accurately updates farmland information through multi-source data fusion and time series analysis. The method first acquires multispectral sensor data and high-resolution imagery, performs time series fusion, and standardizes the data. Farmland boundaries and terrain features are then extracted, and a digital elevation model is introduced to supplement complex terrain data when necessary. A time series analysis algorithm is then used to determine crop growth and soil condition trends, extracting dynamic change features. Finally, an incremental update algorithm is employed to fuse the new features with high-precision map data to generate a dynamic farmland map. The present invention can promptly capture farmland change information, improve map update efficiency and accuracy, and provide reliable data support for precision agriculture management and decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 The figure is a flowchart of an embodiment of a method for automatically drawing a map according to the present invention. DETAILED DESCRIPTION

[0019] In order to better understand the above technical solution, the above technical solution will be described in detail below with reference to the accompanying drawings and specific implementation methods.

[0020] like Figure 1 As shown, the first embodiment of the present invention provides a method for automatically drawing a map, comprising the following steps: Step S100: Acquire multi-source farmland data including multispectral sensor data and high-resolution images, and generate a first farmland dataset through time series fusion.

[0021] Multispectral sensor data is remote sensing data collected by multispectral sensors, covering multiple discrete spectral bands (usually from visible light to near infrared or shortwave infrared bands).

[0022] High-resolution imagery refers to image data acquired by remote sensing sensors with a spatial resolution of less than 1 meter per pixel. Its core characteristic is that individual pixels have a relatively small relative size to the ground, allowing for clear visualization of ground features.

[0023] Multi-source farmland data refers to a heterogeneous data set covering the entire agricultural production chain, acquired through various technical means and sources. It is used to support precision agriculture decision-making and farmland management. Its core characteristics are the diversity of data sources and the complexity of data types, and the need for integration and analysis to achieve a multi-dimensional portrayal of farmland conditions.

[0024] Time series fusion is the process of integrating and collaboratively analyzing time series data from multiple sources (such as sensor time series data and periodic remote sensing observations). Through mathematical modeling or algorithm optimization, it eliminates information redundancy and conflict in the time series dimension, generating a consistent and highly reliable comprehensive time series representation. Its core goal is to uncover potential correlations within multi-source time series data to support dynamic trend forecasting and decision optimization.

[0025] The First Farmland Dataset is a standardized nationwide data product on cultivated land resources with high temporal and spatial resolution, generated by integrating multi-source remote sensing data and land survey results. It aims to provide an authoritative benchmark for agricultural resource management, ecological protection, and food security assessment. Its core feature is that it breaks the limitations of traditional single data sources and enables the precise integration and analysis of multi-dimensional farmland information.

[0026] Step S200: Standardize the first farmland dataset according to a pre-established unified data model to obtain a second farmland dataset.

[0027] Normalization is a key step in data processing. It involves mathematically transforming the raw data to a uniform scale (e.g., a distribution with a mean of 0 and a standard deviation of 1) to eliminate dimensional differences between different features or indicators. This improves data comparability and analytical efficiency. Its core goal is to enable comprehensive comparison and analysis of heterogeneous data from multiple sources using the same benchmark through normalization.

[0028] The Second Farmland Dataset is an upgraded version of the First Farmland Dataset, focusing on integrating high-resolution time-series remote sensing data with real-time farmland sensor network data to build a dynamic monitoring and intelligent decision-making support platform covering cultivated land resources nationwide. Its core goal is to achieve real-time and precise farmland resource management by integrating multi-source heterogeneous data (such as crop growth, soil moisture conditions, and disaster warnings).

[0029] Step S300: extracting farmland boundaries and terrain features from the second farmland dataset. If the complexity of the terrain features exceeds a preset threshold, a digital elevation model is introduced to supplement the terrain data to generate a third farmland dataset.

[0030] The Terrain Complexity Index (TCI) is a comprehensive indicator that quantitatively describes the heterogeneity and complexity of surface morphology and structure. Through mathematical modeling, it combines terrain geometric features (such as slope, curvature, and relief) with the processes of internal and external forces to characterize the fragmentation and diversity of a region's surface. This index is both a digital abstraction of the terrain's spatial form and a dynamic mapping of its evolutionary processes (such as erosion and accumulation).

[0031] A Digital Elevation Model (DEM) is a mathematical model that discretizes the continuous spatial distribution of surface elevation through digital means. It stores land surface elevation information in the form of a regular grid (raster) or triangulated irregular network (TIN). It is core geographic data that represents the relief of terrain. Essentially, it is a two-dimensional elevation map of a three-dimensional geographic space, providing key terrain parameters for terrain analysis, hydrological simulation, and engineering construction.

[0032] The Third Farmland Dataset is a dynamic knowledge base of all elements of national cultivated land resources, built on the core results of the Third National Land Survey (Third National Land Survey). It deeply integrates soil profile physical and chemical properties, cultivated land production potential models, and the Rural Revitalization Big Data Platform. This dataset aims to achieve coordinated monitoring and full lifecycle management of cultivated land resources across four dimensions: "space, land, and network." Its core features include: leveraging the DiLei TuBan Database (DLTB) of the Third National Land Survey to express the four-dimensional attributes of cultivated land: land type, ownership, quality, and ecology. Furthermore, it leverages multi-scale models to support the delineation of cultivated land protection red lines, grain production capacity calculations, and ecological restoration decisions.

[0033] Step S400: Acquire dynamic change characteristics based on the third farmland dataset, use a time series analysis algorithm to determine the trend of crop growth and soil state changes, and generate a fourth farmland dataset.

[0034] Dynamic change characteristics refer to the patterns and processes by which land use types, spatial distribution, quantity, and quality evolve over time within a specific spatial and temporal context. These characteristics are characterized by both diversity and continuity across multiple dimensions and spatiotemporal scales, and require quantitative expression through remote sensing monitoring, time series data analysis, and model deduction. In the context of land management, these characteristics focus on the interactive response mechanisms between human activities and natural succession, such as the conversion of cultivated land to non-grain use, the expansion of construction land, and the restoration of ecological land.

[0035] Time series analysis algorithms are a set of mathematical methods that analyze discrete observational data collected in chronological order. By modeling its inherent temporal dependencies, trends, periodicity, and random perturbations, they aim to reveal the evolutionary patterns of systems and enable predictions. In remote sensing and land management, their core goal is to interpret the temporal variations in parameters such as land use / cover, surface temperature, and vegetation indices, supporting decision-making for dynamic farmland monitoring, ecological restoration assessment, and disaster warning.

[0036] The Fourth Farmland Dataset, based on the Third Farmland Dataset, is a next-generation intelligent farmland resource management database constructed by deeply integrating sub-meter high-resolution remote sensing imagery, time-series data on the planting distribution of staple crops, and an AIGC (Artificial Intelligence Generated Content)-driven soil optimization strategy generation model. Its core breakthrough lies in establishing a five-dimensional collaborative perception system encompassing air, space, space, and internet, enabling dynamic, closed-loop optimization of the four key elements of farmland resources: production, quality, ecology, and policy. This supports the transition of farmland protection from static regulation to intelligent decision-making.

[0037] Step S500: extract update features from the fourth farmland dataset, use an incremental update algorithm to fuse preset high-precision map data, and generate a dynamic farmland map.

[0038] Update features refer to a set of features that, in a dynamic spatiotemporal environment, enable real-time or periodic revisions of cultivated land resource attributes (such as land type, quality, and utilization patterns) through periodic data collection, rule base iteration, and multi-source information fusion. Their core value lies in ensuring the timeliness and accuracy of national land spatial decision-making. This feature captures the non-stationary evolution of cultivated land systems by decoupling the differences between a "static baseline" and "dynamic increments."

[0039] The incremental update algorithm is a dynamic data processing method that continuously optimizes and evolves the system state by processing only newly added or changed data, rather than the entire dataset. Its core lies in constructing a difference-driven update mechanism that uses the incremental difference between historical calculation results and new input data to make local adjustments, avoiding repeated calculations and significantly reducing time and space complexity. In spatiotemporal data management (such as cultivated land resource monitoring), this algorithm integrates online learning and data assimilation techniques to support real-time capture and precise modeling of dynamic features such as cultivated land classification changes and cropping pattern evolution.

[0040] Dynamic farmland maps are a spatiotemporally coupled intelligent decision-making tool. By integrating high-precision remote sensing imagery (such as GF-2 / 6 and Sentinel-1), IoT sensor networks, and multi-source spatiotemporal data assimilation technology, they enable minute-by-minute updates and three-dimensional visualization of farmland resource attributes (land type, fertility, crop distribution, etc.). Their core function is to depict the dynamic evolution of farmland systems (such as crop rotation patterns, abandoned land reclamation events, and the risk of non-grain conversion), supporting the transition of agricultural planning from static record management to real-time responsive decision-making.

[0041] Furthermore, in the automatic map drawing method provided in this embodiment, step S100 includes: Step S110: Collect farmland spectral data using a multispectral sensor and image data using a high-resolution camera, use a timestamp to record the acquisition time, and generate an initial data set containing spectral features and spatial information.

[0042] For example, when generating initial datasets using multispectral sensors and high-resolution cameras to collect farmland data, this can be implemented in rice-growing areas. Multispectral sensors collect spectral data from visible to near-infrared wavelengths, recording vegetation indices such as the Normalized Difference Vegetation Index (NDVI), which reflects crop health. High-resolution cameras capture field images, capturing spatial information such as plot boundaries and crop distribution.

[0043] Timestamp recording ensures data synchronization, for example, data is collected every 10 minutes from 8:00 to 10:00 on May 9, 2025.

[0044] The initial dataset contains spectral features such as NDVI value of 0.7 and crop coverage of 80% in image data, providing a basis for subsequent analysis.

[0045] Step S120: Using a time series analysis method, align the spectral data and image data in the initial data set according to the timestamps, fuse the multi-source features through a weighted average method, and generate a first farmland data set.

[0046] Specifically, when aligning data in time series analysis, spectral and image data can be matched by timestamp. For example, an NDVI value of 0.7 in spectral data at 8:00 AM is paired with an 80% green pixel ratio in image data at the same time.

[0047] If timestamps deviate slightly, linear interpolation can be used to fill in the gaps and ensure data consistency. This aligned dataset can fully reflect the characteristics of farmland over time, such as the changing trends of crop growth stages from tillering to heading, which facilitates precision agriculture management.

[0048] In one embodiment, multi-source features are fused using a weighted average method to generate a first farmland dataset. The NDVI value of the spectral data reflects crop health and is weighted at 0.6; the coverage of the image data reflects spatial distribution and is weighted at 0.4. For example, the NDVI value of a field at 8:00 a.m. is 0.7, and the coverage is 80%. The fused feature value is 0.7 × 0.6 + 80 × 0.4 = 74. This value comprehensively reflects the crop status and spatial information. The weight can be adjusted according to the crop type. For example, rice is more dependent on spectral features, so the weight can be increased to 0.7. The fused dataset reduces redundancy and improves analysis efficiency. It should be noted that the use of fused datasets can improve the accuracy of farmland monitoring. For example, a field with a fused feature value of 74 may indicate good crop growth, but if the feature value drops to 60 at a later time point, it may indicate insufficient water and the need for irrigation adjustment.

[0049] This method, by integrating spectral and image information, enhances the ability to perceive farmland dynamics, supports precise fertilization and irrigation decisions, and reduces resource waste. For example, in rice field monitoring, the fused dataset can be used for early warning of pests and diseases. If the fused feature value of a particular field plot continues to decline, for example, from 74 to 50, and the image shows localized yellowing, it can be inferred that rice blast may occur. Combined with historical data analysis, disease trends can be identified and preventive measures can be implemented in advance. In this way, the fused dataset not only improves monitoring accuracy but also provides data support for agricultural decision-making, ensuring yields.

[0050] In one embodiment, time series analysis can be combined with meteorological data to further enrich the dataset. For example, by adding 10 mm of rainfall data at 8:00 AM and analyzing it together with spectral and image data, the impact of rainfall on crop growth can be determined. The fused dataset may reveal that the NDVI value rose to 0.75 after the rainfall, indicating improved crop health. This comprehensive analysis supports more informed farmland management decisions.

[0051] It can be understood that the above method constructs a high-precision first farmland dataset through multi-source data fusion and time series analysis, significantly improving the efficiency of agricultural monitoring.

[0052] The fused eigenvalues ​​intuitively reflect the status of farmland, and time series analysis captures dynamic changes, jointly supporting the application of precision agriculture, optimizing resource utilization, and ensuring food security.

[0053] Furthermore, in the automatic map drawing method provided in this embodiment, step S200 includes: Step S210: Obtain a first farmland dataset, use a data cleaning tool to process missing values ​​and outliers, and if a data point exceeds a preset threshold, replace it with the time series mean to obtain a cleaned dataset.

[0054] For example, after acquiring the first farmland dataset, data cleaning is a key step in ensuring data quality. Data cleaning tools can identify missing and outliers, such as when the NDVI value at a certain time point is null or abnormally high, reaching 1.2, outside the normal range of 0 to 1. During cleaning, if a missing NDVI value at 8:00 AM is detected for a particular field, it can be filled with the mean of 0.65 from the preceding and subsequent time points, such as 7:50 AM and 8:10 AM. If an outlier value of 1.2 is detected, it can be replaced with the time series mean of 0.7. This cleaned dataset is complete and reliable, laying the foundation for subsequent processing.

[0055] Step S220: extract spectral features and spatial features from the cleaned dataset according to the field definition of the unified data model, align them into a standard format using a format conversion tool, and obtain a feature-aligned dataset.

[0056] In one possible implementation, spectral and spatial features are extracted based on a unified data model. The data model defines fields such as NDVI and coverage. From the cleaned dataset, the NDVI value of 0.7 at 8:00 AM and the image coverage of 80% are extracted.

[0057] Using a format conversion tool, we aligned the spectral data's timestamp format from "2025-05-09 08:00:00" to standard UTC (Coordinated Universal Time). The spatial resolution of the image data was standardized to 1 meter to ensure feature alignment. This feature-aligned dataset maintains temporal and spatial consistency, facilitating fusion analysis.

[0058] Step S230: For the feature alignment dataset, a data verification tool is used to check the data integrity and consistency. If data is missing, it is filled by interpolation to obtain a verified dataset.

[0059] For example, the data verification tool checks the integrity of the feature-aligned dataset. If the coverage data for a particular field at 9:00 AM is missing, the verification tool uses linear interpolation to extrapolate the coverage at 9:00 AM to 80% based on the coverage values ​​of 78% and 82% at 8:50 and 9:10 AM, respectively. Consistency checks ensure that NDVI values ​​match coverage trends; for example, as NDVI increases, coverage should also increase. This verified dataset ensures that there are no missing values ​​and that data logic is consistent, improving analytical reliability.

[0060] Step S240: fuse the multi-source features in the verified dataset using a weighted average algorithm, determine the fused feature values, and obtain a second farmland dataset.

[0061] In one possible implementation, the fused, validated datasets are averaged using a weighted averaging algorithm. The spectral NDVI feature is weighted at 0.6, and the spatial coverage feature is weighted at 0.4. For example, if a field has an NDVI value of 0.7 at 8:00 AM and a coverage of 80%, the fused feature value is 0.7 × 0.6 + 80 × 0.4 = 74. This fused second farmland dataset comprehensively reflects crop health and spatial distribution, reducing data redundancy.

[0062] The weight can be adjusted based on the crop type. For example, if coverage is more important for corn, the weight can be set to 0.5. The fused dataset provides unified features for precision agriculture, making it easier to monitor crop status.

[0063] It should be noted that the above process generates a high-quality second farmland dataset through cleaning, feature extraction, alignment, validation, and fusion. Cleaning ensures data integrity, feature alignment unifies the format, validation ensures consistency, and fusion integrates multi-source information. Each step is closely linked and logically coherent, collectively ensuring accurate and reliable farmland monitoring and providing a data foundation for agricultural management.

[0064] Preferably, in the automatic map drawing method provided in this embodiment, step S300 includes: Step S310: Obtain farmland boundary data from the second farmland dataset, use edge detection tools to extract boundary lines, and if the continuity of the boundary lines is lower than a preset threshold, fill in the discontinuities by interpolation to obtain a boundary extraction dataset.

[0065] After acquiring the second farmland dataset, extracting farmland boundary data is a key step. The edge detection tool identifies field boundaries by analyzing changes in pixel values ​​within the image data. For example, if a field image shows a sudden change in NDVI value from 0.7 to 0.2 at the boundary, the tool draws the boundary line accordingly. If the boundary continuity falls below a threshold, such as a 5-meter gap in a section of the boundary due to vegetation obstruction, interpolation can be used to infer and fill in the gaps based on the coordinates of adjacent boundary points, forming a complete boundary extraction dataset. This ensures the accuracy of the field boundaries and facilitates subsequent terrain analysis.

[0066] Step S320: Obtain terrain feature data from the boundary extraction dataset, calculate slope and elevation distribution using terrain analysis tools, and if the complexity of the terrain features exceeds a preset threshold, supplement the high-resolution terrain data with a digital elevation model to obtain a terrain enhancement dataset.

[0067] In one possible implementation, terrain feature data is extracted from a boundary extraction dataset. The terrain analysis tool calculates slope and elevation based on a digital elevation model. For example, the elevation within a field boundary gradually changes from 50 meters to 55 meters, with an average slope of 5 degrees.

[0068] If the terrain is complex, such as when slope changes frequently exceed a threshold, a high-resolution digital elevation model can be introduced to supplement the 1-meter-resolution terrain data to form a terrain enhancement dataset. This approach can refine terrain features and improve the accuracy of field delineation.

[0069] Step S330: For the terrain enhancement dataset, a data fusion tool is used to weighted average boundary features and terrain features, and the dataset is aligned to a standard format using a format conversion tool to obtain a feature fusion dataset.

[0070] For terrain enhancement datasets, the Data Fusion tool combines boundary and terrain features using a weighted average. Boundary features reflect the shape of the field, while terrain features describe the slope distribution. For example, if a field has a boundary length of 200 meters and an average slope of 5 degrees, the weights are set to 0.5 and 0.5, and the resulting fusion generates a composite feature value.

[0071] The format conversion tool aligns the data to a standard format, such as converting elevation data timestamps to UTC and adjusting the spatial resolution to 1 meter, to produce a feature-fused dataset. This ensures consistency across multiple data sources and facilitates subsequent verification.

[0072] Step S340: obtain verification data from the feature fusion dataset, use a data verification tool to check the integrity and consistency, and if the verification data is found to be missing, fill it in using linear interpolation to obtain a third farmland dataset.

[0073] In one possible implementation, verification data is obtained from the feature fusion dataset, and a data verification tool checks its completeness and consistency. For example, if slope data for a particular field at a certain point in time is missing, the verification tool uses linear interpolation to infer the missing value as 5 degrees based on slopes of 4 degrees and 6 degrees at the previous and subsequent points. If a boundary length mismatches with the terrain features—for example, if the boundary is extended but the slope remains unchanged—the tool flags the anomaly and corrects it. The verified third farmland dataset is complete and consistent, providing a reliable data foundation for farmland management.

[0074] As you can see, the above process generates a high-quality third-level farmland dataset through boundary extraction, terrain analysis, feature fusion, and data verification. Each step is closely linked: boundary extraction defines the field boundaries, terrain enhancement refines features, multi-source information is integrated and unified, and verification ensures data reliability. This logically rigorous processing method provides accurate terrain and boundary data support for precision agriculture.

[0075] Furthermore, in the automatic map drawing method provided in this embodiment, step S400 includes: Step S410: Acquire crop growth data and soil state data from the third farmland dataset, and use a data extraction tool to separate crop height, leaf area index, and soil moisture and nutrient content data in the time dimension to obtain a dynamic feature dataset.

[0076] For example, when extracting crop growth and soil condition data from the third farmland dataset, the data extraction tool analyzes multi-temporal imagery and sensor data to isolate information such as crop height, leaf area index, soil moisture, and nutrient content. Crop height can be calculated using lidar point cloud data; the average crop height for a particular field at a given point in time is 0.8 meters. The leaf area index, obtained through spectral image analysis, is 3.5. Soil moisture sensors record surface moisture of 25% and nitrogen content of 150 mg / kg. This data is organized by time to form a dynamic feature dataset that reflects the dynamic changes in crop and soil conditions, providing a foundation for subsequent analysis.

[0077] Step S420: For the dynamic feature dataset, a time series analysis tool is used to calculate the periodic change characteristics of crop growth and soil state. If the fluctuation amplitude of the change characteristics exceeds a preset threshold, the data is smoothed by a moving average method to obtain a smoothed feature dataset.

[0078] In one possible implementation, a time series analysis tool is used to identify periodic variations in crop growth and soil conditions within a dynamic feature dataset. The analysis tool uses Fourier transforms to identify periods of variation. For example, it may discover that crop height in a particular field increases periodically by 0.2 meters every 30 days. If the fluctuation is excessive, such as a sudden change in soil moisture from 20% to 35% within 7 days, exceeding a preset threshold, the data is smoothed using a moving average. This smoothing results in a more gradual change in the moisture data, generating a smoothed feature dataset. This smoothing process effectively reduces noise interference and improves the reliability of data analysis.

[0079] Step S430: Obtain time series data from the smoothed feature data set, and use an autoregressive moving average algorithm to calculate the changing trends of crop growth stages and soil conditions to obtain a trend analysis data set.

[0080] For example, after acquiring time series data from a smoothed feature dataset, the autoregressive moving average algorithm is used to predict crop growth stages and soil condition trends. Based on historical data analyzing crop height growth rates, the algorithm infers that crops in a particular field are in the jointing stage and are expected to enter the heading stage 15 days later. A trend of decreasing soil moisture indicates a need for irrigation strategy adjustments. The resulting trend analysis dataset clearly reflects the dynamic evolution of crops and soil, providing a basis for decision-making in agricultural management.

[0081] Step S440: For the trend analysis dataset, a data format conversion tool is used to align the crop growth trend and the soil state trend into a standard format to obtain a fourth farmland dataset.

[0082] In one possible implementation, a data format conversion tool aligns crop growth trends and soil condition trends into a standard format for trend analysis datasets. For example, crop height trend data timestamps are standardized to UTC format, the spatial resolution is adjusted to 1 meter, and soil moisture trend data is converted to a JSON structure.

[0083] The aligned format of the fourth farmland dataset is consistent and compatible, making it easier to integrate with other agricultural systems and improve data sharing efficiency.

[0084] As you can see, the above process, through data extraction, time series analysis, trend prediction, and format conversion, generates a high-quality fourth-level farmland dataset. Each step is tightly integrated: data extraction provides multidimensional dynamic information, time series analysis reveals cyclical patterns, trend prediction supports management decisions, and format conversion ensures data interoperability. This logically rigorous processing approach provides reliable crop and soil dynamic data support for precision agriculture.

[0085] Furthermore, the automatic map drawing method provided in this embodiment includes step S500: Step S510: extract crop distribution data and soil property data from the fourth farmland dataset, and use a data extraction tool to separate the crop type, coverage rate, and soil nutrients and moisture data in the spatial dimension to obtain a spatial feature dataset.

[0086] For example, when extracting crop distribution data and soil property data from the fourth farmland dataset, the data extraction tool separates information in the spatial dimension by analyzing high-resolution satellite imagery and soil sensor data.

[0087] Crop types can be classified using spectral characteristics. For a particular field, wheat coverage accounts for 60% and corn for 30%. Coverage, calculated based on image pixel statistics, is 85% for wheat. Soil nutrient data, acquired from sensors, show a nitrogen content of 140 mg / kg and soil moisture of 22% in a particular area. These data are organized into spatial grids to form a spatial feature dataset, reflecting the spatial heterogeneity of crops and soils within a field, providing a foundation for precision agriculture management.

[0088] Step S520: For the spatial feature dataset, an incremental update algorithm is used to match the preset high-precision map data. If the spatial coordinate deviation between the crop distribution data and the high-precision map exceeds a preset threshold, the coordinates are adjusted by spatial interpolation to obtain an aligned feature dataset.

[0089] In one possible implementation, an incremental update algorithm is used to match spatial feature datasets to high-precision map data. The incremental update algorithm compares the spatial coordinates of crop distribution data with the reference coordinates of the map to detect deviations. For example, the distribution coordinates of wheat in a particular field may deviate by 2 meters from the high-precision map, exceeding a preset threshold of 1 meter.

[0090] Spatial interpolation adjusts coordinates based on a weighted average of neighboring grid points to generate an aligned feature dataset. This adjustment reduces coordinate deviation to as little as 0.5 meters, ensuring spatial consistency between crop distribution and the map, facilitating subsequent integration and analysis.

[0091] Step S530: Based on the aligned feature dataset, a data fusion tool is used to merge the crop distribution data, soil property data, and spatial grid data of the high-precision map to obtain a fused feature dataset.

[0092] For example, for aligned feature datasets, the data fusion tool merges crop distribution data and soil property data with the 1-meter resolution grid of the high-precision map.

[0093] The fusion process integrates wheat coverage, nitrogen content, and moisture data through spatial overlay to generate a fused feature dataset. For example, a grid point shows a wheat coverage of 80%, a nitrogen content of 145 mg / kg, and a moisture content of 20%, precisely corresponding to the map grid. This fused dataset is spatially complete and supports precision agricultural decision-making.

[0094] Step S540: By fusing the feature data sets, a map generation tool is used to map the crop distribution data and soil property data to a spatial grid to obtain a dynamic farmland map.

[0095] In one possible implementation, a map generation tool fuses feature datasets and maps the data onto a spatial grid, generating dynamic farmland maps. For example, the tool renders the distribution of wheat and corn as a color-coded map, with green representing wheat, yellow representing corn, and a blue gradient representing soil moisture. A particular field map shows that wheat is concentrated in the north, where moisture levels are low, indicating that irrigation is a priority. Dynamic farmland maps intuitively display the spatial patterns of crops and soil, supporting precision fertilization and irrigation management.

[0096] It's important to note that the above process generates dynamic farmland maps through extraction, matching, fusion, and mapping. Each step is logically rigorous. Spatial feature extraction provides diverse information, incremental updates ensure coordinate accuracy, data fusion improves grid integrity, and map generation supports intuitive decision-making. This approach provides spatially refined data support for agricultural management.

[0097] The present invention relates to an automatic map drawing system, which is used to implement the above-mentioned automatic map drawing method. The automatic map drawing system includes a first generation module, a second generation module, a third generation module, a fourth generation module and a fifth generation module, wherein the first generation module obtains multi-source farmland data including multispectral sensor data and high-resolution images, and generates a first farmland data set through time series fusion; the second generation module standardizes the first farmland data set according to a pre-established unified data model to obtain a second farmland data set; the third generation module extracts farmland boundaries and terrain features from the second farmland data set, and if the complexity of the terrain features exceeds a preset threshold, introduces a digital elevation model to supplement the terrain data to generate a third farmland data set; the fourth generation module obtains dynamic change features based on the third farmland data set, uses a time series analysis algorithm to judge the trend of crop growth and soil state changes, and generates a fourth farmland data set; the fifth generation module extracts update features from the fourth farmland data set, uses an incremental update algorithm to fuse preset high-precision map data, and generates a dynamic farmland map.

[0098] Furthermore, in the automatic map drawing system provided in this embodiment, the first generation module includes a first generation unit and a second generation unit, wherein the first generation unit is used to collect farmland spectral data and image data through a multispectral sensor and a high-resolution camera, and use a timestamp to record the collection time to generate an initial data set containing spectral characteristics and spatial information; the second generation unit is used to use a time series analysis method to align the spectral data and image data in the initial data set according to the timestamp, and fuse multi-source features through a weighted average method to generate a first farmland data set.

[0099] Preferably, in the automatic map drawing system provided in this embodiment, the second generation module includes a first acquisition unit, a second acquisition unit, a third acquisition unit and a fourth acquisition unit, wherein the first acquisition unit is used to acquire the first farmland data set, and use a data cleaning tool to process missing values ​​and outliers. If a data point exceeds a preset threshold, it is replaced by the time series mean to obtain a cleaned data set; the second acquisition unit is used to extract spectral features and spatial features from the cleaned data set according to the field definition of the unified data model, and align them to a standard format through a format conversion tool to obtain a feature-aligned data set; the third acquisition unit is used to check the data integrity and consistency of the feature-aligned data set using a data verification tool. If data is found to be missing, it is filled by interpolation to obtain a verified data set; the fourth acquisition unit is used to fuse the multi-source features in the verified data set through a weighted average algorithm, determine the fused feature value, and obtain the second farmland data set.

[0100] Furthermore, in the automatic map drawing system provided in this embodiment, the third generation module includes a fifth acquisition unit, a sixth acquisition unit, a seventh acquisition unit and an eighth acquisition unit, wherein the fifth acquisition unit is used to obtain farmland boundary data from the second farmland dataset, and use an edge detection tool to extract the boundary line. If the continuity of the boundary line is lower than a preset threshold, the discontinuity is filled by interpolation to obtain a boundary extraction dataset; the sixth acquisition unit is used to obtain terrain feature data from the boundary extraction dataset, and use a terrain analysis tool to calculate the slope and elevation distribution. If the complexity of the terrain feature exceeds the preset threshold, the high-resolution terrain data is supplemented by a digital elevation model to obtain a terrain enhancement dataset; the seventh acquisition unit is used to use a data fusion tool to weightedly average the boundary features and terrain features of the terrain enhancement dataset, and align them into a standard format through a format conversion tool to obtain a feature fusion dataset; the eighth acquisition unit is used to obtain verification data from the feature fusion dataset, and use a data verification tool to check the integrity and consistency. If the verification data is found to be missing, it is filled by linear interpolation to obtain a third farmland dataset.

[0101] Compared with the existing technology, the automatic map drawing method and system provided by this embodiment achieves accurate updating of farmland information through multi-source data fusion and time series analysis. This embodiment first obtains multispectral sensor data and high-resolution images, performs time series fusion and standardization processing. Then, the farmland boundaries and terrain features are extracted, and a digital elevation model is introduced to supplement complex terrain data when necessary. Then, a time series analysis algorithm is used to determine the trend of crop growth and soil state changes, and dynamic change features are extracted. Finally, an incremental update algorithm is used to fuse the new features with high-precision map data to generate a dynamic farmland map. This embodiment can capture farmland change information in a timely manner, improve the efficiency and accuracy of map updates, and provide reliable data support for precision agricultural management and decision-making.

[0102] Although preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they are aware of the basic inventive concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the invention. Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the invention. Thus, the present invention is intended to include such changes and modifications as fall within the scope of the claims and their equivalents.

Claims

1. A method for automatic map drawing, characterized in that: The following steps are involved: Acquire multi-source farmland data including multispectral sensor data and high-resolution images, and generate the first farmland dataset through time series fusion; performing standardization processing on the first farmland dataset according to a pre-established unified data model to obtain a second farmland dataset; Extracting farmland boundaries and terrain features from the second farmland dataset; if the complexity of the terrain features exceeds a preset threshold, introducing a digital elevation model to supplement the terrain data to generate a third farmland dataset; obtaining dynamic change characteristics based on the third farmland dataset, using a time series analysis algorithm to determine crop growth and soil state change trends, and generating a fourth farmland dataset; Update features are extracted from the fourth farmland dataset, and preset high-precision map data are fused using an incremental update algorithm to generate a dynamic farmland map.

2. The method for automatically drawing a map according to claim 1, wherein: The step of acquiring multi-source farmland data including multispectral sensor data and high-resolution images and generating a first farmland dataset by time series fusion includes: The spectral data of farmland are collected by multispectral sensors and image data by high-resolution cameras. The acquisition time is recorded by timestamp to generate an initial dataset containing spectral characteristics and spatial information. A time series analysis method is used to align the spectral data and image data in the initial data set according to timestamps, and multi-source features are fused through a weighted average method to generate the first farmland data set.

3. The method for automatically drawing a map according to claim 1, wherein: The step of performing standardization processing on the first farmland dataset according to the pre-established unified data model to obtain the second farmland dataset includes: Obtain the first farmland dataset, use a data cleaning tool to process missing values ​​and outliers, and if a data point exceeds a preset threshold, replace it with the time series mean to obtain a cleaned dataset; According to the field definition of the unified data model, spectral features and spatial features are extracted from the cleaned dataset and aligned to a standard format using a format conversion tool to obtain a feature-aligned dataset. For the feature alignment dataset, a data verification tool is used to check the data integrity and consistency. If data is missing, it is filled by interpolation to obtain a verified dataset; The multi-source features in the verified dataset are fused through a weighted average algorithm, and the fused feature values ​​are determined to obtain the second farmland dataset.

4. The method for automatically drawing a map according to claim 1, wherein: The steps of extracting farmland boundaries and terrain features from the second farmland dataset, and introducing a digital elevation model to supplement the terrain data if the complexity of the terrain features exceeds a preset threshold, to generate a third farmland dataset include: Acquire farmland boundary data from the second farmland dataset, extract boundary lines using an edge detection tool, and if the continuity of the boundary lines is lower than a preset threshold, fill in the discontinuities using an interpolation method to obtain a boundary extraction dataset; Obtaining terrain feature data from the boundary extraction dataset, calculating slope and elevation distribution using terrain analysis tools, and if the complexity of the terrain features exceeds a preset threshold, supplementing the high-resolution terrain data with a digital elevation model to obtain a terrain enhancement dataset; For the terrain enhancement dataset, a data fusion tool is used to weighted average the boundary features and the terrain features, and the data are aligned into a standard format using a format conversion tool to obtain a feature fusion dataset; Verification data is obtained from the feature fusion dataset, and a data verification tool is used to check the integrity and consistency. If the verification data is found to be missing, it is filled in by linear interpolation to obtain a third farmland dataset.

5. The method for automatically drawing a map according to claim 1, wherein: The steps of obtaining dynamic change characteristics based on the third farmland dataset and using a time series analysis algorithm to determine the trend of crop growth and soil state changes to generate a fourth farmland dataset include: Acquire crop growth data and soil state data from the third farmland dataset, and use a data extraction tool to separate crop height, leaf area index, and soil moisture and nutrient content data in the time dimension to obtain a dynamic feature dataset; For the dynamic feature dataset, a time series analysis tool is used to calculate the periodic change characteristics of crop growth and soil state. If the fluctuation amplitude of the change characteristics exceeds a preset threshold, the data is smoothed using a moving average method to obtain a smoothed feature dataset. Acquiring time series data from the smoothed feature data set, calculating the changing trends of crop growth stages and soil conditions using an autoregressive moving average algorithm, and obtaining a trend analysis data set; For the trend analysis dataset, a data format conversion tool is used to align the crop growth trend and the soil state trend into a standard format to obtain a fourth farmland dataset.

6. The method for automatically drawing a map according to claim 1, wherein: The steps of extracting update features from the fourth farmland dataset and fusing preset high-precision map data with an incremental update algorithm to generate a dynamic farmland map include: Extracting crop distribution data and soil property data from the fourth farmland dataset, and using a data extraction tool to separate crop type, coverage, and soil nutrient and moisture data in a spatial dimension to obtain a spatial feature dataset; For the spatial feature dataset, an incremental update algorithm is used to match the preset high-precision map data. If the spatial coordinate deviation between the crop distribution data and the high-precision map exceeds a preset threshold, the coordinates are adjusted through spatial interpolation to obtain an aligned feature dataset; According to the aligned feature data set, a data fusion tool is used to merge the crop distribution data, the soil property data and the spatial grid data of the high-precision map to obtain a fused feature data set; By using the fused feature data set, a map generation tool is used to map the crop distribution data and soil property data to a spatial grid to obtain a dynamic farmland map.

7. A system for automatically drawing a map, for implementing the method for automatically drawing a map according to any one of claims 1 to 6, characterized in that: The automatic map drawing system includes: The first generation module acquires multi-source farmland data including multispectral sensor data and high-resolution images, and generates the first farmland dataset through time series fusion; a second generation module, performing standardization processing on the first farmland dataset according to a pre-established unified data model to obtain a second farmland dataset; a third generation module, extracting farmland boundaries and terrain features from the second farmland dataset; and introducing a digital elevation model to supplement the terrain data to generate a third farmland dataset if the complexity of the terrain features exceeds a preset threshold; a fourth generation module, which obtains dynamic change characteristics based on the third farmland dataset, uses a time series analysis algorithm to determine the trend of crop growth and soil state changes, and generates a fourth farmland dataset; The fifth generation module extracts update features from the fourth farmland dataset, adopts an incremental update algorithm to fuse preset high-precision map data, and generates a dynamic farmland map.

8. The automated map drawing system according to claim 7, wherein: The first generation module includes: The first generation unit is used to collect farmland spectral data and image data using a multispectral sensor and a high-resolution camera, record the collection time using a timestamp, and generate an initial data set containing spectral characteristics and spatial information; The second generating unit is configured to align the spectral data and image data in the initial data set according to timestamps using a time series analysis method, fuse multi-source features using a weighted average method, and generate the first farmland data set.

9. The automated map drawing system according to claim 7, wherein: The second generation module includes: a first acquisition unit, configured to acquire the first farmland dataset, process missing values ​​and outliers using a data cleaning tool, and if a data point exceeds a preset threshold, replace it with a time series mean to obtain a cleaned dataset; The second acquisition unit is used to extract spectral features and spatial features from the cleaned data set according to the field definition of the unified data model, and align them into a standard format through a format conversion tool to obtain a feature-aligned data set; a third acquisition unit, configured to use a data verification tool to check the integrity and consistency of the feature alignment dataset, and to fill in missing data by interpolation to obtain a verified dataset; The fourth acquisition unit is used to fuse the multi-source features in the verified data set through a weighted average algorithm, determine the fused feature value, and obtain the second farmland data set.

10. The automated map drawing system according to claim 7, wherein: The third generation module includes: a fifth acquisition unit, configured to acquire farmland boundary data from the second farmland dataset, extract boundary lines using an edge detection tool, and fill in discontinuities using an interpolation method if the continuity of the boundary lines is lower than a preset threshold, thereby obtaining a boundary extraction dataset; a sixth acquisition unit, configured to acquire terrain feature data from the boundary extraction dataset, calculate slope and elevation distribution using a terrain analysis tool, and, if the complexity of the terrain features exceeds a preset threshold, supplement the high-resolution terrain data with a digital elevation model to obtain a terrain enhancement dataset; a seventh acquisition unit, configured to perform weighted averaging of the boundary features and the terrain features using a data fusion tool on the terrain enhancement dataset, and align the dataset into a standard format using a format conversion tool to obtain a feature fusion dataset; The eighth acquisition unit is used to obtain verification data from the feature fusion dataset, use a data verification tool to check the integrity and consistency, and if the verification data is found to be missing, fill it by linear interpolation to obtain a third farmland dataset.

Images