Ploughing plot dynamic updating method and product

Abstract

The invention discloses a method and product for dynamically updating a cultivated land parcel, and relates to the field of cultivated land right affirmation.The method comprises the steps that historical remote sensing image data pairs of rural cultivated land and corresponding farmland boundary data are obtained, and a data set is constructed; according to the data set, based on a super-resolution network and a parcel segmentation network, constructing a super-resolution-parcel joint model; determining current super-resolution data and current field data of the rural cultivated land based on the trained super-resolution-field joint model according to a current medium-resolution remote sensing image of the rural cultivated land; further determining current field piece vector data of the rural cultivated land; performing association matching on the current field piece vector data and the ownership database, and performing right confirmation on the rural cultivated land; according to the method, the land right determination precision and efficiency can be improved, and the cultivated land parcels are dynamically updated regularly.

Description

Technical Field

[0001] The present application relates to the field of farmland rights confirmation, and in particular to a method and product for dynamically updating farmland plots. Background Art

[0002] In recent years, with the rapid development of remote sensing technology, deep learning, and geographic information systems (GIS), automation technologies have been gradually introduced into agricultural informatization and land management to improve efficiency. In the field of farmland rights confirmation, existing technologies mainly include medium- and low-resolution satellite remote sensing imagery, super-resolution reconstruction, and instance-based segmentation models. Medium- and low-resolution satellite remote sensing imagery (such as Landsat and Sentinel-2) offers wide coverage and is relatively low-cost, making it widely used for large-scale farmland monitoring. However, its resolution (10-30 meters) makes it difficult to accurately extract field boundaries, especially in complex scenes with intersecting ridges and crop obstruction, where boundary blurring is particularly prominent. While drones can acquire high-resolution imagery (0.1-2 meters), they are limited by flight costs, data processing cycles, and reliance on manual interpretation, making large-scale application difficult. Traditional image segmentation methods (such as threshold segmentation and edge detection algorithms) have high false positive rates in complex farmland scenarios and cannot meet the requirements for high-precision land rights confirmation. Existing research on super-resolution reconstruction techniques (such as ESRGAN and SRCNN) has primarily focused on general scenarios and lacks optimization for farmland texture features, resulting in artifacts and loss of detail in reconstructed images. While instance segmentation models (such as Mask R-CNN and U-Net) have seen initial application in farmland segmentation, their performance significantly degrades with low-resolution imagery. Furthermore, they lack deep integration with land ownership data, making them ineffective in directly supporting land ownership confirmation processes.

[0003] Currently, land rights confirmation is still primarily a manual process, relying on field measurements, paper archive management, and manual verification. This is inefficient, costly, and susceptible to subjective factors. Some semi-automated systems attempt to combine GIS and remote sensing technologies, but data update cycles are long (typically one to two years), making dynamic management impossible. They also lack a reliable evidence storage mechanism, making them difficult to address the needs of land transfer and dispute arbitration. Despite certain technological advances, rural land rights confirmation still faces key bottlenecks: First, there is a conflict between image resolution and boundary accuracy. Low-cost satellite imagery has insufficient resolution, resulting in an error rate exceeding 10% in extracting field boundaries. High-resolution drone imagery is expensive and difficult to update frequently, making it uneconomical for large-scale land rights confirmation. Second, algorithms lack adaptability. General super-resolution models are not optimized for farmland edges (such as ridges and ditches), resulting in artifacts and blurred details in reconstructed images. Segmentation models are susceptible to interference from crop cover and shadows in complex farmland scenarios, resulting in high rates of missed and false positives, especially with significant performance fluctuations during crop growth periods. Third, data updates for land rights confirmation lag. The existing system relies on periodic manual surveys, with data updates taking up to one to two years. This system fails to promptly reflect changes in cultivated land (such as encroachment, abandonment, and land consolidation), leading to a disconnect between ownership information and actual conditions and increasing the risk of disputes. Fourthly, there are challenges with data credibility and traceability. Land ownership data is often stored in centralized databases, which pose a risk of tampering and lack an immutable record of changes. This makes it difficult to support the data authenticity and traceability requirements of scenarios such as land transfers and dispute arbitration.

[0004] Therefore, based on the above problems, there is an urgent need to provide a method for dynamically updating cultivated land plots, which can improve the accuracy and efficiency of land rights confirmation and regularly update rural cultivated land plots. Summary of the Invention

[0005] The purpose of this application is to provide a method and product for dynamically updating cultivated land plots, which can improve the accuracy and efficiency of land rights confirmation and regularly update rural cultivated land plots.

[0006] To achieve the above objectives, this application provides the following solutions: In a first aspect, the present application provides a method for dynamically updating a farmland plot, the method comprising: Acquire historical remote sensing image data pairs of rural cultivated land and corresponding farmland boundary data, and construct a data set; the remote sensing image data pairs include: medium-resolution remote sensing images and high-resolution remote sensing images; According to the data set, a super-resolution-field joint model is constructed based on a super-resolution network and a field segmentation network; the super-resolution-field joint model is used to output super-resolution data and field data corresponding to rural cultivated land; the super-resolution-field joint model includes: an encoder, a decoder, an output head and a loss function; the encoder includes a plurality of positive pyramid cascaded convolution blocks, and is used to extract image features of remote sensing image data pairs; the decoder includes a plurality of inverted pyramid cascaded convolution blocks, and is used to decode the image features; the convolution blocks of the encoder and the decoder are symmetrically arranged and the number is the same; the loss function includes: mean absolute error loss, adversarial loss, edge enhancement loss, cross entropy loss and loss; According to the current medium-resolution remote sensing images of rural cultivated land, the current super-resolution data and current field data of rural cultivated land are determined based on the trained super-resolution-field joint model; Determine the current field vector data of rural cultivated land based on the current super-resolution data and the current field data; The current field vector data is associated and matched with the ownership database to confirm the ownership of rural arable land.

[0007] Optionally, the step of obtaining a pair of historical remote sensing image data of rural cultivated land and corresponding farmland boundary data and constructing a data set specifically includes: Preprocessing historical remote sensing image data to determine farmland areas; the preprocessing process includes: radiometric correction, geometric registration, and farmland area extraction; the farmland area extraction process includes: extracting regions of interest in the farmland area using cultivated land classification in a land classification dataset, and clipping non-farmland areas; Aligning the spatial coordinates of the farmland area and the ownership database to obtain an aligned remote sensing image data pair; A dataset is constructed based on the aligned remote sensing image data pairs and the corresponding farmland boundary data.

[0008] Optionally, the process of determining the loss function includes: Using the formula Determine the over-resolution loss ; Using the formula Determine segmentation loss ; Using the formula Determine the loss function ; in, is the mean absolute error loss coefficient, is the mean absolute error loss, To combat the loss factor, To combat losses, is the edge enhancement loss coefficient, is the edge enhancement loss, is the cross entropy loss coefficient, is the cross entropy loss, for Loss coefficient, for loss.

[0009] Optionally, determining the current field vector data of the rural cultivated land based on the current super-resolution data and the current field data specifically includes: Align and correct the current super-resolution data and the current field data to obtain aligned field data; The aligned field data are vectorized to obtain current field vector data of the rural cultivated land.

[0010] Optionally, the correlating and matching the current field vector data with the ownership database to confirm the ownership of rural cultivated land specifically includes: Overlapping the current field vector data with the longitude and latitude coordinates of the ownership database; and determining the area error between the current field vector data and the ownership database; The matching rule when the latitude and longitude coordinate overlap is greater than 90% and the area error is less than 5% is used as the optimal matching rule, and a matching result is obtained; The matching results are used to confirm the ownership of rural arable land.

[0011] Optionally, the current field vector data is associated and matched with the ownership database to confirm the ownership of the rural cultivated land, and then further includes: Based on the results of rural cultivated land title confirmation, determine whether there are any abnormal situations; if so, an alarm is issued to prompt manual intervention; if not, the results of rural cultivated land title confirmation are stored; the abnormal situations include: the degree of change in field area exceeds the area change threshold, field ownership matching conflicts and newly added unregistered cultivated land.

[0012] In a second aspect, the present application provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps of any one of the above-described methods for dynamically updating cultivated land plots.

[0013] According to the specific embodiments provided in this application, this application has the following technical effects: The present application provides a method and product for dynamically updating cultivated land plots. By integrating a super-resolution network and a field segmentation network, a super-resolution-field joint model is constructed for unified optimization training, and the loss function is adaptively improved to make it more suitable for farmland scenes, solving the problem of blurred boundaries caused by low-resolution images, enhancing farmland edge information, and improving the extraction accuracy of field data; the field vector data and the ownership database are correlated and matched to generate field vector data with spatial position, area and ownership information, further improving the accuracy and efficiency of land rights confirmation. When confirming the ownership of rural cultivated land, each input is the current medium-resolution remote sensing image of the rural cultivated land, that is, the current rural cultivated land is confirmed on a regular basis, achieving the effect of dynamically updating the rural cultivated land. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0015] Figure 1 This is a flow chart of a method for dynamically updating cultivated land plots in one embodiment of the present application; Figure 2 This is a specific flow chart of a method for dynamically updating cultivated land plots in one embodiment of the present application; Figure 3 Schematic diagram of historical remote sensing image data pairs and corresponding farmland boundary data in one embodiment of the present application ( Figure 3 Part (a) is a schematic diagram of medium-resolution remote sensing images. Figure 3 Part (b) is a schematic diagram of high-resolution remote sensing images. Figure 3 Part (c) is a schematic diagram of farmland boundary data); Figure 4 This is a schematic diagram of the super-resolution-field joint model in one embodiment of the present application; Figure 5 Schematic diagram of current medium-resolution remote sensing image, super-resolution data and field data in one embodiment of the present application ( Figure 5 Part (a) is a schematic diagram of the current medium-resolution remote sensing image. Figure 5 Part (b) is a schematic diagram of the current super-resolution data. Figure 5 Part (c) is a schematic diagram of the current field data); Figure 6 This is a schematic diagram of the rural farmland rights confirmation results in one embodiment of the present application. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0017] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0018] In an exemplary embodiment, Figure 1 and Figure 2 As shown, a method for dynamically updating cultivated land plots is provided, which includes the following S1 to S5. S1: Obtain historical remote sensing image data pairs of rural cultivated land and corresponding farmland boundary data, and construct a dataset.

[0019] S1 specifically includes: S11: Obtain historical remote sensing image data pairs of rural cultivated land and corresponding farmland boundary data.

[0020] like Figure 3 As shown, the remote sensing image data pairs include medium-resolution and high-resolution remote sensing images. The medium-resolution remote sensing images are Sentinel-2 (S2) data, available for download from the European Space Agency (ESA), and the high-resolution remote sensing images are Gaofen-2 (GF-2) data, available for download from the Land Observation Satellite Data Service Platform. The temporal error of the remote sensing image data pairs is less than two days, and the spatial error is less than 10 meters.

[0021] The farmland boundary data is the open source Field Boundaries for Agriculture (Fiboa).

[0022] S12: Preprocess the historical remote sensing image data to determine the farmland area.

[0023] Historical remote sensing image data pairs were radiometrically corrected and geometrically registered, and reprojected to a geographic coordinate system to ensure consistency with the Geographic Information System (GIS) coordinate system. A land classification dataset (European Space Agency World Cover, ESA World Cover) was downloaded from Google Earth Engine (GEE). Regions of Interest (ROIs) were extracted from farmland areas based on the cultivated land classification within the dataset. Non-farmland areas, including houses and roads, were cropped using the ENVI (Environment for Visualizing Images) software.

[0024] S13: Align the spatial coordinates of the farmland area and the ownership database to obtain an aligned remote sensing image data pair.

[0025] The coordinate system of the farmland area is projected into the coordinate system of the ownership database, and then the ENVI software is used to add control points to spatially align the two so that their coordinates are completely aligned to obtain the aligned remote sensing image data pair.

[0026] S14: Construct a dataset based on the aligned remote sensing image data pairs and the corresponding farmland boundary data.

[0027] S2: Based on the dataset, a super-resolution-field segmentation joint model is constructed based on the super-resolution network and the field segmentation network.

[0028] This application combines the traditional Super-Resolution Convolutional Neural Network (SRCNN) and the Field Parcel Segmentation Network (FPSN) into a new joint super-resolution-field parcel segmentation model. This model adopts a U-net network architecture and consists of an encoder, a decoder, an output head, and a loss function. The encoder consists of multiple convolutional blocks cascaded in a positive pyramid configuration. Through several convolutional downsampling operations, image features are compressed into a high-level semantic feature space. The encoder is used to extract image features from remote sensing image data pairs. The decoder consists of multiple convolutional blocks cascaded in an inverted pyramid configuration. The encoder and decoder blocks are arranged symmetrically and in equal numbers. Through several convolutional upsampling operations, the extracted image features are decoded back into target features. The output head consists of two parallel multi-layer convolutions: one for translating target features into higher-resolution super-resolved data (the super-resolution task), and the other for translating target features into parcel data (the semantic segmentation task). Loss functions include: mean absolute error loss, adversarial loss, edge enhancement loss, cross entropy loss and Loss, the loss function is determined as follows: ; in, is the excess loss, , is the segmentation loss, , is the loss function, is the mean absolute error loss coefficient, is the mean absolute error loss, To combat the loss factor, To combat losses, is the edge enhancement loss coefficient, is the edge enhancement loss, is the cross entropy loss coefficient, is the cross entropy loss, for Loss coefficient, for loss.

[0029] In summary, this application jointly trains the super-resolution network and the field segmentation network, builds a joint framework of "feature extractor-spatial attention-feature decoder-multi-task head", and constructs a super-resolution-field segmentation joint model with a U-net network structure. Figure 4 As shown, Figure 4The number in represents the number of convolutional layer channels. This application integrates the Convolutional Block Attention Module (CBAM) as a whole convolutional block into the decoder and encoder, enabling the super-resolution-field-block joint model to reconstruct field edge information and internal texture information with high precision, thereby improving the recognition accuracy of fields (also known as cultivated land plots). The output head configuration enables the super-resolution-field-block joint model to simultaneously perform super-resolution image recognition and field recognition, further improving the accuracy of field extraction. Experiments have found that compared to traditional single-task learning, the recognition accuracy of this application's model has increased by 5% to 8%.

[0030] In addition, this application combines the classic loss functions in super-resolution tasks and field extraction tasks, and improves the loss function according to the task objectives of this application, using mean absolute error loss, adversarial loss, edge enhancement loss, cross entropy loss and The loss is determined as super-resolution loss and segmentation loss respectively, and the super-resolution loss and segmentation loss are weightedly fused to obtain the loss function (weight ratio 6:4). The improvement of the loss function makes it more suitable for farmland scenes, solves the problem of blurred farmland boundaries caused by low-resolution images, enhances the information of farmland edges, improves the prediction accuracy of the super-resolution-field joint model, and further improves the field extraction accuracy.

[0031] When training the super-resolution-field joint model, the input is historical aligned remote sensing image data pairs and corresponding farmland boundary data. The purpose is to allow the super-resolution-field joint model to learn to extract target features, decode target features, and learn the translation method of the output head. At the same time, the super-resolution-field joint model is trained using public farmland boundary data and existing farmland boundary data, and remote sensing image data pairs with different lighting conditions and different crop growth cycles are used for simulation training to enhance the data of farmland scenes.

[0032] S3: Based on the current medium-resolution remote sensing images of rural cultivated land and the trained super-resolution-field joint model, determine the current super-resolution data and current field data of rural cultivated land.

[0033] After the super-resolution-field joint model provided in this application is trained, it only needs to input the current medium-resolution remote sensing image, i.e., Sentinel-2 data, to obtain the current super-resolution data and current field data of rural cultivated land, such as Figure 5 As shown, the resolution of the super-resolution data is 2m.

[0034] S4: Determine the current field vector data of the rural cultivated land based on the current super-resolution data and the current field data.

[0035] S4 specifically includes: S41: Align and correct the current super-resolution data and the current field data to obtain aligned field data.

[0036] S42: Vectorizing the aligned field data to obtain current field vector data of the rural cultivated land.

[0037] Morphological optimization is performed on the field vector data, such as using closing operations to fill small holes and remove noise areas with too small an area.

[0038] S5: Associate and match the current field vector data with the ownership database to confirm the ownership of rural arable land.

[0039] S5 specifically includes: S51: Overlapping the current field vector data and the longitude and latitude coordinates of the ownership database; and determining the area error between the current field vector data and the ownership database.

[0040] ArcGis software is used to associate and match the current field vector data with the ownership database. The matching rules include spatial position matching and area consistency verification. Spatial position matching is to overlap the latitude and longitude coordinates of the field vector data and the ownership database. Area consistency verification is to use ArcGis software to calculate the area deviation between the contracted area in the ownership database and the field vector data.

[0041] S52: taking the matching rule when the latitude and longitude coordinate overlap is greater than 90% and the area error is less than 5% as the optimal matching rule, and obtaining a matching result.

[0042] Specifically, the overlap threshold is set to 90%, the area threshold is set to 5%, and the matching rule when the latitude and longitude coordinate overlap is greater than 90% and the area error is less than 5% is used as the optimal matching rule. When the optimal matching rule is met, it is considered that the field vector data and the ownership database match. After the matching conditions are met, the contracted area in the ownership database is assigned to the field vector data to obtain the matching result. The assignment function is implemented in ArcGis software.

[0043] Fields that cannot be automatically matched in the field vector data, such as newly added farmland and farmland with ownership disputes, are marked as "farmland awaiting manual verification."

[0044] S53: Use the matching results to confirm the ownership of rural cultivated land.

[0045] The matching results are used to confirm the ownership of rural arable land and obtain the confirmation results, that is, the confirmed field data. The confirmation results include vector data such as the field spatial location, area and right holder information. After the confirmation results are linked to the GIS platform, a standard electronic certificate template can be automatically generated, and PDF and blockchain evidence storage are supported.

[0046] Specifically, obtain the latest medium-resolution remote sensing images every quarter, re-execute the title confirmation process, and update the title confirmation results, such as Figure 6 As shown, comparing historical and current land title confirmation results can automatically identify farmland changes, such as land parcel mergers and divisions, and trigger ownership information updates, enabling automated, incremental updates of farmland parcel data. Furthermore, manual screening and prompting rules are used to handle anomalies. This involves comparing historical and current land title confirmation results to determine if an anomaly exists. If so, an alert is issued, prompting manual intervention. If not, the rural farmland title confirmation results are stored. Anomalies include changes in parcel area exceeding a threshold, conflicts in parcel ownership matching, and newly added unregistered farmland. The threshold for area change is 10%.

[0047] In summary, this application can realize the integrated process design of super-division, segmentation, title confirmation and blockchain evidence storage.

[0048] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0049] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this 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 limiting this application.

Claims

1. A method for dynamically updating cultivated land plots, characterized in that: The method for dynamically updating cultivated land plots includes: Acquire historical remote sensing image data pairs of rural cultivated land and corresponding farmland boundary data, and construct a data set; the remote sensing image data pairs include: medium-resolution remote sensing images and high-resolution remote sensing images; According to the data set, a super-resolution-field joint model is constructed based on a super-resolution network and a field segmentation network; the super-resolution-field joint model is used to output super-resolution data and field data corresponding to rural cultivated land; the super-resolution-field joint model includes: an encoder, a decoder, an output head and a loss function; the encoder includes a plurality of positive pyramid cascaded convolution blocks, and is used to extract image features of remote sensing image data pairs; the decoder includes a plurality of inverted pyramid cascaded convolution blocks, and is used to decode the image features; the convolution blocks of the encoder and the decoder are symmetrically arranged and the number is the same; the loss function includes: mean absolute error loss, adversarial loss, edge enhancement loss, cross entropy loss and loss; According to the current medium-resolution remote sensing images of rural cultivated land, the current super-resolution data and current field data of rural cultivated land are determined based on the trained super-resolution-field joint model; Determine the current field vector data of rural cultivated land based on the current super-resolution data and the current field data; The current field vector data is associated and matched with the ownership database to confirm the ownership of rural arable land.

2. The method for dynamically updating cultivated land plots according to claim 1, characterized in that: The acquisition of historical remote sensing image data pairs of rural cultivated land and corresponding farmland boundary data, and the construction of a data set, specifically includes: Preprocessing historical remote sensing image data to determine farmland areas; the preprocessing process includes: radiometric correction, geometric registration, and farmland area extraction; the farmland area extraction process includes: extracting regions of interest in the farmland area using cultivated land classification in a land classification dataset, and clipping non-farmland areas; Aligning the spatial coordinates of the farmland area and the ownership database to obtain an aligned remote sensing image data pair; A dataset is constructed based on the aligned remote sensing image data pairs and the corresponding farmland boundary data.

3. The method for dynamically updating cultivated land plots according to claim 1, characterized in that: The process of determining the loss function includes: Using the formula Determine the over-resolution loss ; Using the formula Determine segmentation loss ; Using the formula Determine the loss function ; in, is the mean absolute error loss coefficient, is the mean absolute error loss, To combat the loss factor, To combat losses, is the edge enhancement loss coefficient, is the edge enhancement loss, is the cross entropy loss coefficient, is the cross entropy loss, for Loss coefficient, for loss.

4. The method for dynamically updating cultivated land plots according to claim 1, characterized in that: Determining the current field vector data of rural cultivated land based on the current super-resolution data and the current field data specifically includes: Align and correct the current super-resolution data and the current field data to obtain aligned field data; The aligned field data are vectorized to obtain current field vector data of the rural cultivated land.

5. The method for dynamically updating cultivated land plots according to claim 1, characterized in that: The current field vector data is associated and matched with the ownership database to confirm the ownership of rural cultivated land, specifically including: Overlapping the current field vector data with the longitude and latitude coordinates of the ownership database; and determining the area error between the current field vector data and the ownership database; The matching rule when the latitude and longitude coordinate overlap is greater than 90% and the area error is less than 5% is used as the optimal matching rule, and a matching result is obtained; The matching results are used to confirm the ownership of rural arable land.

6. The method for dynamically updating cultivated land plots according to claim 1, characterized in that: The current field vector data is associated and matched with the ownership database to confirm the ownership of the rural cultivated land, and then the following steps are further included: Based on the results of rural cultivated land title confirmation, determine whether there are any abnormal situations; if so, an alarm is issued to prompt manual intervention; if not, the results of rural cultivated land title confirmation are stored; the abnormal situations include: the degree of change in field area exceeds the area change threshold, field ownership matching conflicts and newly added unregistered cultivated land.

7. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method for dynamically updating cultivated land plots according to any one of claims 1 to 6 is implemented.

Images