Land parcel identification method and device, equipment, storage medium and computer program product
By combining multispectral and sub-meter satellite remote sensing data processing methods, and using crop vegetation index and image segmentation models, the problem of insufficient accuracy in planting plot identification in satellite remote sensing technology was solved, and high-precision plot identification was achieved.
Patent Information
- Application Number
- CN202510646790.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-19
- Publication Date
- 2025-09-19
AI Technical Summary
Existing satellite remote sensing technology has the problem of poor recognition accuracy in planting plot identification, especially due to the lack of recognition accuracy caused by the same spectrum of different objects and low spatial resolution.
Combining multispectral satellite and sub-meter satellite remote sensing data, through the joint processing of geometric registration, classification model and image segmentation model, using the dual threshold classification model of crop vegetation index YVI and GNDVI, combined with random forest classification and SAM image segmentation model, the refined identification of planting plots can be achieved.
It improves the accuracy of planting plot identification, overcomes the problem of insufficient sensitivity of traditional spectral indicators to crops, and achieves a leap in identification accuracy from pixel level to sub-meter level.
Smart Images
Figure CN120673268A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of image recognition technology, and in particular to a land parcel recognition method, apparatus, device, storage medium, and computer program product. Background Art
[0002] Crop insurance primarily covers staple crops like wheat, corn, and rice, as well as oilseed crops like peanuts, soybeans, and rapeseed. Insurance and claims settlements are based on each farmer's cultivated land, which can be as small as a few mu (approximately 1 acre). Crop insurance underwriting is conducted on-site, with the insured area typically provided by the farmer or measured on-site.
[0003] Using satellite remote sensing technology to obtain planting information of plots is currently a convenient and efficient means. Specifically, satellite remote sensing technology is used to collect the spectral characteristics of plots, and based on the spectral characteristics, planted plots are distinguished from non-planted plots. For example, taking the identification of rapeseed planting plots through satellite remote sensing technology as an example, since rapeseed has spectral characteristics that are different from other crops during its flowering period, this spectral characteristic can be used to identify rapeseed planting plots.
[0004] However, the existing solutions for identifying cultivated plots using satellite remote sensing technology have the following problems: on the one hand, due to the problem of different objects having the same spectrum, it is difficult to accurately distinguish mixed pixels with similar spectra, such as roads and bare land around rapeseed fields, using spectral features alone, resulting in poor recognition accuracy of cultivated plots; on the other hand, since the current identification of cultivated plots often uses medium-resolution multi-light source satellite remote sensing images such as Sentinel-2 and Landsat as data sources, their spatial resolution is low and cannot meet the requirements of crop insurance for plot-level crop interpretation, while sub-meter high-resolution satellite images usually have only four bands (blue, green, red, and near-red), and therefore cannot accurately identify multiple target crops, which in turn leads to low accuracy in plot-level crop interpretation.
[0005] It can be seen that how to use satellite remote sensing technology to accurately and efficiently identify planting plots has become an urgent problem to be solved. Summary of the Invention
[0006] The embodiment of the present application provides a plot identification method to solve the problem of poor identification accuracy when using existing satellite remote sensing technology to identify planting plots.
[0007] The embodiment of the present application also provides a plot identification device to solve the problem of poor recognition accuracy when using existing satellite remote sensing technology to identify planting plots.
[0008] The embodiment of the present application also provides a plot identification device to solve the problem of poor recognition accuracy when using existing satellite remote sensing technology to identify planting plots.
[0009] The embodiment of the present application also provides a computer-readable storage medium to solve the problem of poor recognition accuracy when using existing satellite remote sensing technology to identify planting plots.
[0010] A computer program product is used to solve the problem of poor recognition accuracy when using existing satellite remote sensing technology to identify planting plots.
[0011] The embodiments of this application adopt the following technical solutions: A land parcel identification method includes: obtaining first satellite remote sensing data and second satellite remote sensing data corresponding to a region to be identified, wherein the first satellite remote sensing data is a remote sensing image collected based on a multispectral satellite, and the second satellite remote sensing data is a remote sensing image collected based on a sub-meter satellite; geometrically registering the first satellite remote sensing data based on the second satellite remote sensing data to obtain third satellite remote sensing data; classifying the third satellite remote sensing data based on a pre-trained classification model to obtain an initial crop classification result for the region to be identified; performing a first segmentation processing on the second satellite remote sensing data based on an image segmentation model to obtain land parcel classification information corresponding to the region to be identified, wherein the land parcel classification information represents different types of land parcel information contained in the second satellite remote sensing data and corresponding contour ranges of the land parcels; performing a second segmentation processing on the second satellite remote sensing data based on the initial crop classification result using the image segmentation model to obtain planting plot classification information corresponding to the region to be identified; generating crop planting plot data based on the planting plot classification information and the land parcel classification information to complete identification of the crop planting plots in the region to be identified.
[0012] A land parcel identification device comprises: a remote sensing image acquisition unit, for acquiring first satellite remote sensing data and second satellite remote sensing data corresponding to a to-be-identified area, wherein the first satellite remote sensing data is a remote sensing image acquired based on a multispectral satellite, and the second satellite remote sensing data is a remote sensing image acquired based on a sub-meter-level satellite; a registration unit, for performing geometric registration on the first satellite remote sensing data based on the second satellite remote sensing data as a benchmark, to obtain third satellite remote sensing data; a classification unit, for performing classification processing on the third satellite remote sensing data based on a pre-trained classification model, to obtain an initial crop classification result of the to-be-identified area; an image segmentation unit, for performing classification processing on the third satellite remote sensing data based on an image segmentation model A model is used to perform a first segmentation processing on the second satellite remote sensing data to obtain the plot classification information corresponding to the area to be identified, wherein the plot classification information represents the different types of plot information contained in the second satellite remote sensing data, and the contour range corresponding to the plot; an image segmentation unit is used to perform a second segmentation processing on the second satellite remote sensing data using the image segmentation model based on the initial crop classification result to obtain the planting plot classification information corresponding to the area to be identified; a plot identification unit is used to generate crop planting plot data according to the planting plot classification information and the plot classification information, and complete the identification of the crop planting plots in the area to be identified.
[0013] A land parcel identification device, comprising: A processor; and a memory arranged to store computer-executable instructions, wherein the executable instructions, when executed, cause the processor to perform the following operations: obtain first satellite remote sensing data and second satellite remote sensing data corresponding to the area to be identified, wherein the first satellite remote sensing data is a remote sensing image collected based on a multispectral satellite, and the second satellite remote sensing data is a remote sensing image collected based on a sub-meter satellite; based on the second satellite remote sensing data as a benchmark, geometrically align the first satellite remote sensing data to obtain third satellite remote sensing data; classify the third satellite remote sensing data according to a pre-trained classification model to obtain an initial crop of the area to be identified Classification result; according to the image segmentation model, the second satellite remote sensing data is segmented for the first time to obtain the land parcel classification information corresponding to the area to be identified, wherein the land parcel classification information represents the different types of land parcel information contained in the second satellite remote sensing data, and the contour range corresponding to the land parcel; based on the initial crop classification result, the second satellite remote sensing data is segmented for the second time using the image segmentation model to obtain the planting plot classification information corresponding to the area to be identified; based on the planting plot classification information and the plot classification information, crop planting plot data is generated to complete the identification of the crop planting plots in the area to be identified.
[0014] A computer-readable storage medium stores one or more programs, which, when executed by an electronic device including multiple application programs, causes the electronic device to perform the following operations: obtain first satellite remote sensing data and second satellite remote sensing data corresponding to a to-be-identified area, wherein the first satellite remote sensing data is a remote sensing image collected based on a multispectral satellite, and the second satellite remote sensing data is a remote sensing image collected based on a sub-meter satellite; based on the second satellite remote sensing data, geometrically align the first satellite remote sensing data to obtain third satellite remote sensing data; and classify the third satellite remote sensing data according to a pre-trained classification model to obtain to the initial crop classification result of the area to be identified; according to the image segmentation model, the second satellite remote sensing data is segmented for the first time to obtain the plot classification information corresponding to the area to be identified, wherein the plot classification information represents the different types of plot information contained in the second satellite remote sensing data, and the contour range corresponding to the plot; based on the initial crop classification result, the second satellite remote sensing data is segmented for the second time using the image segmentation model to obtain the planting plot classification information corresponding to the area to be identified; according to the planting plot classification information and the plot classification information, crop planting plot data is generated to complete the identification of the crop planting plots in the area to be identified.
[0015] A computer program product includes a computer program, which, when executed by a processor, implements the following: obtaining first satellite remote sensing data and second satellite remote sensing data corresponding to a to-be-identified area, wherein the first satellite remote sensing data is a remote sensing image collected based on a multispectral satellite, and the second satellite remote sensing data is a remote sensing image collected based on a sub-meter satellite; performing geometric registration on the first satellite remote sensing data based on the second satellite remote sensing data to obtain third satellite remote sensing data; performing classification processing on the third satellite remote sensing data based on a pre-trained classification model to obtain an initial crop classification result for the to-be-identified area; and performing classification processing on the third satellite remote sensing data based on a pre-trained classification model to obtain an initial crop classification result for the to-be-identified area. The image segmentation model is used to perform a first segmentation processing on the second satellite remote sensing data to obtain the plot classification information corresponding to the area to be identified, wherein the plot classification information represents the different types of plot information contained in the second satellite remote sensing data, and the contour range corresponding to the plot; based on the initial crop classification result, the image segmentation model is used to perform a second segmentation processing on the second satellite remote sensing data to obtain the planting plot classification information corresponding to the area to be identified; based on the planting plot classification information and the plot classification information, crop planting plot data is generated to complete the identification of the crop planting plots in the area to be identified.
[0016] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects: By adopting the plot identification method provided in the embodiment of the present application, when it is necessary to identify a designated crop planting plot through satellite remote sensing images, the plot identification system can respectively obtain first satellite remote sensing data collected by a multispectral satellite and second satellite remote sensing data collected by a sub-meter satellite. After geometrically registering the first satellite remote sensing data based on the second satellite remote sensing data, the multispectral remote sensing data collected by the multispectral satellite is first classified using the spectral information contained in the first satellite remote sensing data and a pre-trained classification model to obtain an initial crop classification result for the area to be identified, thereby realizing the interpretation of the crop planting plots in the area to be identified. When the image segmentation model is used, the second satellite remote sensing data is subjected to full-map unsupervised segmentation processing to obtain the land parcel classification information corresponding to the area to be identified. The accuracy of remote sensing images collected by sub-meter satellites can be used to accurately determine the types of each plot in the area to be identified (for example, roads, farmlands, playgrounds, etc.); secondly, according to the initial crop classification results, the second satellite remote sensing data collected by the sub-meter satellite is subjected to a second segmentation processing to obtain the planting plot classification information corresponding to the area to be identified; finally, the planting plot classification information and the plot classification information can be used to determine the crop planting plot data in the area to be identified, thereby completing the identification of the crop planting plots in the area to be identified. The land parcel identification method provided in the embodiment of the present application, on the one hand, proposes a dual-threshold classification model training method based on the crop vegetation index YVI and the normalized green vegetation index GNDVI according to the spectral characteristics of crops. The classification model trained by this method focuses on strengthening the high reflectivity characteristics of crops in the green light band, overcoming the problem of insufficient sensitivity of traditional spectral indicators (such as EVI and SAVI) to crops, and improving the recognition accuracy; on the other hand, when performing land parcel identification, multispectral remote sensing images collected by multispectral satellites are combined with high spatial resolution remote sensing images collected by sub-meter satellites. Multispectral remote sensing images are used for spectral classification, while spatial resolution remote sensing images are used to refine land parcel boundaries, solving the problem that medium and low resolution images cannot extract small land parcels, thereby greatly improving land parcel identification accuracy; finally, when performing land parcel identification, the embodiment of the present application is based on the joint iteration of the spectral index random forest classification model and the image segmentation model SAM based on sub-meter remote sensing images. The random forest is first used to roughly extract the crop area, and then the SAM model is combined with point prompt information for secondary conditional segmentation processing, achieving a leap in vector land parcel accuracy from pixel level to sub-meter level. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1A schematic diagram of a specific process of a land parcel identification method provided in an embodiment of the present application; Figure 2 A schematic diagram of the specific structure of a land parcel identification device provided in an embodiment of the present application; Figure 3 A schematic diagram of the specific structure of a land parcel identification device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0018] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. 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.
[0019] In order to solve the problem of poor recognition accuracy when using existing satellite remote sensing technology to identify planting plots, an embodiment of the present application provides a plot recognition system and a data processing method based on the data processing system.
[0020] The execution subject of the plot identification method provided in the embodiment of the present application may be, but is not limited to, at least one of a plot identification server, an underwriting server, a claims server, or a remote sensing image processing server; or, the execution subject of the method may also be a plot identification system running on a server; in addition, the execution subject of the method may also be a terminal device used by a user or a backend manager, or even a plot identification application installed on a terminal device. For ease of description, the embodiments of the present invention are all described by taking the execution subject as a plot identification system running on an underwriting server as an example. It can be understood that the execution subject of the method being a plot identification system on an underwriting server is only an exemplary description and should not be understood as a limitation on the method.
[0021] The specific implementation flow chart of the land parcel identification method provided in this application is as follows Figure 1 As shown, it mainly includes the following steps: Step 11, obtaining first satellite remote sensing data and second satellite remote sensing data corresponding to the area to be identified; Among them, the first satellite remote sensing data is multispectral remote sensing images collected by multispectral satellites (for example, Sentinel-29 Earth observation satellite or Landsat-8 Earth observation satellite), and the second satellite remote sensing data is high spatial resolution remote sensing images collected by sub-meter satellites (for example, WorldView series satellites, PlanetScope satellites and SuperDove satellites, etc.).
[0022] It should be noted here that the embodiment of the present application does not limit the specific satellite platform from which the first satellite remote sensing data and the second satellite remote sensing data come. It only needs to ensure that the first satellite remote sensing data is a multispectral remote sensing image, and the second satellite remote sensing data is a high spatial resolution remote sensing image collected by a sub-meter (0.5m) satellite.
[0023] Step 12, performing geometric registration on the first satellite remote sensing data based on the second satellite remote sensing data to obtain third satellite remote sensing data; In one embodiment, the land parcel recognition system may perform geometric registration according to the following sub-steps, including: Sub-step 1201, performing orthorectification on the first satellite remote sensing data; Through orthorectification, the obvious geometric distortion caused by terrain, camera geometric characteristics and sensor-related errors in the first satellite remote sensing data can be processed, and the output orthorectified image will be an orthophotod true planar image.
[0024] It should be noted that the embodiments of the present application do not specifically limit the specific algorithm used for orthorectification.
[0025] Sub-step 1202, pre-processing the second satellite remote sensing data to obtain geographic location information; In an embodiment of the present application, the geographic location information may be directly provided by a data provider corresponding to the second satellite remote sensing data.
[0026] Sub-step 1203, taking the second satellite remote sensing data as a reference and directly using its projection parameters; For example, if the second satellite remote sensing data adopts the WGS-84 ellipsoid coordinate system (World Geodetic System, WGS), the land parcel recognition system can perform geometric registration on the first satellite remote sensing data based on the projection parameters of the WGS-84 coordinate system.
[0027] In sub-step 1204 , based on the principle of spatial coverage priority, ground features such as field boundaries and road intersections in the crop planting area (with clear geometric outlines and high spectral contrast) are selected as control points (GCPs).
[0028] Sub-step 1205 , using a Scale-Invariant Feature Transform (SIFT) algorithm, performs a coarse match between the second satellite remote sensing data and the first satellite remote sensing data in the visible light band (blue, green, and red) to extract coarse matching points; In sub-step 1206 , a Random Sample Consensus (RANSAC) algorithm is used to perform fine matching, and incorrect matching points are eliminated from the coarse matching points obtained by executing sub-step 1205 , ensuring that the root mean square error (RMSE) of the same-name points is less than 0.5 pixels.
[0029] Sub-step 1207 , manually select points for mountainous areas / cloud-covered areas where automatic matching is insufficient (≥20 control points per image).
[0030] In sub-step 1208, based on the matching points obtained by executing the above sub-steps 1205 to 1207, a second-order polynomial transformation algorithm is used to perform coordinate transformation and resample the first satellite remote sensing data to obtain the first satellite remote sensing data after geometric registration. For the sake of convenience, the first satellite remote sensing data after geometric registration will be referred to as the third satellite remote sensing data in the following text.
[0031] Step 13: classify the third satellite remote sensing data acquired by executing step 12 according to the pre-trained classification model to obtain an initial crop classification result for the area to be identified; It should be noted here that since different vegetation has different reflectivity in different color bands (for example, rapeseed has a higher reflectivity in the green band and a lower reflectivity in the red and blue bands), when training a classification model for a specified crop planting area, by emphasizing the high reflectivity characteristics of the crop in a specific color band, the trained classification model's ability to distinguish this type of crop can be improved. Based on this, the embodiment of the present application proposes a crop vegetation index YVI and a normalized green vegetation index GNDVI, and uses the crop vegetation index YVI and the normalized green vegetation index GNDVI to mark training samples, thereby improving the trained classification model's ability to distinguish specific crops and the accuracy of distinction.
[0032] For example, taking the training of a classification model for rapeseed plots as an example, in the embodiment of the present application, the plot recognition system can train the classification model according to the following sub-steps, including: Sub-step 1301, sampling the third satellite remote sensing data to obtain a plurality of sample data; In one embodiment, the land parcel identification system may extract sample data from the third satellite remote sensing data at a certain pixel interval, for example, extracting one sample data at every 5-pixel interval.
[0033] Sub-step 1302 , reading each sample data separately, and obtaining the reflectance corresponding to the red light band, green light band, blue light band and near infrared band respectively corresponding to each sample data; Sub-step 1303, determining the crop vegetation index YVI corresponding to each sample data according to the red light band reflectance, the green light band reflectance, and the blue light band reflectance; Since rapeseed has a higher reflectivity in the green light band and a lower reflectivity in the red and blue light bands, the green light band can be used to determine the rapeseed vegetation index YVI and the rapeseed normalized green light vegetation index GNDVI when training a classification model for rapeseed fields.
[0034] In an embodiment of the present application, the land parcel identification system can sum the red light band reflectivity and the blue light band reflectivity corresponding to the sample data to obtain the red and blue reflectivity corresponding to the sample data; and determine the crop vegetation index YVI corresponding to the sample data based on the difference between three times the green light band reflectivity and the red and blue reflectivity.
[0035] Specifically, the land parcel recognition system can standardize the green light band reflectance according to the following formula [1]: [1] in, , as well as They represent the red light band reflectivity, green light band reflectivity and blue light band reflectivity respectively. Normalized green light reflectance.
[0036] The land parcel recognition system can standardize the red light band reflectance and the blue light band reflectance according to the following formula [2]: [2] Then the plot identification system can determine the crop vegetation index YVI according to the following formula [3]:
[0037] Sub-step 1304 , determining the normalized green vegetation index (GNDVI) corresponding to each sample data according to the green band reflectance and the near-infrared band reflectance; Specifically, in the embodiment of the present application, the land parcel recognition system can sum the near-infrared band reflectivity and the green band reflectivity corresponding to the sample data according to the following formula [4] to obtain the first reflectivity corresponding to the sample data: [4] The land parcel recognition system can determine the second reflectivity corresponding to the sample data according to the difference between the near-infrared band reflectivity and the green band reflectivity corresponding to the sample data according to the following formula [5]: [5] Furthermore, the land parcel recognition system can determine the normalized green vegetation index GNDVI corresponding to the sample data according to the ratio of the second reflectivity to the first reflectivity according to the following formula [6]: [6] Sub-step 1305: Classify and label the sample data obtained by executing sub-step 1301 based on the crop vegetation index YVI and the normalized green vegetation index GNDVI obtained by executing sub-steps 1303 and 1304 above, to obtain planted plot sample data and non-planted plot sample data; Specifically, the plot identification system can calculate the crop vegetation index YVI and the normalized green vegetation index GNDVI corresponding to each sample data according to the above formulas [2] and [6], and determine the first quartile (Q1, the value in the top 25% after sorting the data set from small to large) and the second quartile (Q2, also known as the median, the value in the top 50% after sorting the data set from small to large) of YVI and GNDVI respectively based on the crop vegetation index YVI and the normalized green vegetation index GNDVI corresponding to each sample data.
[0038] According to business needs, the plot identification system can flexibly set the thresholds corresponding to YVI and GNDVI based on Q1 or Q2. Furthermore, the plot identification system can mark sample data whose YVI and GNDVI are both greater than their respective thresholds as planted plot sample data (i.e., rapeseed planted plots), and mark the remaining sample data as non-planted plot sample data (i.e., non-rapeseed planted plots).
[0039] Sub-step 1306 , training a random forest classifier based on the plantation plot sample data and the non-plantation plot sample data marked by executing sub-step 1305 , to obtain a classification model.
[0040] Specifically, the plot identification system can divide the planted plot sample data and non-planted plot sample data marked by executing sub-step 1305 into a training set and a test set according to a certain ratio (for example, 70% for training and 30% for testing), use the hyperparameter grid search (GridSearchCV) module in Python to find the best hyperparameter combination, and use the random forest classifier (RandomForestClassifier) module of the python random forest library to retrain the model with the best hyperparameters, and predict the test set to obtain the model evaluation result, and use the highest scoring model as the trained classification model.
[0041] Then, the trained classification model is used to classify the third satellite remote sensing data, and the multispectral remote sensing impact corresponding to the third satellite remote sensing data is divided into rapeseed pixels and non-rape pixels. The initial crop classification results of the area to be identified are obtained, and the classification results are exported as raster data. At the same time, the raster data is binarized. In the binarization result, 1 represents the target area (i.e., the rapeseed planting area) and 0 represents the background area (the non-rape planting area). The binarization result is saved for subsequent processing.
[0042] Step 14: performing a first segmentation process on the second satellite remote sensing data according to the image segmentation model to obtain land parcel classification information corresponding to the area to be identified; The land parcel classification information represents different types of land parcel information contained in the second satellite remote sensing data, as well as the contour range corresponding to the land parcel.
[0043] Specifically, the land parcel recognition system can use the SAM segmentation algorithm (Segment Anything Model) to perform unsupervised segmentation on the second satellite remote sensing data to obtain land parcel classification information corresponding to the area to be identified (including roads, factories, schools, farmland and other areas corresponding to the area to be identified), and convert the land parcel classification information into a vector to obtain a vector diagram of the first rapeseed planting area.
[0044] Step 15, based on the initial crop classification result obtained by executing step 13, performing a second segmentation process on the second satellite remote sensing data using the image segmentation model to obtain the planting plot classification information corresponding to the area to be identified; In one embodiment, the plot identification system can determine point prompt information based on the initial crop classification result obtained by executing step 13, and then use the image segmentation model to perform conditional segmentation processing on the second satellite remote sensing data based on the point prompt information to obtain the planting plot classification information corresponding to the area to be identified.
[0045] Specifically, in an embodiment of the present application, the plot identification system can convert the binarization processing result obtained by executing step 13 into a vector format to obtain a crop classification vector result, sample the data located in the crop planting area in the crop classification vector result to obtain a second sample, determine the point prompt information of the second sample, and then convert the second satellite remote sensing data into a 3-band (RGB) 8-bit image. The converted second satellite remote sensing data is input into the SAM segmentation model, and conditional segmentation is performed based on the point prompt information (point prompt) determined by the second sample to obtain the planting plot classification information corresponding to the area to be identified, and the planting plot classification information is converted into a vector to obtain a second rapeseed planting area vector diagram.
[0046] Step 16: Generate crop planting plot data based on the planting plot classification information and the plot classification information, and complete the identification of the crop planting plots in the area to be identified.
[0047] Specifically, the plot identification system can perform spatial analysis based on the second rapeseed planting area vector diagram obtained by executing step 15 and the first rapeseed planting area vector diagram obtained by executing step 14, using the first rapeseed planting area vector diagram as the data source, and selecting the plot vector in the first rapeseed planting area vector diagram that intersects with the second rapeseed planting area vector diagram, which is the identified rapeseed planting plot.
[0048] By adopting the plot identification method provided in the embodiment of the present application, when it is necessary to identify a designated crop planting plot through satellite remote sensing images, the plot identification system can respectively obtain first satellite remote sensing data collected by a multispectral satellite and second satellite remote sensing data collected by a sub-meter satellite. After geometrically registering the first satellite remote sensing data based on the second satellite remote sensing data, the multispectral remote sensing data collected by the multispectral satellite is first classified using the spectral information contained in the first satellite remote sensing data and a pre-trained classification model to obtain an initial crop classification result for the area to be identified, thereby realizing the interpretation of the crop planting plots in the area to be identified. When the image segmentation model is used, the second satellite remote sensing data is subjected to full-map unsupervised segmentation processing to obtain the land parcel classification information corresponding to the area to be identified. The accuracy of remote sensing images collected by sub-meter satellites can be used to accurately determine the types of each plot in the area to be identified (for example, roads, farmlands, playgrounds, etc.); secondly, according to the initial crop classification results, the second satellite remote sensing data collected by the sub-meter satellite is subjected to a second segmentation processing to obtain the planting plot classification information corresponding to the area to be identified; finally, the planting plot classification information and the plot classification information can be used to determine the crop planting plot data in the area to be identified, thereby completing the identification of the crop planting plots in the area to be identified. The land parcel identification method provided in the embodiment of the present application, on the one hand, proposes a dual-threshold classification model training method based on the crop vegetation index YVI and the normalized green vegetation index GNDVI according to the spectral characteristics of crops. The classification model trained by this method focuses on strengthening the high reflectivity characteristics of crops in the green light band, overcoming the problem of insufficient sensitivity of traditional spectral indicators (such as EVI and SAVI) to crops, and improving the recognition accuracy; on the other hand, when performing land parcel identification, multispectral remote sensing images collected by multispectral satellites are combined with high spatial resolution remote sensing images collected by sub-meter satellites. Multispectral remote sensing images are used for spectral classification, while spatial resolution remote sensing images are used to refine land parcel boundaries, solving the problem that medium and low resolution images cannot extract small land parcels, thereby greatly improving land parcel identification accuracy; finally, when performing land parcel identification, the embodiment of the present application is based on the joint iteration of the spectral index random forest classification model and the image segmentation model SAM based on sub-meter remote sensing images. The random forest is first used to roughly extract the crop area, and then the SAM model is combined with point prompt information for secondary conditional segmentation processing, achieving a leap in vector land parcel accuracy from pixel level to sub-meter level.
[0049] In one embodiment, the present application also provides a plot identification device to solve the problem of poor recognition accuracy when using existing satellite remote sensing technology to identify planting plots. The specific structural diagram of the plot identification device is as follows Figure 2 As shown, it includes: a remote sensing image acquisition unit 21, a registration unit 22, a classification unit 23, an image segmentation unit 24 and a land parcel recognition unit 25.
[0050] The remote sensing image acquisition unit 21 is configured to acquire first satellite remote sensing data and second satellite remote sensing data corresponding to the area to be identified, wherein the first satellite remote sensing data is a remote sensing image acquired based on a multispectral satellite, and the second satellite remote sensing data is a remote sensing image acquired based on a sub-meter satellite; a registration unit 22 for performing geometric registration on the first satellite remote sensing data based on the second satellite remote sensing data to obtain third satellite remote sensing data; a classification unit 23 configured to classify the third satellite remote sensing data according to a pre-trained classification model to obtain an initial crop classification result for the area to be identified; an image segmentation unit 24 configured to perform a first segmentation process on the second satellite remote sensing data according to an image segmentation model to obtain land parcel classification information corresponding to the area to be identified, wherein the land parcel classification information represents information of different types of land parcels contained in the second satellite remote sensing data and contour ranges corresponding to the land parcels; An image segmentation unit 24 is configured to perform a second segmentation process on the second satellite remote sensing data using the image segmentation model based on the initial crop classification result to obtain classification information of the planting plot corresponding to the area to be identified; The plot identification unit 25 is used to generate crop planting plot data according to the planting plot classification information and the plot classification information, and complete the identification of the crop planting plots in the area to be identified.
[0051] In one embodiment, it also includes a model training unit, which is specifically used to: sample the third satellite remote sensing data to obtain multiple sample data; read each of the sample data separately to obtain the reflectance corresponding to the red light band, green light band, blue light band and near-infrared band of each of the sample data; determine the crop vegetation index YVI corresponding to each of the sample data according to the red light band reflectance, the green light band reflectance and the blue light band reflectance; determine the normalized green light vegetation index GNDVI corresponding to each of the sample data according to the green light band reflectance and the near-infrared band reflectance; classify and label the sample data according to the crop vegetation index YVI and the normalized green light vegetation index GNDVI to obtain planted plot sample data and non-planted plot sample data; train a random forest classifier based on the planted plot sample data and the non-planted plot sample data to obtain the classification model.
[0052] In one embodiment, the model training unit is specifically used to: sum the red light band reflectivity and the blue light band reflectivity corresponding to the sample data to obtain the red and blue reflectivity corresponding to the sample data; and determine the crop vegetation index YVI corresponding to the sample data based on the difference between three times the green light band reflectivity and the red and blue reflectivity.
[0053] In one embodiment, the model training unit is specifically used to: sum the near-infrared band reflectivity and the green band reflectivity corresponding to the sample data to obtain a first reflectivity corresponding to the sample data; determine a second reflectivity corresponding to the sample data based on the difference between the near-infrared band reflectivity and the green band reflectivity corresponding to the sample data; and determine a normalized green vegetation index GNDVI corresponding to the sample data based on the ratio of the second reflectivity to the first reflectivity.
[0054] In one embodiment, the image segmentation unit 24 is specifically used to: determine point prompt information based on the initial crop classification result; and perform conditional segmentation processing on the second satellite remote sensing data using the image segmentation model based on the point prompt information to obtain the planting plot classification information corresponding to the area to be identified.
[0055] In one embodiment, the image segmentation unit 24 is specifically used to: binarize the initial crop classification result to obtain a binarized result, wherein 1 in the binarized result represents a crop planting area and 0 represents a background area; convert the binarized result into a vector format to obtain a crop classification vector result; sample the data located in the crop planting area in the crop classification vector result to obtain a second sample; and determine point prompt information based on the second sample.
[0056] By using the plot identification device provided in the embodiment of the present application, when it is necessary to identify a designated crop planting plot through satellite remote sensing images, the plot identification system can respectively obtain first satellite remote sensing data collected by a multispectral satellite and second satellite remote sensing data collected by a sub-meter satellite. After geometrically registering the first satellite remote sensing data based on the second satellite remote sensing data, the multispectral remote sensing data collected by the multispectral satellite is first classified using the spectral information contained in the first satellite remote sensing data and a pre-trained classification model to obtain an initial crop classification result for the area to be identified, thereby realizing the interpretation of the crop planting plots in the area to be identified. At the same time, When the image segmentation model is used, the second satellite remote sensing data is subjected to full-map unsupervised segmentation processing to obtain the land parcel classification information corresponding to the area to be identified. The accuracy of remote sensing images collected by sub-meter satellites can be used to accurately determine the types of each plot in the area to be identified (for example, roads, farmlands, playgrounds, etc.); secondly, according to the initial crop classification results, the second satellite remote sensing data collected by the sub-meter satellite is subjected to a second segmentation processing to obtain the planting plot classification information corresponding to the area to be identified; finally, the planting plot classification information and the plot classification information can be used to determine the crop planting plot data in the area to be identified, thereby completing the identification of the crop planting plots in the area to be identified. The land parcel identification method provided in the embodiment of the present application, on the one hand, proposes a dual-threshold classification model training method based on the crop vegetation index YVI and the normalized green vegetation index GNDVI according to the spectral characteristics of crops. The classification model trained by this method focuses on strengthening the high reflectivity characteristics of crops in the green light band, overcoming the problem of insufficient sensitivity of traditional spectral indicators (such as EVI and SAVI) to crops, and improving the recognition accuracy; on the other hand, when performing land parcel identification, multispectral remote sensing images collected by multispectral satellites are combined with high spatial resolution remote sensing images collected by sub-meter satellites. Multispectral remote sensing images are used for spectral classification, while spatial resolution remote sensing images are used to refine land parcel boundaries, solving the problem that medium and low resolution images cannot extract small land parcels, thereby greatly improving land parcel identification accuracy; finally, when performing land parcel identification, the embodiment of the present application is based on the joint iteration of the spectral index random forest classification model and the image segmentation model SAM based on sub-meter remote sensing images. The random forest is first used to roughly extract the crop area, and then the SAM model is combined with point prompt information for secondary conditional segmentation processing, achieving a leap in vector land parcel accuracy from pixel level to sub-meter level.
[0057] Figure 3 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present application. Figure 3At the hardware level, the electronic device includes a processor and, optionally, an internal bus, a network interface, and memory. The memory may include internal memory, such as high-speed random-access memory (RAM), or non-volatile memory, such as at least one disk drive. Of course, the electronic device may also include other hardware required for its services.
[0058] The processor, network interface, and memory can be interconnected via an internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus. The bus can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, Figure 3 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or one type of bus.
[0059] The memory is used to store programs. Specifically, the program may include program code, which includes computer operating instructions. The memory may include internal memory and non-volatile memory, and provides instructions and data to the processor.
[0060] The processor reads the corresponding computer program from the non-volatile memory into the internal memory and then runs it, forming a land parcel identification device at the logical level. The processor executes the program stored in the memory and is specifically used to perform the following operations: Acquire first satellite remote sensing data and second satellite remote sensing data corresponding to the area to be identified, wherein the first satellite remote sensing data is a remote sensing image collected based on a multispectral satellite, and the second satellite remote sensing data is a remote sensing image collected based on a sub-meter satellite; perform geometric registration on the first satellite remote sensing data based on the second satellite remote sensing data to obtain third satellite remote sensing data; classify the third satellite remote sensing data based on a pre-trained classification model to obtain an initial crop classification result for the area to be identified; perform a first segmentation processing on the second satellite remote sensing data based on an image segmentation model to obtain land parcel classification information corresponding to the area to be identified, wherein the land parcel classification information represents different types of land parcel information contained in the second satellite remote sensing data and the corresponding contour range of the land parcel; based on the initial crop classification result, perform a second segmentation processing on the second satellite remote sensing data using the image segmentation model to obtain planting plot classification information corresponding to the area to be identified; generate crop planting plot data based on the planting plot classification information and the land parcel classification information to complete the identification of the crop planting plots in the area to be identified.
[0061] The above application Figure 3The methods performed by electronic devices for land parcel identification disclosed in the illustrated embodiments can be applied to or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. During implementation, the steps of the aforementioned methods can be performed by hardware integrated logic circuits within the processor or by software instructions. These processors can be general-purpose processors, including central processing units (CPUs) and network processors (NPs); they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The methods, steps, and logic block diagrams disclosed in the embodiments of this application can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly executed by a hardware decoding processor or by a combination of hardware and software modules within the decoding processor. The software module can be located in a storage medium well-known in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in the memory, and the processor reads the information in the memory and, in conjunction with its hardware, completes the steps of the above method.
[0062] Of course, in addition to software implementation, the electronic device of this application does not exclude other implementation methods, such as logic devices or a combination of software and hardware, etc. That is to say, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.
[0063] The embodiment of the present application also provides a computer-readable storage medium, which stores one or more programs, wherein the one or more programs include instructions, which, when executed by a portable electronic device including multiple application programs, can enable the portable electronic device to execute Figure 1 The method of the embodiment shown is specifically used to perform the following operations: Acquire first satellite remote sensing data and second satellite remote sensing data corresponding to the area to be identified, wherein the first satellite remote sensing data is a remote sensing image collected based on a multispectral satellite, and the second satellite remote sensing data is a remote sensing image collected based on a sub-meter satellite; perform geometric registration on the first satellite remote sensing data based on the second satellite remote sensing data to obtain third satellite remote sensing data; classify the third satellite remote sensing data based on a pre-trained classification model to obtain an initial crop classification result for the area to be identified; perform a first segmentation processing on the second satellite remote sensing data based on an image segmentation model to obtain land parcel classification information corresponding to the area to be identified, wherein the land parcel classification information represents different types of land parcel information contained in the second satellite remote sensing data and the corresponding contour range of the land parcel; based on the initial crop classification result, perform a second segmentation processing on the second satellite remote sensing data using the image segmentation model to obtain planting plot classification information corresponding to the area to be identified; generate crop planting plot data based on the planting plot classification information and the land parcel classification information to complete the identification of the crop planting plots in the area to be identified.
[0064] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0065] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0066] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.
[0067] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.
[0068] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.
[0069] Memory may include non-permanent storage in a computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.
[0070] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can be implemented using any method or technology to store information. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change RAM (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media such as modulated data signals and carrier waves.
[0071] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.
[0072] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0073] The foregoing is merely an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.
Claims
1. A land parcel identification method, characterized in that: include: Acquire first satellite remote sensing data and second satellite remote sensing data corresponding to the area to be identified, wherein the first satellite remote sensing data is a remote sensing image collected based on a multispectral satellite, and the second satellite remote sensing data is a remote sensing image collected based on a sub-meter satellite; Based on the second satellite remote sensing data, the first satellite remote sensing data is geometrically aligned to obtain third satellite remote sensing data; classifying the third satellite remote sensing data according to a pre-trained classification model to obtain an initial crop classification result for the area to be identified; performing a first segmentation process on the second satellite remote sensing data according to an image segmentation model to obtain land parcel classification information corresponding to the to-be-identified area, wherein the land parcel classification information represents information of different types of land parcels contained in the second satellite remote sensing data and contour ranges corresponding to the land parcels; Based on the initial crop classification result, the second satellite remote sensing data is subjected to a second segmentation process using the image segmentation model to obtain the planting plot classification information corresponding to the area to be identified; According to the planting plot classification information and the plot classification information, crop planting plot data is generated to complete the identification of the crop planting plots in the area to be identified.
2. The method according to claim 1, characterized in that Pre-training the classification model specifically includes: Sampling the third satellite remote sensing data to obtain a plurality of sample data; Reading each of the sample data respectively to obtain the reflectance corresponding to the red light band, the green light band, the blue light band and the near-infrared band respectively corresponding to each of the sample data; Determining the crop vegetation index YVI corresponding to each of the sample data according to the red light band reflectivity, the green light band reflectivity, and the blue light band reflectivity; Determine the normalized green vegetation index GNDVI corresponding to each of the sample data according to the green light band reflectivity and the near-infrared band reflectivity; Classifying and marking the sample data according to the crop vegetation index YVI and the normalized green vegetation index GNDVI to obtain plantation plot sample data and non-plantation plot sample data; A random forest classifier is trained based on the planted plot sample data and the non-planted plot sample data to obtain the classification model.
3. The method according to claim 2, characterized in that Determining the crop vegetation index YVI corresponding to each sample data according to the red light band reflectivity, the green light band reflectivity, and the blue light band reflectivity specifically includes: Summing the red light band reflectivity and the blue light band reflectivity corresponding to the sample data to obtain the red and blue reflectivity corresponding to the sample data; The crop vegetation index YVI corresponding to the sample data is determined according to the difference between three times the green light band reflectivity and the red and blue reflectivity.
4. The method according to claim 2, characterized in that Determining the normalized green vegetation index (GNDVI) corresponding to each sample data according to the green light band reflectivity and the near-infrared band reflectivity specifically includes: Summing the near-infrared band reflectivity and the green band reflectivity corresponding to the sample data to obtain a first reflectivity corresponding to the sample data; Determining a second reflectivity corresponding to the sample data according to a difference between a near-infrared band reflectivity and a green band reflectivity corresponding to the sample data; A normalized green vegetation index (GNDVI) corresponding to the sample data is determined according to a ratio of the second reflectivity to the first reflectivity.
5. The method according to claim 1, wherein The step of performing a second segmentation process on the second satellite remote sensing data based on the initial crop classification result using the image segmentation model to obtain the planting plot classification information corresponding to the area to be identified specifically includes: determining point prompt information according to the initial crop classification result; According to the point prompt information, the image segmentation model is used to perform conditional segmentation processing on the second satellite remote sensing data to obtain the planting plot classification information corresponding to the area to be identified.
6. The method according to claim 5, characterized in that Determining point prompt information according to the initial crop classification result specifically includes: Binarizing the initial crop classification result to obtain a binarized result, wherein 1 in the binarized result represents a crop planting area and 0 represents a background area; Converting the binarization result into a vector format to obtain a crop classification vector result; Sampling data located in the crop planting area in the crop classification vector result to obtain a second sample; Point prompt information is determined according to the second sample.
7. A land parcel identification device, characterized in that: include: A remote sensing image acquisition unit, configured to acquire first satellite remote sensing data and second satellite remote sensing data corresponding to the area to be identified, wherein the first satellite remote sensing data is a remote sensing image acquired based on a multispectral satellite, and the second satellite remote sensing data is a remote sensing image acquired based on a sub-meter satellite; a registration unit, configured to perform geometric registration on the first satellite remote sensing data based on the second satellite remote sensing data to obtain third satellite remote sensing data; a classification unit, configured to perform classification processing on the third satellite remote sensing data according to a pre-trained classification model to obtain an initial crop classification result of the area to be identified; an image segmentation unit, configured to perform a first segmentation process on the second satellite remote sensing data according to an image segmentation model to obtain land parcel classification information corresponding to the area to be identified, wherein the land parcel classification information represents information of different types of land parcels contained in the second satellite remote sensing data and contour ranges corresponding to the land parcels; an image segmentation unit, configured to perform a second segmentation process on the second satellite remote sensing data using the image segmentation model based on the initial crop classification result, to obtain classification information of the planting plot corresponding to the area to be identified; The plot identification unit is used to generate crop planting plot data based on the planting plot classification information and the plot classification information, and complete the identification of the crop planting plots in the area to be identified.
8. A land parcel identification device comprising: processor; as well as a memory arranged to store computer-executable instructions which, when executed, cause the processor to: Acquire first satellite remote sensing data and second satellite remote sensing data corresponding to the area to be identified, wherein the first satellite remote sensing data is a remote sensing image collected based on a multispectral satellite, and the second satellite remote sensing data is a remote sensing image collected based on a sub-meter satellite; Based on the second satellite remote sensing data, the first satellite remote sensing data is geometrically aligned to obtain third satellite remote sensing data; classifying the third satellite remote sensing data according to a pre-trained classification model to obtain an initial crop classification result for the area to be identified; performing a first segmentation process on the second satellite remote sensing data according to an image segmentation model to obtain land parcel classification information corresponding to the to-be-identified area, wherein the land parcel classification information represents information of different types of land parcels contained in the second satellite remote sensing data and contour ranges corresponding to the land parcels; Based on the initial crop classification result, the second satellite remote sensing data is subjected to a second segmentation process using the image segmentation model to obtain the planting plot classification information corresponding to the area to be identified; According to the planting plot classification information and the plot classification information, crop planting plot data is generated to complete the identification of the crop planting plots in the area to be identified.
9. A computer-readable storage medium storing one or more programs, which, when executed by an electronic device comprising a plurality of application programs, enables the electronic device to execute the land parcel identification method according to any one of claims 1 to 6.
10. A computer program product, characterized in that The invention comprises a computer program, which, when executed by a processor, implements the land parcel identification method according to any one of claims 1 to 6.