An oilseed rape and sunflower automatic recognition method and device based on cooperation of agronomic knowledge and optical radar

Abstract

The application discloses a rape and sunflower automatic identification method and device based on cooperation of agronomic knowledge and optical radar. The method is based on Sentinel-1 radar and Sentinel-2 optical time series data, first, the crop growth cycle and the key discrimination window are adaptively determined according to the vegetation-bare soil time series change; then, around the flower crown layer and the high and large crown layer structure characteristics of rape and sunflower, the flower crown layer optical response index, the high and large crown layer volume scattering enhanced index and the crown layer scattering density index are constructed; finally, the multi-level discrimination rule is established, and the rape and sunflower automatic identification is realized. The method has the advantages of strong interpretability, low sample requirement and good cross-region applicability, and can be used for large-scale oil crop monitoring.

Description

Technical Field

[0001] This invention relates to the field of agricultural remote sensing technology, and in particular to an automatic identification method and device for rapeseed and sunflower based on agronomic knowledge and the collaboration of optical radar. Background Technology

[0002] Rapeseed and sunflower are both important oilseed crops, serving as crucial sources of rapeseed oil and sunflower seed oil, respectively. Rapeseed is one of the world's major oilseed crops, providing edible oil, feed, and ecological value through its flowering period. It is widely cultivated in countries and regions such as China, Canada, India, and Europe. In my country, rapeseed is a major oilseed crop and a significant source of edible vegetable oil for residents, with the Yangtze River Basin and the Huang-Huai-Hai Plain being key planting areas. Sunflower, as an important oilseed and economic crop, possesses both ornamental and edible value. Originally from South America, it is now cultivated in countries such as Ukraine, Russia, Argentina, and China. In my country, sunflowers are mainly distributed in North and Northwest China, exhibiting characteristics such as light preference and drought resistance, playing a vital role in conserving agricultural water resources and ensuring food supply. Since the 21st century, the planting area of ​​both rapeseed and sunflower in my country has fluctuated, with sunflower cultivation maintaining an overall growth trend after periodic adjustments. Timely and accurate monitoring of the planting distribution of these two crops is beneficial for further optimizing the layout and structure of oilseed crops, expanding the industry scale, and ensuring national food and oil security.

[0003] Traditionally, information on crop planting area distribution has relied primarily on manual field surveys or sampling statistics, which is costly, inefficient, and susceptible to subjective influences. With the increasing abundance and quality of time-series remote sensing imagery, remote sensing-based crop identification technology has become a crucial means for high-precision automatic acquisition of agricultural information. Existing research has proposed various crop mapping indices for major crops such as rice, winter wheat, and corn, combining cultivation management, growth and development characteristics, and their remote sensing response relationships. This has enabled the extraction of large-area crop distribution under conditions of limited or no samples. However, current research on crop remote sensing mapping still mainly focuses on major crops, with relatively insufficient research on the collaborative identification of rapeseed and sunflower. Rapeseed exhibits significant phenological differences across different regions and is easily confused with winter wheat and other flowering crops. Existing identification methods largely rely on optical images during the flowering period, which can lead to serious omissions when images are missing during key periods. Sunflower, as a minor oilseed crop, has a scattered planting distribution, and field samples and statistical data are relatively scarce. Currently, there are few mature methods suitable for large-scale automatic identification. Overall, existing technologies generally suffer from problems such as strong dependence on fixed phenological dates, insufficient utilization of multi-source remote sensing data, and limited cross-regional migration capabilities. There is still a lack of an automatic identification method for oil crops that can integrate optical and radar data, take into account the collaborative identification of rapeseed and sunflower, and be applicable to large-scale applications. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide an automatic identification method and device for rapeseed and sunflower based on agronomic knowledge and the collaboration of optical radar. Starting from the characteristics of the corolla layer, the canopy structure attributes and the differences in their temporal remote sensing responses, an automatic identification technology solution for rapeseed and sunflower is constructed.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: an automatic identification method for rapeseed and sunflower based on agronomic knowledge and optical radar, comprising the following steps:

[0006] Step S01: Construct a time-series dataset of radar and optical images;

[0007] Step S02: Establish a spatiotemporal-feature three-dimensional data structure for optical radar fusion;

[0008] Step S03: Determine the crop growth cycle based on the temporal changes in vegetation and bare soil;

[0009] Step S04: Determine the key phenological window by combining agronomic knowledge;

[0010] Step S05: Extract crop time-series features within the key phenological window;

[0011] Step S06: Construct the optical response index of the crop canopy layer;

[0012] Step S07: Construct a crop tall canopy scattering enhancement index;

[0013] Step S08: Construct the crop canopy scattering density index;

[0014] Step S09: Establish a rapeseed and sunflower grading framework driven by agronomic knowledge;

[0015] Step S10: Automatic identification of rapeseed and sunflower based on the hierarchical discrimination framework.

[0016] In a preferred embodiment, in step S02, based on the radar and optical image time-series dataset constructed in step S01, the normalized vegetation index, the corrected normalized differential bare soil index, and the VV and VH backscattering coefficient features of the radar images are extracted. The extracted optical index features and radar backscattering features are registered and aligned according to a unified time series to establish a spatiotemporal-feature three-dimensional data structure for characterizing pixel temporal changes. The first dimension is the spatial dimension, corresponding to the image pixel position; the second dimension is the temporal dimension, corresponding to the time of acquisition of the time-series images; and the third dimension is the feature dimension, including optical index features and radar backscattering features.

[0017] In a preferred embodiment, in step S03, based on the normalized vegetation index and the modified normalized difference bare soil index extracted in step S02, a crop cover index (GPCCI) for growth stages is constructed; the crop cover start time, crop cover end time, and cover duration are identified according to the time-series variation curve of the GPCCI to determine the crop growth cycle; and the crop growth cycle is adaptively adjusted in conjunction with the normalized vegetation index to determine the key growth stages of the crop; in step S04, based on the crop growth cycle determined in step S03, the temporal response differences of rapeseed, sunflower, and other crops in the target area are compared and analyzed in conjunction with the phenological characteristics and canopy structure differences of rapeseed and sunflower, and the time period with the most significant difference between the target crop and non-target crops is determined as the key phenological discrimination window.

[0018]

[0019] Wherein, NDVI is the Normalized Difference Vegetation Index, and MNDBI is the Modified Normalized Difference Bare Soil Index. This is a compensation factor.

[0020] In a preferred embodiment, in step S05, based on the key phenological window determined in step S04, optical and radar time-series features of the crop are extracted; wherein the optical time-series features include at least the red-edge band features and the green chlorophyll vegetation index (GCVI) features in Sentinel-2 multispectral data, and the radar time-series features include at least the VV and VH backscattering coefficient features of Sentinel-1 imagery, so as to form the optical and radar time-series features of the target crop within the key phenological window;

[0021]

[0022] in, , These are the green light band reflectance and near-infrared band reflectance of Sentinel-2, respectively.

[0023] In a preferred embodiment, in step S06, based on the optical time-series features extracted in step S05, the red-edge band response information and chlorophyll concentration response information are fused to construct the crop corolla optical response index CLORI, which is used to characterize the difference in corolla optical response between rapeseed and sunflower within the key discriminative phenological window, thereby achieving the distinction between flowering crops and non-flowering crops.

[0024]

[0025] in, The red-edge reflectivity of Sentinel-2.

[0026] In a preferred embodiment, in step S07, based on the radar time-series features extracted in step S05, the difference between the backscattering coefficients of VV and VH is used to construct the crop tall canopy volume scattering enhancement index TCVSEI, which is used to enhance the volume scattering characteristics of tall canopy crops and suppress the scattering interference of dwarf flowering crops, thereby achieving the distinction between tall flowering crops and dwarf flowering crops.

[0027]

[0028] Where VV and VH represent the backscattering coefficients of radar VV and VH polarizations, respectively.

[0029] In a preferred embodiment, in step S08, based on the radar time-series features extracted in step S05, the VV and VH backscattering coefficients are fused to construct the crop canopy scattering density index CSDI, which is used to characterize the difference in canopy scattering density between rapeseed and sunflower, thereby achieving the distinction between rapeseed and sunflower.

[0030]

[0031] Where VV and VH represent the backscattering coefficients of radar VV and VH polarizations, respectively.

[0032] In a preferred embodiment, in step S09, based on agronomic knowledge, the crop canopy optical response index CLORI constructed in step S06, the crop tall canopy body scattering enhancement index TCVSEI constructed in step S07, and the crop canopy scattering density index CSDI constructed in step S08 are integrated to establish a multi-level classification discrimination rule; the multi-level classification discrimination rule includes the exclusion of non-flowering crops, the exclusion of dwarf flowering crops, and the distinction between rapeseed and sunflower, so as to form a classification discrimination framework for rapeseed and sunflower.

[0033] In a preferred embodiment, in step S10, the discrimination thresholds for the crop canopy optical response index CLORI, the crop tall canopy body scattering enhancement index TCVSEI, and the crop canopy scattering density index CSDI are determined based on field sample data. A discrimination function for rapeseed and sunflower is then constructed based on these thresholds to achieve automatic identification of rapeseed and sunflower. The output categories of the discrimination function include at least non-flowering crops, dwarf flowering crops, rapeseed, and sunflower.

[0034]

[0035] The output categories of the discrimination functions for rapeseed and sunflower include at least non-flowering crops, dwarf flowering crops, rapeseed, and sunflower, and the discrimination threshold is determined based on the distribution characteristics of field sample data. , , For fixed values, , , The suggested values ​​are 13.5±0.5, 0.1±0.5, and 0.35±0.5, respectively.

[0036] This invention also provides an automatic identification device for rapeseed and sunflower based on agronomic knowledge and optical radar, which executes the aforementioned automatic identification method for rapeseed and sunflower based on agronomic knowledge and optical radar; including:

[0037] The optical radar time-series data processing module is used to acquire and preprocess radar and optical remote sensing time-series data;

[0038] The 3D data structure establishment module is used to perform time alignment of optical features and radar features, and to establish a spatiotemporal-feature 3D data structure for optical-radar fusion.

[0039] The crop growth cycle determination module is used to determine the crop growth cycle based on the temporal relationship between vegetation and bare soil.

[0040] The critical phenological window determination module is used to determine the critical phenological window by combining agronomic knowledge;

[0041] The temporal feature extraction module is used to extract the temporal features of crop optical radar within the key phenological discrimination window;

[0042] The crop canopy optical response index construction module is used to construct the crop canopy optical response index.

[0043] A module for constructing crop tall canopy scattering enhancement indices is used to construct crop tall canopy scattering enhancement indices.

[0044] The crop canopy scattering density index construction module is used to construct the crop canopy scattering density index.

[0045] The grading and discrimination framework establishment module is used to establish a grading and discrimination framework for rapeseed and sunflower based on the crop canopy optical response index, crop tall canopy scattering enhancement index and crop canopy scattering density index.

[0046] An automatic identification module is used to automatically identify rapeseed and sunflower based on the hierarchical discrimination framework.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] (1) An adaptive identification mechanism for key phenological periods of crops is proposed: by integrating the temporal change information of vegetation and bare soil, the crop growth cycle and key phenological period discrimination window are adaptively determined, which reduces the dependence of the method on fixed phenological dates and regional prior phenological calendars.

[0049] (2) A multi-level knowledge-driven method for distinguishing rapeseed and sunflower is proposed: by constructing the crop canopy optical response index, the crop tall canopy scattering enhancement index and the crop canopy scattering density index, and combining the progressive discrimination logic of canopy characteristics, tall canopy scattering characteristics and canopy scattering density differences, rapeseed and sunflower can be automatically identified, which has strong interpretability and potential for promotion and application.

[0050] (3) Achieving multi-source synergy between optics and SAR and cross-regional promotion: Fully leveraging the complementary advantages of optical data's sensitivity to changes in the canopy layer and chlorophyll, and radar data's stable response to canopy structure and volume scattering, this invention still possesses good potential for cross-regional and cross-year applications even with limited sample sizes. Compared to existing identification methods that rely solely on the rapeseed flowering index, this invention can still characterize crop canopy structure differences using radar data even when optical images of key flowering periods are missing. This effectively reduces the risk of missed identification of rapeseed and improves the stability of sunflower identification, enabling large-scale, high-precision automatic mapping of rapeseed and sunflower. Attached Figure Description

[0051] Figure 1 This is a schematic diagram illustrating the implementation process of an embodiment of the present invention;

[0052] Figure 2 Radar timing curves for rapeseed and sunflower;

[0053] Figure 3 This is a schematic diagram of the device structure according to an embodiment of the present invention;

[0054] Figure 4 This invention presents a spatial distribution map of rapeseed and sunflower in a certain city of a certain province. Detailed Implementation

[0055] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0056] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0057] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations according to this application; as used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise; furthermore, it should be understood that when the terms “comprising” and / or “including” are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0058] This invention provides an automatic identification method for rapeseed and sunflower based on agronomic knowledge-driven and optical radar-assisted identification, comprising the following steps (see...). Figures 1-4 ):

[0059] Step S01: Construct a time-series dataset of radar and optical images;

[0060] Step S02: Establish a spatiotemporal-feature three-dimensional data structure for optical radar fusion;

[0061] Step S03: Determine the crop growth cycle based on the temporal changes in vegetation and bare soil;

[0062] Step S04: Determine the key phenological window by combining agronomic knowledge;

[0063] Step S05: Extract crop time-series features within the key phenological window;

[0064] Step S06: Construct the optical response index of the crop canopy layer;

[0065] Step S07: Construct a crop tall canopy scattering enhancement index;

[0066] Step S08: Construct the crop canopy scattering density index;

[0067] Step S09: Establish a rapeseed and sunflower grading framework driven by agronomic knowledge;

[0068] Step S10: Automatic identification of rapeseed and sunflower based on the hierarchical discrimination framework.

[0069] This invention also provides an automatic identification device for rapeseed and sunflower based on agronomic knowledge and in conjunction with optical radar, comprising:

[0070] Based on the rapeseed and sunflower automatic identification method driven by agronomic knowledge and coordinated by optical radar described in steps S01 to S10 above, an automated monitoring device is designed, which includes an optical radar time-series data processing module, a three-dimensional data structure establishment module, a crop growth cycle determination module, a key discrimination phenological window determination module, a time-series feature extraction module, a crop canopy optical response index construction module, a crop tall canopy body scattering enhancement index construction module, a crop canopy scattering density index construction module, a hierarchical discrimination framework establishment module, and an automatic identification module.

[0071] The following is a detailed implementation process of the present invention.

[0072] Step S01: Construct a time-series dataset of radar and optical images;

[0073] Sentinel-1 SAR and Sentinel-2 MSI image data that meet the research criteria were selected on the remote sensing cloud platform according to the research time range and research area. Edge masking, denoising, speckle filtering suppression, and radiation topography correction were applied to the Sentinel-1 SAR image data to construct a radar time-series dataset; cloud masking, temporal smoothing reconstruction, and temporal synthesis were applied to the Sentinel-2 MSI image data to construct an optical time-series dataset.

[0074] Step S02: Establish a spatiotemporal-feature three-dimensional data structure for optical radar fusion;

[0075] Based on the optical radar time-series dataset constructed in step S01, the normalized vegetation index, the modified normalized differential bare soil index, and the VV and VH backscattering coefficient features of the radar image are extracted. The extracted optical index features and radar backscattering features (i.e., the normalized vegetation index, the modified normalized differential bare soil index, and the VV and VH backscattering coefficient features of the radar image) are registered and aligned according to a unified time series to establish a spatiotemporal-feature three-dimensional data structure for characterizing the temporal changes of pixels. The first dimension is the spatial dimension, corresponding to the image pixel position; the second dimension is the temporal dimension, corresponding to the time of acquisition of the time-series image; and the third dimension is the feature dimension, including the optical index features and the radar backscattering features.

[0076] Step S03: Determine the crop growth cycle based on the temporal changes in vegetation and bare soil;

[0077] Based on Sentinel-2 MSI multi-temporal time-series remote sensing imagery, a Growth Period Crop Coverage Index (GPCCI) is constructed by comprehensively utilizing complementary information from the Normalized Difference Vegetation Index (NDVI) and the Modified Normalized Difference Bare Soil Index (MNDBI). The time-series variation curve of the GPCCI is used to identify the start time, end time, and duration of crop cover coverage, thereby determining the crop growth cycle. Adaptive adjustments are then made to the key growth stages of the crop in conjunction with the NDVI. The specific calculation formula is as follows:

[0078]

[0079] Wherein, NDVI is the Normalized Difference Vegetation Index, and MNDBI is the Modified Normalized Difference Bare Soil Index. This is a compensation factor.

[0080] Step S04: Determine the key phenological window by combining agronomic knowledge;

[0081] Based on the crop growth cycle determined in step S03, and combined with the phenological characteristics and canopy structure differences of rapeseed and sunflower, the temporal response differences of rapeseed, sunflower and other crops in the target area are compared and analyzed. The time period with the most significant difference between the target crop and non-target crops is determined as the key phenological window to reduce interference from other crops and improve the identification accuracy.

[0082] Step S05: Extract crop time-series features within the key phenological window;

[0083] Different crops exhibit significant differences under different wavelengths and indices. Based on the key phenological discriminant window determined in step S04, optical temporal characteristics of crop growth stages were constructed using the red-edge band and Green Chlorophyll Vegetation Index (GCVI) from Sentinel-2 multispectral data. The red-edge band primarily reflects chlorophyll content, leaf area index (LAI), or photosynthetic activity. Compared to other crops, sunflowers and rapeseed show significantly higher values ​​in the red-edge band of Sentinel-2, mainly due to a significant increase in the proportion of floral organs in the canopy and a decrease in relative chlorophyll content. The GCVI (Green Chlorophyll Vegetation Index) is mainly used to characterize the relative chlorophyll content in the vegetation canopy; its value is usually closely related to the chlorophyll concentration, leaf area index, and growth vigor of crop leaves. Sunflowers, rapeseed, and other crops show differences in leaf structure or growth process, resulting in relatively lower chlorophyll accumulation levels. Secondly, based on the temporal variation of the Sentinel-1 radar backscattering coefficient, different crops exhibit significant differences under VV and VH polarization. Therefore, based on the VV and VH time-series curves of rapeseed and sunflower, which differ from other crops, radar time-series characteristics of their growth period were constructed. Microwave signals are sensitive to crop canopy structure, leaf water content, and soil surface structure and moisture. Their backscattering intensity depends on leaf size, shape, spatial orientation, and water content, and is also affected by canopy height, volume, and structural complexity. For VV polarization, it is highly sensitive to vegetation canopy structure, vegetation water content, and soil moisture. Due to its high canopy height, tilted leaves, and abundant water, sunflower's VV value is significantly higher than other vertically structured or sparsely canopied crops such as corn and wheat. In contrast, peppers, potatoes, and rapeseed have higher VH values ​​than sunflowers, while soybeans, corn, and wheat have lower VH values. This indicates that VH polarization can better capture the canopy complexity and volume scattering characteristics of different crops, complementing VV polarization. Finally, based on optical and radar time-series datasets, optical and radar time-series features of rapeseed and sunflower growth periods are constructed. The specific calculation formula is as follows:

[0084]

[0085] in, , These are the green light band reflectance and near-infrared band reflectance of Sentinel-2, respectively.

[0086] Step S06: Construct the optical response index of the crop canopy layer;

[0087] As learned from step S05, rapeseed and sunflower exhibit high values ​​in the red-edge band and GCVI. A Corolla Layer Optical Response Index (CLORI) is designed for key growth stages, using GCVI as the numerator and the red-edge band as the denominator. This index integrates the red-edge response of the crop canopy with chlorophyll concentration information, highlighting the canopy layer characteristic of rapeseed and sunflower. After sunflowers and rapeseed enter their flowering or reproductive growth stages, the proportion of floral organs in the canopy increases significantly, the relative chlorophyll content decreases, and the red-edge band reflection strengthens, resulting in a significant decrease in the CLORI value. This explains the significantly lower CLORI values ​​observed in rapeseed and sunflower during this period. The specific calculation formula is as follows:

[0088]

[0089] in, The red-edge reflectivity of Sentinel-2.

[0090] Step S07: Construct a crop tall canopy scattering enhancement index;

[0091] During the critical growth periods of rapeseed and sunflower, there exist dwarf flowering crops that meet the optical response index of the crop canopy. From the radar time-series characteristics obtained in step S05, there are significant differences in the radar backscattering characteristics between tall and dwarf flowering crops. Tall flowering crops have higher plant height, more developed canopy structure, and greater canopy thickness. Their backscattering coefficients under both VV and VH polarization modes are at a moderate to low level, with the VH polarization coefficient significantly lower than that under VV polarization. They exhibit a high proportion of volume scattering and stable temporal changes in backscattering. Dwarf flowering crops have shorter plant height and shallower canopies, making it difficult for electromagnetic waves to form effective volume scattering. They primarily rely on surface scattering and are greatly affected by soil scattering. Their backscattering coefficients are generally higher and exhibit large temporal fluctuations, with little difference in the dual polarization coefficients and an extremely low proportion of volume scattering. Based on the above differences, the Tall Canopy Volume Scattering Enhancement Index (TCVSEI) can amplify the volume scattering characteristics of tall flowering crops and suppress the interference of dwarf crops, thereby accurately distinguishing and screening tall flowering crops, laying the foundation for subsequent identification of rapeseed and sunflower. The specific calculation formula is as follows:

[0092]

[0093] Wherein, VV and VH represent the backscattering coefficients of radar VV and VH polarization, respectively. By performing square normalization on VV, the common scattering characteristics caused by changes in biomass and background structure can be effectively weakened, while the differential response of VH in different crop growth stages can be significantly amplified.

[0094] Step S08: Construct the crop canopy scattering density index;

[0095] Rapeseed and sunflower are both tall flowering crops that primarily exhibit volume scattering, but their canopy structures differ significantly. Rapeseed has a relatively compact canopy structure and moderate foliage density, resulting in a milder scattering response and a gradual temporal change in the backscattering VV / VH coefficient. Sunflower, on the other hand, has a taller and more expansive canopy with relatively sparse foliage but a larger canopy volume per plant, leading to a more pronounced temporal change in the backscattering coefficient. Based on these different canopy response characteristics, a crop canopy scattering density index (CSDI) was constructed, with rapeseed exhibiting a lower canopy scattering density and sunflower a higher canopy scattering density. The specific calculation formula is as follows:

[0096]

[0097] Where VV and VH represent the backscattering coefficients of radar VV and VH polarizations, respectively.

[0098] Step S09: Establish a rapeseed and sunflower grading framework driven by agronomic knowledge;

[0099] Based on agronomical knowledge, a multi-level classification and discrimination rule is established by integrating the crop canopy optical response index constructed in step S06, the crop tall canopy scattering enhancement index constructed in step S07, and the crop canopy scattering density index constructed in step S08. The classification and discrimination rule includes the exclusion of non-flowering crops, the exclusion of dwarf flowering crops, and the distinction between rapeseed and sunflower, so as to form a classification and discrimination framework for rapeseed and sunflower.

[0100] Step S10: Automatic identification of rapeseed and sunflower based on the hierarchical discrimination framework;

[0101] The agronomic knowledge-driven discrimination framework for rapeseed and sunflower constructed in step S09 is used to determine the optimal thresholds for each feature index in conjunction with field crop point samples. Based on the field sample data, the canopy optical response index (CLORI), tall canopy volume scattering enhancement index (TCVSEI), and canopy scattering density index (CSDI) corresponding to the samples in the target area are extracted. The distribution characteristics of each index are analyzed to determine the optimal thresholds and construct the discrimination function for rapeseed and sunflower. The specific calculation formula is as follows:

[0102]

[0103] in, Let be the discriminant function for rapeseed and sunflower, with output values ​​1, 2, 3, and 4 corresponding to non-flowering crops, dwarf flowering crops, rapeseed, and sunflower, respectively; and a threshold value. , , It is determined based on the distribution characteristics of the field sample data. , , The suggested values ​​are 13.5±0.5, 0.1±0.5, and 0.35±0.5, respectively.

[0104] like Figure 3 As shown, the present invention also provides an automatic identification device for rapeseed and sunflower, which is used to perform the monitoring method described in Embodiment 1 above, including:

[0105] The optical radar time-series data processing module is used to acquire and preprocess radar and optical remote sensing time-series data;

[0106] The 3D data structure establishment module is used to perform time alignment of optical features and radar features, and to establish a spatiotemporal-feature 3D data structure for optical-radar fusion.

[0107] The crop growth cycle determination module is used to determine the crop growth cycle based on the temporal relationship between vegetation and bare soil.

[0108] The critical phenological window determination module is used to determine the critical phenological window by combining agronomic knowledge;

[0109] The temporal feature extraction module is used to extract the temporal features of crop optical radar within the key phenological discrimination window;

[0110] The crop canopy optical response index construction module is used to construct the crop canopy optical response index.

[0111] A module for constructing crop tall canopy scattering enhancement indices is used to construct crop tall canopy scattering enhancement indices.

[0112] The crop canopy scattering density index construction module is used to construct the crop canopy scattering density index.

[0113] The grading and discrimination framework establishment module is used to establish a grading and discrimination framework for rapeseed and sunflower based on the crop canopy optical response index, crop tall canopy scattering enhancement index and crop canopy scattering density index.

[0114] An automatic identification module is used to automatically identify rapeseed and sunflower based on the hierarchical discrimination framework.

[0115] The above modules can be configured as computer program instructions stored in memory, and executed by a processor to implement the above functions. This device can be applied to various electronic devices.

[0116] To verify the effectiveness of this invention, a city in a certain province was selected as the study area, and a national standard administrative division vector map was used as the base map. Following the above method steps, a spatial distribution map of rapeseed and sunflower in the study area was created using the monitoring device provided by this invention (e.g., Figure 4 (As shown).

[0117] The above are preferred embodiments of the present invention. Any changes made to the technical solution of the present invention that do not exceed the scope of the technical solution of the present invention shall fall within the protection scope of the present invention.

Claims

1. An automatic identification method for rapeseed and sunflower based on agronomic knowledge and optical radar, characterized in that: Includes the following steps: Step S01: Construct a time-series dataset of radar and optical images; Step S02: Establish a spatiotemporal-feature three-dimensional data structure for optical radar fusion; Step S03: Determine the crop growth cycle based on the temporal changes in vegetation and bare soil; Step S04: Determine the key phenological window by combining agronomic knowledge; Step S05: Extract crop time-series features within the key phenological window; Step S06: Construct the optical response index of the crop canopy layer; Step S07: Construct a crop tall canopy scattering enhancement index; Step S08: Construct the crop canopy scattering density index; Step S09: Establish a rapeseed and sunflower grading framework driven by agronomic knowledge; Step S10: Automatic identification of rapeseed and sunflower based on the hierarchical discrimination framework.

2. The automatic identification method for rapeseed and sunflower based on agronomic knowledge and optical radar collaboration as described in claim 1, characterized in that, In step S02, based on the radar and optical image time-series dataset constructed in step S01, the normalized vegetation index, the corrected normalized differential bare soil index, and the VV and VH backscattering coefficient features of the radar images are extracted. The extracted optical index features and radar backscattering features are registered and aligned according to a unified time series to establish a spatiotemporal-feature three-dimensional data structure for characterizing pixel temporal changes. The first dimension is the spatial dimension, corresponding to the image pixel position; the second dimension is the temporal dimension, corresponding to the time of acquisition of the time-series images; and the third dimension is the feature dimension, including optical index features and radar backscattering features.

3. The automatic identification method for rapeseed and sunflower based on agronomic knowledge and optical radar collaboration as described in claim 1, characterized in that, In step S03, based on the normalized vegetation index and the modified normalized difference bare soil index extracted in step S02, a crop cover index (GPCCI) for growth stages is constructed. The crop cover start time, crop cover end time, and cover duration are identified according to the temporal variation curve of the GPCCI to determine the crop growth cycle. The crop growth cycle is then adaptively adjusted using the normalized vegetation index to determine the key growth stages of the crop. In step S04, based on the crop growth cycle determined in step S03, the temporal response differences between rapeseed, sunflower, and other crops within the target area are compared and analyzed using the phenological characteristics and canopy structure differences between rapeseed and sunflower. The time period with the most significant difference between the target crop and non-target crops is identified as the key phenological discrimination window. Wherein, NDVI is the Normalized Difference Vegetation Index, and MNDBI is the Modified Normalized Difference Bare Soil Index. This is a compensation factor.

4. The automatic identification method for rapeseed and sunflower based on agronomic knowledge and optical radar collaboration as described in claim 1, characterized in that, In step S05, based on the key phenological window determined in step S04, optical and radar time-series features of the crop are extracted. The optical time-series features include at least the red-edge band features and the green chlorophyll vegetation index (GCVI) features in the Sentinel-2 multispectral data, and the radar time-series features include at least the VV and VH backscattering coefficient features of the Sentinel-1 image, so as to form the optical and radar time-series features of the target crop within the key phenological window. in, , These are the green light band reflectance and near-infrared band reflectance of Sentinel-2, respectively.

5. The automatic identification method for rapeseed and sunflower based on agronomic knowledge and optical radar collaboration according to claim 4, characterized in that, In step S06, based on the optical time-series features extracted in step S05, the red-edge band response information and chlorophyll concentration response information are fused to construct the crop corolla optical response index CLORI, which is used to characterize the difference in corolla optical response between rapeseed and sunflower within the key phenological window, thereby enabling the distinction between flowering and non-flowering crops. in, The red-edge reflectivity of Sentinel-2.

6. The automatic identification method for rapeseed and sunflower based on agronomic knowledge and optical radar collaboration according to claim 4, characterized in that, In step S07, based on the radar time-series features extracted in step S05, the difference between the backscattering coefficients of VV and VH is used to construct the crop tall canopy volume scattering enhancement index TCVSEI, which is used to enhance the volume scattering characteristics of tall canopy crops and suppress the scattering interference of dwarf flowering crops, thereby realizing the distinction between tall flowering crops and dwarf flowering crops. Where VV and VH represent the backscattering coefficients of radar VV and VH polarizations, respectively.

7. The automatic identification method for rapeseed and sunflower based on agronomic knowledge and optical radar collaboration as described in claim 1, characterized in that, In step S08, based on the radar time-series features extracted in step S05, the VV and VH backscattering coefficients are fused to construct the crop canopy scattering density index CSDI, which is used to characterize the difference in canopy scattering density between rapeseed and sunflower, thereby enabling the distinction between rapeseed and sunflower. Where VV and VH represent the backscattering coefficients of radar VV and VH polarizations, respectively.

8. The automatic identification method for rapeseed and sunflower based on agronomic knowledge and optical radar collaboration according to claim 1, characterized in that, In step S09, based on agronomic knowledge, the crop canopy optical response index CLORI constructed in step S06, the crop tall canopy body scattering enhancement index TCVSEI constructed in step S07, and the crop canopy scattering density index CSDI constructed in step S08 are integrated to establish a multi-level classification discrimination rule. The multi-level classification discrimination rule includes the exclusion of non-flowering crops, the exclusion of dwarf flowering crops, and the distinction between rapeseed and sunflower, so as to form a classification discrimination framework for rapeseed and sunflower.

9. The automatic identification method for rapeseed and sunflower based on agronomic knowledge and optical radar collaboration according to claim 1, characterized in that, In step S10, the discrimination thresholds for the crop canopy optical response index CLORI, the crop tall canopy body scattering enhancement index TCVSEI, and the crop canopy scattering density index CSDI are determined based on field sample data. Based on these discrimination thresholds, a discrimination function for rapeseed and sunflower is constructed to achieve automatic identification of rapeseed and sunflower. The output categories of the discrimination function include at least non-flowering crops, dwarf flowering crops, rapeseed, and sunflower. The output categories of the discrimination functions for rapeseed and sunflower include at least non-flowering crops, dwarf flowering crops, rapeseed, and sunflower, and the discrimination threshold is determined based on the distribution characteristics of field sample data. , , It is a fixed value.

10. An automatic identification device for rapeseed and sunflower based on agronomic knowledge and optical radar, characterized in that, The method for automatic identification of rapeseed and sunflower based on agronomic knowledge and optical radar collaboration as described in any one of claims 1 to 9 includes: The optical radar time-series data processing module is used to acquire and preprocess radar and optical remote sensing time-series data; The 3D data structure establishment module is used to perform time alignment of optical features and radar features, and to establish a spatiotemporal-feature 3D data structure for optical-radar fusion. The crop growth cycle determination module is used to determine the crop growth cycle based on the temporal relationship between vegetation and bare soil. The critical phenological window determination module is used to determine the critical phenological window by combining agronomic knowledge; The temporal feature extraction module is used to extract the temporal features of crop optical radar within the key phenological discrimination window; The crop canopy optical response index construction module is used to construct the crop canopy optical response index. A module for constructing crop tall canopy scattering enhancement indices is used to construct crop tall canopy scattering enhancement indices. The crop canopy scattering density index construction module is used to construct the crop canopy scattering density index. The grading and discrimination framework establishment module is used to establish a grading and discrimination framework for rapeseed and sunflower based on the crop canopy optical response index, crop tall canopy scattering enhancement index and crop canopy scattering density index. An automatic identification module is used to automatically identify rapeseed and sunflower based on the hierarchical discrimination framework.

Images