Method, device, equipment and medium for monitoring progress of food crop harvesting

Abstract

The application discloses a grain crop harvesting progress monitoring method, device, equipment and medium, and relates to the technical field of agricultural remote sensing monitoring. The method comprises the following steps: acquiring grain crop type distribution map data and multispectral remote sensing image data of a preset number of days before harvesting, calculating a red band depth index and a ground water content index, and calculating reference reflectivity of an unharvested field; during harvesting, acquiring multispectral remote sensing image data of a current time to be monitored, calculating a difference between red light and short-wave infrared band reflectivity of each to-be-monitored pixel of the multispectral remote sensing image data and reference reflectivity of the unharvested field, and obtaining a harvesting index; based on the harvesting index, a parameter intersection point of a double-cluster Gaussian mixture model is fitted to obtain an optimal threshold value; finally, a harvesting state is judged according to the harvesting index and the optimal threshold value, and a monitoring result is obtained. The application realizes large-range and near-real-time grain crop harvesting progress monitoring, and improves monitoring efficiency and accuracy.

Description

Technical Field

[0001] This application relates to the field of agricultural remote sensing monitoring technology, and in particular to a method, device, equipment and medium for monitoring the harvest progress of grain crops. Background Technology

[0002] Harvesting, as the final stage of grain crop cultivation, is a crucial step in extracting yield from the total crop biomass. In the era of agricultural mechanization, large-scale and near-real-time monitoring of harvesting progress is particularly important. Timely monitoring of crop harvesting progress enables dynamic optimization of harvesting equipment scheduling, effectively addresses the impact of extreme weather events such as rainstorms and hail on harvesting operations, and ensures the smooth progress of grain harvesting.

[0003] Currently, monitoring the progress of large-scale crop harvests mainly relies on field surveys and the uploading of agricultural machinery operation information. While this method yields relatively accurate information, it has several drawbacks. On the one hand, field surveys require significant manpower, resources, and time, resulting in low efficiency. On the other hand, uploading agricultural machinery operation information may lead to incomplete data collection, failing to reflect the overall harvest progress in a timely and accurate manner. The development of spaceborne remote sensing technology offers another feasible approach for monitoring the progress of large-scale grain crop harvests. Grain crops, such as rice, corn, and wheat, are typically planted in contiguous areas. At harvest, the aboveground biomass is removed, and the relatively moist photosynthetically active canopy becomes dry stubble or bare soil, causing significant spectral changes that provide crucial signals for remote sensing to capture harvest events. However, there are currently no remote sensing-based algorithms for monitoring the progress of grain crop harvests.

[0004] Therefore, there is an urgent need for a remote sensing-based method for monitoring the harvest progress of grain crops to solve the problems existing in the monitoring methods of related technologies and achieve large-scale, near real-time monitoring of harvest progress. Summary of the Invention

[0005] The purpose of this application is to provide a method, device, equipment, and medium for monitoring the harvest progress of grain crops. This method can solve the problems of existing large-scale crop harvest progress monitoring relying on field surveys and agricultural machinery operation information uploads, which are time-consuming, labor-intensive, and have incomplete data collection. This method can achieve large-scale, near real-time monitoring of grain crop harvest progress, improve monitoring efficiency and accuracy, provide strong support for dynamically optimizing the scheduling of harvesting equipment and responding to the impact of extreme weather, and give full play to the advantages of remote sensing technology in the field of agricultural monitoring.

[0006] To achieve the above objectives, this application provides the following solution:

[0007] Firstly, this application provides a method for monitoring the harvest progress of grain crops, including:

[0008] Acquire distribution map data and multispectral remote sensing image data of grain crop types a preset number of days before harvest;

[0009] Based on the distribution map data of grain crop types and multispectral remote sensing image data of a preset number of days before harvest, the reference reflectance of unharvested fields of each grain crop type is determined by calculating the red band depth index and the surface moisture index; the grain crop types include corn, rice, and wheat.

[0010] During the harvest period, acquire multispectral remote sensing image data to be monitored at the current moment;

[0011] By calculating the difference between the red band reflectance and shortwave infrared band reflectance of each pixel in the multispectral remote sensing image data to be monitored at the current time and the reference reflectance of the unharvested field of the corresponding grain crop type, the harvest index of each grain crop type at the current time is obtained.

[0012] Based on the harvest index of each type of grain crop at the current moment, the intersection of two Gaussian distribution parameters is fitted by a dual-cluster Gaussian mixture model to obtain the optimal threshold for the harvest index of each type of grain crop at the current moment.

[0013] Based on the harvest index of each type of grain crop at the current moment and the corresponding optimal threshold for the harvest index, the harvest status of the grain crop corresponding to each monitored pixel at the current moment is determined, and the monitoring results of the grain crop harvest progress are obtained.

[0014] Secondly, this application provides a monitoring device for the harvesting progress of grain crops, comprising:

[0015] The raw data acquisition module is used to acquire distribution map data of grain crop types and multispectral remote sensing image data for a preset number of days before harvest.

[0016] The reference reflectance determination module is used to determine the reference reflectance of unharvested fields of each type of grain crop based on the distribution map data of grain crop types and multispectral remote sensing image data of a preset number of days before harvest, by calculating the red band depth index and the surface moisture index; the grain crop types include corn, rice, and wheat.

[0017] The remote sensing image data acquisition module is used to acquire multispectral remote sensing image data to be monitored at the current time during the harvest period.

[0018] The harvest index calculation module is used to calculate the harvest index of each type of grain crop at the current time by performing a difference calculation between the red band reflectance and shortwave infrared band reflectance of each pixel in the multispectral remote sensing image data to be monitored at the current time and the reference reflectance of the unharvested field of the corresponding grain crop type.

[0019] The optimal threshold determination module is used to obtain the optimal threshold for the harvest index of each type of grain crop at the current time by fitting the intersection of two Gaussian distribution parameters through a dual-cluster Gaussian mixture model based on the harvest index of each type of grain crop at the current time.

[0020] The harvest progress monitoring module is used to determine the harvest status of the grain crop corresponding to each monitored pixel at the current time based on the harvest index of each type of grain crop at the current time and the corresponding optimal threshold of the harvest index, and to obtain the grain crop harvest progress monitoring results.

[0021] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method for monitoring the harvest progress of grain crops as described above.

[0022] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method for monitoring the harvest progress of grain crops as described above.

[0023] According to the specific embodiments provided in this application, this application has the following technical effects:

[0024] This application provides a method, apparatus, equipment, and medium for monitoring the harvest progress of grain crops. By acquiring data including distribution maps of grain crop types and multispectral remote sensing imagery data from a preset number of days before harvest, it solves the problems of traditional monitoring methods, such as limited data sources, difficulty in acquisition, and inability to cover large areas, thus achieving efficient and convenient acquisition of large-scale monitoring data. By calculating the red band depth index and surface moisture index based on the above data, the reference reflectance of unharvested fields for each type of grain crop is determined. The harvest index is then calculated using the difference between the reflectance of pixels in the current monitoring data and the reference reflectance, solving the problem of unclear and inaccurate extraction of crop harvest feature information in remote sensing data, and achieving effective quantitative characterization of crop harvest status-related features. By using a dual-cluster Gaussian mixture model to fit the parameter intersections based on the harvest index to determine the optimal threshold, and using this to judge the harvest status of the grain crop corresponding to each monitoring pixel, the problem of ambiguous harvest status judgment criteria and insufficient accuracy is solved, achieving accurate and reliable monitoring of grain crop harvest progress. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is an application environment diagram of a method for monitoring the harvest progress of grain crops according to an embodiment of this application.

[0027] Figure 2 This is a flowchart illustrating a method for monitoring the harvest progress of grain crops, provided as an embodiment of this application.

[0028] Figure 3 This is a schematic diagram of the main process of a method for monitoring the harvest progress of grain crops, provided as another embodiment of this application.

[0029] Figure 4 This is a schematic diagram comparing the spectra of harvested and unharvested fields according to an embodiment of this application; wherein, Figure 4 (a) A schematic diagram showing the distribution of harvested and unharvested rice fields in the rice planting area; Figure 4 (b) indicates Figure 4 (a) Comparison of spectral reflectance of different types of fields in rice-growing areas; Figure 4 (c) shows a schematic diagram of the distribution of harvested and unharvested corn fields in the corn planting area; Figure 4 (d) indicates Figure 4 (c) Comparison of spectral reflectance of different types of fields in the maize planting area; Figure 4 (e) shows a schematic diagram of the distribution of harvested and unharvested wheat fields in the wheat planting area; Figure 4 (f) indicates Figure 4 (e) Comparison of spectral reflectance of different types of fields in wheat-growing areas.

[0030] Figure 5 This is a schematic diagram of a method for determining the optimal threshold of the harvest index provided in an embodiment of this application.

[0031] Figure 6 This is a schematic diagram of the study area provided for one embodiment of this application.

[0032] Figure 7 for Figure 6 Spatial distribution map of harvest dates in the study area.

[0033] Figure 8 This is a schematic diagram of the functional modules of a grain crop harvesting progress monitoring device provided in one embodiment of this application.

[0034] Figure 9This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0035] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0036] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0037] The method for monitoring the harvest progress of grain crops provided in this application embodiment can be applied to, for example... Figure 1In the application environment shown, terminal 101 communicates with server 102 via a network. A data storage system can store the data that server 102 needs to process. The data storage system can be set up independently, integrated into server 102, or placed in the cloud or on other servers. Terminal 101 can send the distribution map data of grain crop types and multispectral remote sensing image data for a preset number of days before harvest to server 102. After receiving the distribution map data and multispectral remote sensing image data for the preset number of days before harvest, server 102, based on the distribution map data and multispectral remote sensing image data for the preset number of days before harvest, determines the reference reflectance of unharvested fields for each grain crop type by calculating the red band depth index and surface moisture index; the grain crop types include corn, rice, and wheat; during the harvest period, it acquires the multispectral remote sensing image data to be monitored at the current time; and through the multispectral remote sensing image data to be monitored at the current time... The red band reflectance and shortwave infrared reflectance of each pixel to be monitored in the spectral remote sensing image data are compared with the reference reflectance of the corresponding unharvested field of grain crop type to obtain the harvest index of each grain crop type at the current time. Based on the harvest index of each grain crop type at the current time, the intersection of the two Gaussian distribution parameters is fitted by a dual-cluster Gaussian mixture model to obtain the optimal threshold of the harvest index of each grain crop type at the current time. Based on the harvest index of each grain crop type at the current time and the corresponding optimal threshold, the harvest status of the grain crop corresponding to each pixel to be monitored at the current time is determined to obtain the grain crop harvest progress monitoring result. The server 102 can feed back the obtained grain crop harvest progress monitoring result to the terminal 101. In addition, in some embodiments, the method for monitoring the harvest progress of grain crops can also be implemented by the server 102 or the terminal 101 separately. For example, the terminal 101 can directly perform harvest progress monitoring processing on the grain crop type distribution map data and multispectral remote sensing image data for a preset number of days before harvest. Alternatively, the server 102 can obtain the grain crop type distribution map data and multispectral remote sensing image data for a preset number of days before harvest from the data storage system, and perform harvest progress monitoring processing on the grain crop type distribution map data and multispectral remote sensing image data for a preset number of days before harvest.

[0038] The terminal 101 can be, but is not limited to, various desktop computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. The server 102 can be implemented using a standalone server or a server cluster consisting of multiple servers, or it can be a cloud server.

[0039] In one exemplary embodiment, such as Figure 2As shown, a method for monitoring the harvest progress of grain crops is provided. This method is executed by computer equipment, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method is applied to... Figure 1 Taking server 102 as an example, the explanation includes the following steps 201 to 206. Wherein:

[0040] Step 201: Obtain the distribution map data of grain crop types and multispectral remote sensing image data for the preset number of days before harvest.

[0041] Step 202: Based on the distribution map data of grain crop types and multispectral remote sensing image data of a preset number of days before harvest, the reference reflectance of unharvested fields of each grain crop type is determined by calculating the red band depth index and the surface moisture index; the grain crop types include corn, rice, and wheat.

[0042] Step 203: During the harvest period, acquire the multispectral remote sensing image data to be monitored at the current moment.

[0043] Step 204: By calculating the difference between the red band reflectance and shortwave infrared band reflectance of each pixel in the multispectral remote sensing image data to be monitored at the current time and the reference reflectance of the unharvested field of the corresponding grain crop type, the harvest index of each grain crop type at the current time is obtained.

[0044] Step 205: Based on the harvest index of each type of grain crop at the current time, fit the intersection of two Gaussian distribution parameters using a bi-cluster Gaussian mixture model to obtain the optimal threshold for the harvest index of each type of grain crop at the current time.

[0045] Step 206: Based on the harvest index of each type of grain crop at the current moment and the corresponding optimal threshold of the harvest index, determine the harvest status of the grain crop corresponding to each monitored pixel at the current moment, and obtain the grain crop harvest progress monitoring results.

[0046] By implementing steps 201 to 206 above, this application can fully utilize the computing power of computer equipment to efficiently process data on the distribution map of grain crop types and multispectral remote sensing image data. By accurately calculating the red band depth index and surface moisture index to determine the reference reflectance, and using difference calculations to obtain the harvest index, the spectral characteristic changes of crop harvest status can be accurately captured. Based on this, a near-equilibrium buffer zone is determined, and an optimal threshold is obtained using a bi-cluster Gaussian mixture model to determine the harvest status. This achieves large-scale, near-real-time monitoring of grain crop harvest progress, effectively solving the problems of time-consuming, labor-intensive, and incomplete data collection associated with traditional monitoring methods. It improves the accuracy and efficiency of monitoring, providing strong support for scientific decision-making in agricultural production.

[0047] In another exemplary embodiment of this application, step 202 specifically includes:

[0048] The red band depth index and surface moisture index are calculated using the following formulas:

[0049] .

[0050] .

[0051] .

[0052] in, Indicates the red band depth index; , , and These represent the reflectance in the red light band, green light band, near-infrared band, and short-wave infrared band, respectively. Indicates the interpolated reflectance in the red light band; , , and represent the center wavelengths of the near-infrared band, the green band, and the red band, respectively; This represents the surface moisture index.

[0053] Based on the distribution map data of grain crop types with a preset number of days before harvest, pixels of each grain crop type whose red band depth index and surface moisture index are both greater than the preset unharvested threshold are selected as reference unharvested pixels.

[0054] The average red band reflectance and shortwave infrared reflectance of all reference unharvested pixels corresponding to each type of grain crop were calculated to obtain the average red band reflectance and average shortwave infrared reflectance of each type of grain crop.

[0055] The average red band reflectance and average shortwave infrared band reflectance corresponding to each type of grain crop are used as the reference reflectance for the red band and shortwave infrared band of the corresponding grain crop type, respectively.

[0056] In another exemplary embodiment of this application, the harvest index is calculated in step 204 using the following formula:

[0057] .

[0058] in, Harvest Index; and These represent the reflectivity in the red light band and the reflectivity in the short-wave infrared band, respectively. and These represent the reference reflectance in the red light band and the reference reflectance in the short-wave infrared band, respectively.

[0059] In another exemplary embodiment of this application, step 205 specifically includes:

[0060] Based on the harvest index of each type of grain crop at the current moment, a class-balanced buffer is determined for the harvest status of each type of grain crop at the current moment; the harvest status includes harvested status and unharvested status.

[0061] Based on the class-balanced buffer for each type of grain crop at the current moment, the intersection of two Gaussian distribution parameters is fitted by a dual-cluster Gaussian mixture model to obtain the optimal threshold for the harvest index of each type of grain crop at the current moment.

[0062] In another exemplary embodiment of this application, a class-balanced buffer for the harvest status of each type of grain crop at the current moment is determined based on the harvest index of each type of grain crop at the current moment, specifically including:

[0063] For each type of grain crop, the following steps are used to obtain the class-balanced buffer of the harvest status of each type of grain crop at the current moment:

[0064] In the multispectral remote sensing image data to be monitored at the current moment, the pixels to be monitored that have a harvest index greater than a preset high threshold for the current type of grain crop are extracted as potential harvested pixels for the current type of grain crop.

[0065] Spatial clustering algorithm is used to cluster the potential harvested pixels of the current grain crop type, dividing the potential harvested pixels into different clusters and obtaining the spatial cluster center of each cluster.

[0066] Using each cluster center as an anchor point, square buffers are gradually expanded outward at preset area intervals until the preset maximum area is reached, resulting in multiple buffers of different areas.

[0067] Within each buffer, the harvest index distribution is fitted using a bi-cluster Gaussian mixture model and a uni-cluster Gaussian model, respectively, and the likelihood ratio between the bi-cluster Gaussian mixture model and the uni-cluster Gaussian model is calculated.

[0068] Sort all buffers by likelihood ratio from largest to smallest, and select the buffers whose likelihood ratio is within a preset percentage threshold as the class balance buffers for the current crop type at the current moment.

[0069] In another exemplary embodiment of this application, the harvest index distribution is fitted using a bi-cluster Gaussian mixture model and a uni-cluster Gaussian model within each buffer, and the likelihood ratio between the bi-cluster Gaussian mixture model and the uni-cluster Gaussian model is calculated, specifically including:

[0070] The likelihood functions of the unicluster Gaussian mixture model and the bicluster Gaussian mixture model are calculated using the following formulas:

[0071] .

[0072] .

[0073] in, and Let represent the likelihood function values ​​of the unicluster Gaussian model and the bicluster Gaussian mixture model, respectively. The first in the class balanced buffer i N represents the total number of cells in the class balance buffer; This represents the probability density function of the normal distribution. Harvest Index; This represents the mean parameter of a single-cluster Gaussian model; This represents the variance parameter of a single-cluster Gaussian model; The weight parameter represents the unharvested pixels; This represents the mean parameter of the unharvested pixels; The variance parameter representing the unharvested pixels; The weight parameters represent the pixels that have been captured; The mean parameter representing the number of pixels already captured; This represents the variance parameter of the pixels that have been harvested.

[0074] For each buffer, the likelihood ratio of the bi-cluster Gaussian mixture model to the uni-cluster Gaussian model is calculated using the following formula:

[0075] .

[0076] in, It represents the likelihood ratio.

[0077] In another exemplary embodiment of this application, based on the class-balanced buffer for each type of grain crop at the current time, the intersection of two Gaussian distribution parameters is fitted using a bi-cluster Gaussian mixture model to obtain the optimal threshold for the harvest index of each type of grain crop at the current time, specifically including:

[0078] Calculate the harvest index for all monitored pixels within the class balance buffer for each type of grain crop.

[0079] For each type of grain crop, the expectation-maximization algorithm is used to fit the harvest index of all monitored pixels in the balanced buffer using a dual-cluster Gaussian mixture model to obtain the Gaussian distribution parameters of the harvested clusters and the Gaussian distribution parameters of the unharvested clusters at the current time. The Gaussian distribution parameters include the mean, variance, and weight of each distribution.

[0080] For each type of grain crop, based on the Gaussian distribution parameters of the harvested clusters and the unharvested clusters at the current time, the probability density functions of the harvested clusters and the unharvested clusters at the current time are constructed respectively.

[0081] For each type of grain crop, the harvest index value corresponding to the intersection of the probability density function of the harvested cluster and the probability density function of the unharvested cluster at the current moment is taken as the optimal threshold for the grain crop type, thus obtaining the optimal threshold for the harvest index of each type of grain crop at the current moment.

[0082] The following example illustrates this application using a specific process for monitoring the harvest progress of grain crops.

[0083] The monitoring data for the current season in this application requires two input data: crop type distribution map data and multispectral remote sensing image data; specifically, it can be a crop type distribution map for the current season and a Sentinel-2 multispectral image acquired in near real-time.

[0084] This method mainly includes the following steps. A schematic diagram of the main process (steps 1-2) of one method for monitoring the harvest progress of grain crops is shown below. Figure 3 As shown.

[0085] Step 1: Obtain the distribution map of seasonal food crop types and Sentinel-2 multispectral imagery.

[0086] Specifically, a distribution map of seasonal food crop types is obtained through mid-season remote sensing crop classification techniques or field surveys. Near-real-time Sentinel-2 multispectral remote sensing data is downloaded from the European Space Agency website (https: / / browser.dataspace.copernicus.eu / ) as the harvest season approaches. The revisit frequency of the remote sensing images (ideally within 10 days, otherwise progress monitoring will be difficult) and the presence of a 1.6µm shortwave infrared band characterizing moisture absorption are required.

[0087] Step 2: Based on the crop type and multispectral imagery from Step 1, determine the reference reflectance of unharvested fields for each type of grain crop.

[0088] Specifically, the Red Band Depth (RBD) and Land Surface Water Index are calculated for all pixels in the image, as shown in the following formulas:

[0089] .

[0090] .

[0091] .

[0092] in, Indicates the red band depth index; , , and These represent the reflectivity of the red light band, the green light band, the near-infrared band, and the short-wave infrared band (1.6µm), respectively (corresponding to bands 4, 3, 8, and 11 of the Sentinel-2 sensor). This represents the red light band interpolated reflectance, which is the reflectance obtained by linearly interpolating the green light and near-infrared band reflectances according to wavelength. Since unharvested fields have higher photosynthetic biomass and water content, their RBD and LSWI values ​​are correspondingly higher. Based on the crop type distribution map, pixels with RBD and LSWI indices higher than the top 5% for each crop type were selected as representative reference unharvested pixels. The average red light and short-wave infrared reflectance of these reference unharvested pixels were calculated as the reference red light reflectance and reference short-wave infrared reflectance for each crop type. , ); , , and represent the center wavelengths of the near-infrared band (833nm), the green band (560nm), and the red band (665nm), respectively. This represents the surface moisture index. It is the reflectance obtained by linear interpolation of green light and near-infrared reflectance based on wavelength.

[0093] Step 3: Calculate the harvest index based on the reference reflectance of unharvested fields for each type of grain crop in Step 2, and the grain crop type and multispectral image in Step 1.

[0094] Specifically, by calculating the position of the pixel to be monitored in the red light band (Sentinel 2 band 4), ) and shortwave infrared band (Sentinel 2 band 11, The reflectance of the crop and the reference reflectance of the unharvested field of the corresponding grain crop type obtained in step 2 are compared. , The Harvest Index is constructed by multiplying the absolute differences of the two factors. HI As shown in the following formula:

[0095] .

[0096] Because harvested remote sensing pixels exhibit significant differences in spectral characteristics between harvested and unharvested pixels in the red and shortwave infrared (SWIR1) bands, a schematic diagram comparing the spectra of harvested and unharvested fields was drawn, as shown below. Figure 4 As shown. Figure 4 (a) is a schematic diagram of the distribution of harvested and unharvested rice fields in the rice planting area. Different colors are used to mark the specific locations and ranges of unharvested fields, harvested fields (stubble), and harvested fields (bare soil). Figure 4 (b) is Figure 4 (a) Comparison of spectral reflectance of different types of fields in the rice planting area clearly shows the changes in spectral reflectance of various fields with wavelength under Red and SWIR1 bands. Figure 4 (c) A schematic diagram of the distribution of harvested and unharvested corn fields is presented, with different colors used to distinguish fields in different harvest states. Figure 4 (d) is Figure 4 (c) Comparison of spectral reflectance of different types of fields in the maize planting area helps to analyze the differences in spectral characteristics of maize fields at different harvest stages. Figure 4 (e) is a schematic diagram showing the distribution of harvested and unharvested fields in the wheat planting area, which clarifies the distribution of the harvest status of wheat fields. Figure 4 (f) is Figure 4 (e) shows a comparison of spectral reflectance of different types of wheat fields in the wheat-growing area, reflecting the spectral reflectance characteristics of wheat fields under different harvest conditions. In actual analysis, the larger the harvest index (HI) of a pixel, the greater the likelihood that the pixel has been harvested. By combining these spectral comparison maps and the HI index, the harvest status of the field can be determined more accurately.

[0097] Step 4: Based on the harvest index in Step 3, automatically focus on the harvested / unharvested class balance buffer.

[0098] During the crop harvesting process, the proportion of harvested and unharvested pixels within the study area does not remain balanced but is dynamically changing. For example, in the early stages of harvesting, the proportion of unharvested pixels is large, while in the late stages, harvested pixels dominate, resulting in a significant imbalance between the two types. In this case, the harvest index histogram often exhibits a unimodal distribution, increasing the difficulty of threshold identification. To address this issue, a method is designed to automatically focus on a buffer zone containing a balanced number of harvested / unharvested pixels (with a clearly bimodal distribution within the buffer zone) to obtain the optimal threshold. This threshold is then applied to the entire area. A schematic diagram of the optimal threshold determination method for the harvest index is shown below. Figure 5As shown in the diagram, specifically, firstly, pixels with a harvest index higher than the top 5% quantile (referred to as "long-tail data") within the entire remote sensing image are extracted as potential harvested pixels; secondly, based on the spatial coordinates of these pixels, the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering algorithm is applied to obtain the spatial cluster centers of the potential harvested pixels; finally, square buffers are generated with the spatial cluster centers as anchor points (the area gradually expands from 0.5 km² to 4.5 km²), and based on the HI index data within each buffer area, the distribution is fitted using a dual-cluster Gaussian Mixture Model (GMM) and a single-cluster Gaussian model, respectively.

[0099] The single-cluster Gaussian model fits the HI exponential data distribution within the buffer using a single normal distribution.

[0100] .

[0101] The harvest index represents the single-cluster Gaussian mixture model. The probability density function is used to characterize the distribution of the harvest index within the buffer zone. Among them, It is the standard symbol for the probability density function of the normal distribution; This represents the mean parameter of a single-cluster Gaussian model; This represents the variance parameter of a single-cluster Gaussian model; the normal distribution... mean ( ) and variance The maximum likelihood estimate is the mean and variance of HI for all pixels in the buffer. The harvest index represents the single-cluster Gaussian mixture model. The probability density function is used to characterize the distribution of the harvest index within the buffer. Therefore, the likelihood function of a single-cluster Gaussian mixture model... for:

[0102] .

[0103] in, The first in the class balanced buffer i N represents the total number of cells in the class-balanced buffer; N represents the total number of cells in the class-balanced buffer.

[0104] Biclustered GMMs fit the HI exponential data distribution within a buffer using two normal distributions:

[0105] .

[0106] in, The weight parameter represents the unharvested pixels; This represents the mean parameter of the unharvested pixels; The variance parameter representing the unharvested pixels; The weight parameters represent the pixels that have been captured; The mean parameter representing the number of pixels already captured; This represents the variance parameter of the pixels that have been harvested. and ( ) represent the weight parameters (i.e., pixel proportions) for category 0 (unharvested) and category 1 (harvested), respectively. and These represent the mean values ​​of HI for cells in class 0 and class 1, respectively. and Let HI represent the variances of pixels in class 0 and class 1, respectively. The parameters of the bi-cluster Gaussian mixture model ( It is obtained by the Expectation Maximization (EM) algorithm, and the specific steps are as follows:

[0107] 1) Initialize parameters: , and Take the lower 25% and upper 25% quantile values ​​of HI in the buffer, respectively. and The variance of HI within the buffer is taken, and the iteration step is set to t=0.

[0108] 2) Step E: Calculate each cell i Posterior probabilities of belonging to class 0 and class 1 ( and ).

[0109] .

[0110] 3) M step: based on and Update parameters:

[0111] Weight parameters .

[0112] Mean parameter .

[0113] variance parameter .

[0114] 4) Convergence judgment: Repeat steps 2) and 3) until the changes in the estimated parameters in two iterations are both less than a certain threshold. ),Right now and and and and and When the iteration stops, the iteration stops.

[0115] Based on the estimated distribution parameters, the likelihood function of the bi-cluster Gaussian mixture model can be calculated. .

[0116] .

[0117] For each buffer, calculate the Log-Likelihood Ratio (LLR) between the fitted bi-cluster GMM and the uni-cluster Gaussian distribution:

[0118] .

[0119] Selecting buffers with LLR values ​​greater than the top 5 percentile of all LLR values ​​as quasi-balanced buffers indicates that the HI distribution of the harvest index within these buffers better conforms to the bimodal distribution characteristics.

[0120] Step 5: Based on the balanced buffer in Step 4, determine the optimal threshold for the harvest index using a bi-cluster Gaussian mixture model.

[0121] Specifically, all pixels in the class-balanced buffer obtained in step 4 are summarized, and the parameters of the two Gaussian distributions are determined by fitting a dual-cluster GMM (the fitting algorithm is the same as the EM estimation algorithm in step 4). The HI value corresponding to the intersection of the two Gaussian distributions is used as the optimal classification threshold for the harvest index. ). To find the solution to the following equation:

[0122] .

[0123] Solving for the given information yields:

[0124] .

[0125] in, Indicates intermediate calculation quantities. .

[0126] Step 6: Based on the optimal classification threshold of the harvest index in Step 5, determine whether the pixels have been harvested or not, thereby realizing the monitoring of the harvest progress of grain crops.

[0127] Specifically, the harvest index of all pixels to be monitored in the image is compared with the optimal classification threshold obtained in step 5. In contrast, if the harvest index is higher than the optimal classification threshold ( If a pixel is not harvested, it is considered a harvested pixel; otherwise, it is considered an unharvested pixel, thus achieving near real-time monitoring of the harvest progress of grain crops.

[0128] like Figure 6 As shown, a schematic diagram of the study area is provided. The main food crop in this area is winter wheat, which is typically harvested in early to mid-June. Due to the relatively stable planting structure in this area, a winter wheat map from 2020 was used as the crop distribution map input. Four Sentinel-2 images taken on June 1, 6, 11, and 16, 2022, were used as remote sensing data input to monitor the winter wheat harvest progress in 2022. Furthermore, a field survey was conducted during the 2022 winter wheat harvest to obtain the harvest dates of 57 sample plots, which were used to validate the method described in this application.

[0129] Based on the proposed method, winter wheat harvest identification was performed on four Sentinel-2 images taken on June 1, 6, 11, and 16, 2022, respectively, thus obtaining the harvest date for each pixel. For example... Figure 7 As shown, the proposed harvest period gradually postpones from south to north, which is consistent with the harvest pattern in this region. Based on the verification points from the field survey, the classification accuracy of harvested / unharvested types on images of different dates is shown in Table 1 below. It can be seen that the method of this application has an accuracy of over 90% in identifying harvested pixels on images of different dates (Note: By June 16, the study area was almost fully harvested, and unharvested samples were not included in the field survey points, so there are no valid values ​​for the unharvested type F1 and the average F1), which demonstrates the rationality of the method.

[0130] Table 1. Classification accuracy of harvested / unharvested types on images from different dates.

[0131]

[0132] This application also provides an application scenario in which the above-mentioned method for monitoring the harvest progress of grain crops is applied. Specifically, the method for monitoring the harvest progress of grain crops provided in this embodiment can be applied in the scenario of precision agricultural management and decision support. The scenario of precision agricultural management and decision support includes a data acquisition stage, a data processing and analysis stage, and a decision-making and execution stage. Crop planting-related data enters the system from the data acquisition stage, and after a series of data cleaning, integration, modeling and analysis processes, key information such as crop growth status and harvest progress is obtained, and then enters the decision-making and execution stage. The method for monitoring the harvest progress of grain crops provided in this embodiment is a core step in the data processing and analysis stage. Specifically, by acquiring monitoring data of the current season, including crop type distribution map data and multispectral remote sensing image data, relevant indices are calculated based on these data to determine the reference reflectance and harvest index, and then the optimal threshold of the harvest index is obtained. Finally, the harvest status of the grain crop corresponding to each pixel is determined, providing accurate and timely data support for subsequent agricultural production decisions and helping to achieve scientific and refined management of agricultural production.

[0133] Based on the same inventive concept, this application also provides a grain crop harvest progress monitoring device for implementing the above-described grain crop harvest progress monitoring method. The solution provided by this device is similar to the solution described in the above-described method. Therefore, the specific limitations of one or more grain crop harvest progress monitoring device embodiments provided below can be found in the limitations of the grain crop harvest progress monitoring method described above, and will not be repeated here.

[0134] In one exemplary embodiment, such as Figure 8 As shown, a monitoring device for the harvesting progress of grain crops is provided, comprising:

[0135] The raw data acquisition module 301 is used to acquire the distribution map data of grain crop types and multispectral remote sensing image data for a preset number of days before harvest.

[0136] The reference reflectance determination module 302 is used to determine the reference reflectance of unharvested fields of each type of grain crop based on the distribution map data of grain crop types and multispectral remote sensing image data of a preset number of days before harvest, by calculating the red band depth index and the surface moisture index; the grain crop types include corn, rice and wheat.

[0137] The remote sensing image data acquisition module 303 is used to acquire the multispectral remote sensing image data to be monitored at the current time during the harvest period.

[0138] The harvest index calculation module 304 is used to calculate the harvest index of each type of grain crop at the current time by performing difference calculations on the red band reflectance and shortwave infrared band reflectance of each pixel to be monitored in the multispectral remote sensing image data to be monitored at the current time and the reference reflectance of the unharvested field of the corresponding grain crop type.

[0139] The optimal threshold determination module 305 is used to obtain the optimal threshold for the harvest index of each type of grain crop at the current time by fitting the intersection of two Gaussian distribution parameters through a dual-cluster Gaussian mixture model based on the harvest index of each type of grain crop at the current time.

[0140] The harvest progress monitoring module 306 is used to determine the harvest status of the grain crop corresponding to each monitored pixel at the current time based on the harvest index of each type of grain crop at the current time and the corresponding optimal threshold of the harvest index, and to obtain the grain crop harvest progress monitoring results.

[0141] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 9As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores data related to the progress of grain harvesting. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network. When executed by the processor, the computer program implements a method for monitoring the progress of grain harvesting.

[0142] Those skilled in the art will understand that Figure 9 The structures shown are merely block diagrams of some structures related to the present application and do not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0143] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0144] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0145] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0146] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0147] The technical features of the above embodiments can be combined in any way. For the sake of brevity, 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.

[0148] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for monitoring the harvest progress of grain crops, characterized in that, The methods for monitoring the progress of grain crop harvesting include: Acquire distribution map data and multispectral remote sensing image data of grain crop types a preset number of days before harvest; Based on the distribution map data of grain crop types and multispectral remote sensing image data of a predetermined number of days before harvest, the reference reflectance of unharvested fields for each grain crop type is determined by calculating the red band depth index and surface moisture index, specifically including: The red band depth index and surface moisture index are calculated using the following formulas: ； ； ； in, Indicates the red band depth index; , , and These represent the reflectance in the red light band, green light band, near-infrared band, and short-wave infrared band, respectively. Indicates the interpolated reflectance in the red light band; , , and represent the center wavelengths of the near-infrared band, the green band, and the red band, respectively; Indicates the surface moisture index; Based on the distribution map data of grain crop types with a preset number of days before harvest, pixels of each grain crop type whose red band depth index and surface moisture index are both greater than the preset unharvested threshold are selected as reference unharvested pixels. The average red band reflectance and short-wave infrared reflectance of all reference unharvested pixels corresponding to each type of grain crop were calculated to obtain the average red band reflectance and average short-wave infrared reflectance of each type of grain crop. The average red band reflectance and average shortwave infrared band reflectance corresponding to each type of grain crop are used as the red band reference reflectance and shortwave infrared band reference reflectance for the corresponding grain crop type, respectively; the grain crop types include corn, rice, and wheat. During the harvest period, acquire multispectral remote sensing image data to be monitored at the current moment; By calculating the difference between the red band reflectance and shortwave infrared band reflectance of each pixel in the multispectral remote sensing image data to be monitored at the current time and the reference reflectance of the unharvested field of the corresponding grain crop type, the harvest index of each grain crop type at the current time is obtained. Based on the harvest index of each type of grain crop at the current moment, the intersection of two Gaussian distribution parameters is fitted by a dual-cluster Gaussian mixture model to obtain the optimal threshold for the harvest index of each type of grain crop at the current moment. Based on the harvest index of each type of grain crop at the current moment and the corresponding optimal threshold for the harvest index, the harvest status of the grain crop corresponding to each monitored pixel at the current moment is determined, and the monitoring results of the grain crop harvest progress are obtained.

2. The method for monitoring the harvest progress of grain crops according to claim 1, characterized in that, By calculating the difference between the red band reflectance and shortwave infrared band reflectance of each pixel in the multispectral remote sensing image data to be monitored at the current time and the reference reflectance of the corresponding unharvested field of grain crop type, the harvest index of each grain crop type at the current time is obtained, specifically including: The harvest index is calculated using the following formula: ； in, Indicates the harvest index; and These represent the reflectivity in the red light band and the reflectivity in the short-wave infrared band, respectively. and These represent the reference reflectance in the red light band and the reference reflectance in the short-wave infrared band, respectively.

3. The method for monitoring the harvest progress of grain crops according to claim 1, characterized in that, Based on the harvest index of each type of grain crop at the current moment, the intersection points of the two Gaussian distribution parameters are fitted using a bi-cluster Gaussian mixture model to obtain the optimal threshold for the harvest index of each type of grain crop at the current moment, specifically including: Based on the harvest index of each type of grain crop at the current moment, a class-balanced buffer is determined for the harvest status of each type of grain crop at the current moment; the harvest status includes harvested status and unharvested status. Based on the class-balanced buffer for each type of grain crop at the current moment, the intersection of two Gaussian distribution parameters is fitted by a dual-cluster Gaussian mixture model to obtain the optimal threshold for the harvest index of each type of grain crop at the current moment.

4. The method for monitoring the harvest progress of grain crops according to claim 3, characterized in that, Based on the harvest index of each type of grain crop at the current moment, a class-balanced buffer is determined for the harvest status of each type of grain crop at the current moment, specifically including: For each type of grain crop, the following steps are used to obtain the class-balanced buffer of the harvest status of each type of grain crop at the current moment: In the multispectral remote sensing image data to be monitored at the current moment, the pixels to be monitored that have a harvest index greater than a preset high threshold for the current type of grain crop are extracted as potential harvested pixels of the current type of grain crop. Spatial clustering algorithm is used to cluster the potential harvested pixels of the current grain crop type, dividing the potential harvested pixels into different clusters and obtaining the spatial cluster center of each cluster. Using each cluster center as an anchor point, square buffer zones are gradually expanded outward at preset area intervals until the preset maximum area is reached, resulting in multiple buffer zones of different areas. Within each buffer, the harvest exponent distribution is fitted using a bi-cluster Gaussian mixture model and a uni-cluster Gaussian model, respectively, and the likelihood ratio between the bi-cluster Gaussian mixture model and the uni-cluster Gaussian model is calculated. Sort all buffers by likelihood ratio from largest to smallest, and select the buffers whose likelihood ratio is within a preset percentage threshold as the class balance buffers for the current crop type at the current moment.

5. The method for monitoring the harvest progress of grain crops according to claim 3, characterized in that, Based on the current equilibrium buffer for each type of grain crop, the intersection of two Gaussian distribution parameters is fitted using a bi-cluster Gaussian mixture model to obtain the optimal threshold for the harvest index of each type of grain crop at the current time, specifically including: Calculate the harvest index of all monitored pixels within the class balance buffer for each type of grain crop; For each type of grain crop, the expectation-maximization algorithm is used to fit the harvest index of all pixels to be monitored in the balanced buffer using a dual-cluster Gaussian mixture model to obtain the Gaussian distribution parameters of the harvested clusters and the Gaussian distribution parameters of the unharvested clusters at the current time. The Gaussian distribution parameters include the mean, variance, and weight of each distribution. For each type of grain crop, based on the Gaussian distribution parameters of the harvested clusters and the unharvested clusters at the current time, the probability density functions of the harvested clusters and the unharvested clusters at the current time are constructed respectively. For each type of grain crop, the harvest index value corresponding to the intersection of the probability density function of the harvested cluster and the probability density function of the unharvested cluster at the current moment is taken as the optimal threshold for the grain crop type, thus obtaining the optimal threshold for the harvest index of each type of grain crop at the current moment.

6. The method for monitoring the harvest progress of grain crops according to claim 4, characterized in that, Within each buffer, the harvest exponent distribution is fitted using both a bi-cluster Gaussian mixture model and a uni-cluster Gaussian model. The likelihood ratio between the bi-cluster Gaussian mixture model and the uni-cluster Gaussian model is calculated, specifically including: The likelihood functions of the unicluster Gaussian mixture model and the bicluster Gaussian mixture model are calculated using the following formulas: ； ； in, and Let represent the likelihood function values ​​of the unicluster Gaussian model and the bicluster Gaussian mixture model, respectively. The first in the class balanced buffer i N represents the total number of cells in the class balance buffer; This represents the probability density function of the normal distribution. Indicates the harvest index; This represents the mean parameter of a single-cluster Gaussian model; This represents the variance parameter of a single-cluster Gaussian model; The weight parameter represents the unharvested pixels; This represents the mean parameter of the unharvested pixels; The variance parameter representing the unharvested pixels; The weight parameters represent the pixels that have been captured; The mean parameter representing the number of pixels already captured; The variance parameter represents the number of pixels that have been harvested; For each buffer, the likelihood ratio of the bi-cluster Gaussian mixture model to the uni-cluster Gaussian model is calculated using the following formula: ； in, It represents the likelihood ratio.

7. A monitoring device for the harvest progress of grain crops, characterized in that, The grain crop harvest progress monitoring device uses the grain crop harvest progress monitoring method as described in any one of claims 1-6, and the grain crop harvest progress monitoring device comprises: The raw data acquisition module is used to acquire distribution map data of grain crop types and multispectral remote sensing image data for a preset number of days before harvest. The reference reflectance determination module is used to determine the reference reflectance of unharvested fields for each type of grain crop based on distribution map data of grain crop types and multispectral remote sensing image data from a preset number of days before harvest. This is achieved by calculating the red band depth index and surface moisture index. Specifically, it includes: The red band depth index and surface moisture index are calculated using the following formulas: ； ； ； in, Indicates the red band depth index; , , and These represent the reflectance in the red light band, green light band, near-infrared band, and short-wave infrared band, respectively. Indicates the interpolated reflectance in the red light band; , , and represent the center wavelengths of the near-infrared band, the green band, and the red band, respectively; Indicates the surface moisture index; Based on the distribution map data of grain crop types with a preset number of days before harvest, pixels of each grain crop type whose red band depth index and surface moisture index are both greater than the preset unharvested threshold are selected as reference unharvested pixels. The average red band reflectance and short-wave infrared reflectance of all reference unharvested pixels corresponding to each type of grain crop were calculated to obtain the average red band reflectance and average short-wave infrared reflectance of each type of grain crop. The average red band reflectance and average shortwave infrared band reflectance corresponding to each type of grain crop are used as the red band reference reflectance and shortwave infrared band reference reflectance for the corresponding grain crop type, respectively; the grain crop types include corn, rice, and wheat. The remote sensing image data acquisition module is used to acquire multispectral remote sensing image data to be monitored at the current time during the harvest period. The harvest index calculation module is used to calculate the harvest index of each type of grain crop at the current time by performing a difference calculation between the red band reflectance and shortwave infrared band reflectance of each pixel in the multispectral remote sensing image data to be monitored at the current time and the reference reflectance of the unharvested field of the corresponding grain crop type. The optimal threshold determination module is used to obtain the optimal threshold for the harvest index of each type of grain crop at the current time by fitting the intersection of two Gaussian distribution parameters through a dual-cluster Gaussian mixture model based on the harvest index of each type of grain crop at the current time. The harvest progress monitoring module is used to determine the harvest status of the grain crop corresponding to each monitored pixel at the current time based on the harvest index of each type of grain crop at the current time and the corresponding optimal threshold of the harvest index, and to obtain the grain crop harvest progress monitoring results.

8. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the method for monitoring the harvest progress of grain crops according to any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the method for monitoring the harvest progress of grain crops as described in any one of claims 1-6.

Images