Intelligent extraction method, device and electronic equipment for crop leaf area index

Abstract

The present invention provides a method, device and electronic device for intelligently extracting crop leaf area index, which includes the following steps: acquiring farmland remote sensing images and collecting time-series remote sensing data; determining the plot type of each plot; constructing a spatial distribution map of the plot; identifying target crops and constructing a spatial distribution map of the target crops; using satellite data to invert the first object leaf area index; using unmanned aerial vehicles to collect crop spectra, calculate the second crop leaf area index, and establish a first model training label library; establishing a first mapping model based on a first mapping relationship between the first crop leaf area index and the second crop leaf area index, and generating a converted third crop leaf area index. The present invention realizes the intelligent conversion of crop leaf area index inverted from satellite data to crop leaf area index calculated by near-ground measurement at the crop level, realizes the correction of crop leaf area index inverted from satellite data, and realizes large-scale, high-precision, fine-scale crop leaf area index inversion.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural remote sensing information extraction, and in particular to a method, device and electronic equipment for intelligently extracting crop leaf area index. Background Art

[0002] Precision agriculture requires changing the extensive management methods in the past agricultural production process, improving agricultural production efficiency and reducing the impact of disasters. Accurate crop spatial distribution and growth parameter information can reflect the resource utilization in the agricultural production process, and is an important data support for crop yield prediction and agricultural planting structure adjustment. Among them, the leaf area index (LAI) is an important physiological parameter of crops, which is closely related to evapotranspiration / photosynthesis and biomass accumulation. It can reflect the temporal and spatial heterogeneity of crops and is an important indicator for crop growth monitoring and yield estimation.

[0003] In agricultural remote sensing applications, affected by factors such as atmospheric effects, crop parameters inverted by satellite remote sensing often only reflect the relative differences between different crops; near-ground agricultural monitoring methods represented by drones have high accuracy, but have problems such as small monitoring areas and poor timeliness, and cannot obtain large-scale, high-precision, fine-scale crop growth parameter information. Due to the long distance between satellites and crops and the wide coverage of satellite data, there are problems such as insufficient inversion accuracy and a small near-ground inversion range, and it is impossible to simultaneously achieve large-scale, high-precision, fine-scale crop leaf area index inversion. Summary of the invention

[0004] In order to solve the problem, the purpose of the present invention is to provide a method, device and electronic equipment for intelligent extraction of crop leaf area index.

[0005] In a first aspect, the present disclosure provides an intelligent method for extracting crop leaf area index, comprising the following steps:

[0006] Acquire farmland remote sensing images and collect time series remote sensing data;

[0007] Determine the land type and land space form of each land parcel according to the farmland remote sensing image;

[0008] Constructing a spatial distribution map of the plots according to the plot types and the plot spatial forms;

[0009] Identify the target crop type in the spatial distribution map of the plot, and construct a spatial distribution map of the target crop;

[0010] Inverting the leaf area index of the first crop using satellite data according to the spatial distribution map of the target crop;

[0011] Use drones to collect crop spectra of different crops near the ground and calculate the leaf area index of the second crop;

[0012] Establishing a first training label library using the first crop leaf area index and the second crop leaf area index;

[0013] Determine a first mapping relationship between the first crop leaf area index and the second crop leaf area index, and establish a first mapping relationship model according to the first mapping relationship;

[0014] Using the first training label library to train the first mapping relationship model to obtain a first crop leaf area index conversion model;

[0015] The first crop leaf area index inverted from the satellite data is input into the first crop leaf area index conversion model to generate a converted third crop leaf area index.

[0016] In the second aspect, the present disclosure provides a first acquisition unit, a first determination unit, a first construction unit, a second construction unit, a first calculation processing unit, a second calculation processing unit, a first training label library establishment unit, a first mapping relationship model establishment unit, a first model training unit and a first generation unit:

[0017] The first acquisition unit is used to acquire farmland remote sensing images and collect time series remote sensing data;

[0018] The first determination unit is used to determine the land type of each land parcel according to the farmland remote sensing image;

[0019] The first construction unit is used to construct a spatial distribution map of the land parcels according to the land parcel types and land parcel spatial forms;

[0020] The second construction unit is used to identify the target crop type in the spatial distribution map of the plot and construct the spatial distribution map of the target crop;

[0021] The first calculation processing unit is used to invert the leaf area index of the first crop using satellite data according to the spatial distribution map of the target crop;

[0022] The second calculation processing unit is used to collect crop spectra of different crops near the ground using a drone to calculate a second crop leaf area index;

[0023] The first training label library establishing unit is used to establish a first training label library using the first crop leaf area index and the second crop leaf area index;

[0024] The first mapping relationship model establishing unit is used to determine a first mapping relationship between the first crop leaf area index and the second crop leaf area index, and establish a first mapping relationship model according to the first mapping relationship;

[0025] The first model training unit is used to train the first mapping relationship model using the first training label library to obtain a first crop leaf area index conversion model;

[0026] The first generating unit is used to input the first crop leaf area index into the first crop leaf area index conversion model to generate a converted third crop leaf area index.

[0027] In a third aspect, the present disclosure provides an electronic device, including:

[0028] Processor and memory;

[0029] The memory is used to store computer operation instructions;

[0030] The processor is used to execute the intelligent extraction method of crop leaf area index by calling the computer operation instruction.

[0031] The beneficial effects of the present invention are as follows: the present invention constructs a spatial distribution map of plots according to plot types, constructs a spatial distribution map of target crops according to the target crop types in the spatial distribution map of plots, establishes a mapping relationship between crop leaf area index inverted by using satellite data and crop spectrum calculated by using unmanned aerial vehicles near the ground according to different crops, establishes a mapping relationship model and trains the mapping relationship model, realizes intelligent conversion of crop leaf area index inverted by satellite data to crop leaf area index calculated by near-ground measurement, realizes correction of crop leaf area index inverted by satellite data, overcomes the problem of insufficient inversion accuracy due to large satellite data range and small near-ground inversion range, realizes correction of crop leaf area index inverted by satellite data at the crop level, and realizes large-scale and higher-precision fine-scale crop leaf area index inversion.

[0032] Based on the above technical solution, the present invention can also be improved as follows.

[0033] Furthermore, the crop leaf area index intelligent extraction method also includes:

[0034] collecting crop spectra of each plot type on the ground, and calculating a fourth crop leaf area index;

[0035] Establishing a second training label library using the third crop leaf area index and the fourth crop leaf area index;

[0036] Determine a second mapping relationship between the third crop leaf area index and the fourth crop leaf area index, and establish a second mapping relationship model according to the second mapping relationship;

[0037] Using the second training label library to train the second mapping relationship model to obtain a second crop leaf area index conversion model;

[0038] The third crop leaf area index is input into the transformation model using the second crop leaf area index to generate a transformed fifth crop leaf area index.

[0039] The beneficial effect of adopting the above further scheme is that, by establishing a second mapping relationship model according to the second mapping relationship between the third crop leaf area index and the fourth crop leaf area index, the intelligent conversion of the crop leaf area index inverted from satellite data to the crop leaf area index calculated by near-ground measurement is realized, and the intelligent conversion of the crop leaf area index calculated by near-ground measurement to the crop leaf area index calculated by ground measurement is realized, thereby further improving the inversion accuracy of the crop leaf area index and realizing the correction of the crop leaf area index inverted from satellite data.

[0040] Further, determining the land type of each land plot according to the farmland remote sensing image includes:

[0041] Obtain farmland remote sensing image database and historical farmland remote sensing images in the database;

[0042] Extracting the spatial features of the historical farmland remote sensing image, and determining the land plot type corresponding to the spatial features of the historical farmland remote sensing image;

[0043] Constructing a spatial convolutional neural network model according to the spatial characteristics of the historical farmland remote sensing images and the land plot type;

[0044] Training the spatial convolutional neural network model to obtain a land parcel type extraction model;

[0045] The spatial features of the current farmland remote sensing image of each plot are extracted as input to the plot type extraction model to obtain the plot type of each plot.

[0046] The beneficial effect of adopting the above further scheme is that, based on the spatial characteristics of historical farmland remote sensing images, by constructing a spatial convolutional neural network model, the plot type of each plot can be obtained quickly and accurately.

[0047] Furthermore, the spatial features of the farmland remote sensing image include edge features and texture features; the land plot types include cultivated land, regular terraces and sloping cultivated land.

[0048] The beneficial effect of adopting the above further scheme is that, according to the edge features and texture features of the farmland remote sensing image, it is helpful to accurately obtain the type of plot and construct a spatial distribution map of the plot.

[0049] Furthermore, constructing a spatial distribution map of the land parcels according to the land parcel types and land parcel spatial forms includes:

[0050] Constructing a first land parcel spatial distribution map according to the land parcel type and the land parcel spatial morphology;

[0051] Performing a patch fusion process on the first land parcel spatial distribution map to obtain a second land parcel spatial distribution map;

[0052] The second land parcel spatial distribution map is smoothed, the second land parcel spatial distribution map is reconstructed, and the third land parcel spatial distribution map is determined.

[0053] The beneficial effect of adopting the above further scheme is that by performing patch fusion processing on the spatial distribution map of the first plot of land, it is helpful to improve the quality of the remote sensing image of the farmland; by performing smoothing processing on the spatial distribution map of the second farmland data, it is helpful to connect the discontinuous parts in the image and eliminate random noise in the image.

[0054] Further, the identifying the target crop type in the spatial distribution map of the plot and constructing the spatial distribution map of the target crop includes:

[0055] Acquire the time series characteristics of crops in each plot in the plot spatial distribution map;

[0056] Collecting crop type data of each plot in the plot spatial distribution map on the ground, and using the crop type data collected on the ground as samples to construct a crop type discrimination neural network model;

[0057] Using satellite data to obtain and calculate the temporal characteristics of crops in each plot;

[0058] Inputting the time series features into the crop type discrimination neural network model to determine the crop type of each plot;

[0059] According to the target crop types of all plots, the spatial distribution map of the target crop is determined.

[0060] The beneficial effect of adopting the above further scheme is that by constructing a crop type discrimination neural network model, the crop type of each plot is determined by using the temporal characteristics of the crop, and a subsequent crop leaf area index acquisition model is constructed at the level of each type of crop, thereby avoiding the disadvantage of limited accuracy of constructing a unified leaf area index inversion model, and facilitating the realization of fine-scale crop leaf area index inversion.

[0061] Furthermore, the temporal characteristics of the crop include the normalized difference vegetation index, enhanced vegetation index, land surface water index and greenness normalized difference vegetation index at the plot scale.

[0062] The beneficial effect of adopting the above further solution is that the crop type of each plot can be accurately determined based on the temporal characteristics of the crops. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is a flow chart of the intelligent extraction method of crop leaf area index in Example 1 of the present invention;

[0064] Figure 2 This is a schematic diagram of an intelligent extraction device for crop leaf area index in Example 2 of the present invention;

[0065] Figure 3 This is a schematic diagram of an electronic device in Example 3 of the present invention.

[0066] Icons: 30 - electronic device; 310 - processor; 320 - bus; 330 - memory; 340 - transceiver. DETAILED DESCRIPTION

[0067] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0068] Example 1

[0069] As an example, Figure 1 As shown, in order to solve the above technical problems, this embodiment provides an intelligent extraction method of crop leaf area index, comprising the following steps:

[0070] Acquire farmland remote sensing images and collect time series remote sensing data;

[0071] Determine the land type of each plot based on farmland remote sensing images;

[0072] According to the type of plots, construct a spatial distribution map of plots;

[0073] Identify the target crop types in the spatial distribution map of the plots and construct the spatial distribution map of the target crops;

[0074] According to the spatial distribution map of target crops, the leaf area index of the first crop is inverted using satellite data;

[0075] The crop spectra of different crops are collected by using a drone near the ground to calculate the leaf area index of the second crop; the leaf area index of the first crop and the leaf area index of the second crop are used to establish a first training label library;

[0076] Determine a first mapping relationship between a first crop leaf area index and a second crop leaf area index, and establish a first mapping relationship model according to the first mapping relationship;

[0077] Using the first training label library to train the first mapping relationship model, to obtain a first crop leaf area index conversion model;

[0078] The first crop leaf area index is input into a transformation model using the first crop leaf area index to generate a transformed third crop leaf area index.

[0079] The beneficial effects of the present invention are as follows: the present invention constructs a spatial distribution map of land plots according to land plot types, constructs a spatial distribution map of target crops according to target crop types in the land plot spatial distribution map, and according to different crop types, establishes a mapping relationship between crop leaf area index inverted by using satellite data and crop spectrum calculated by using unmanned aerial vehicles near the ground, establishes a mapping relationship model and trains the mapping relationship model, realizes intelligent conversion of crop leaf area index inverted by satellite data to crop leaf area index calculated by near-ground measurement, realizes correction of crop leaf area index inverted by satellite data, overcomes the problems of insufficient inversion accuracy due to large satellite data range and small near-ground inversion range, and realizes large-scale, high-precision and fine-scale crop leaf area index inversion.

[0080] Optionally, the farmland remote sensing image is a sub-meter satellite remote sensing image. The selected satellite data can be medium-resolution data such as Sentinel-2 and Gaofen-1.

[0081] Optionally, the intelligent extraction method of crop leaf area index also includes:

[0082] Collect crop spectra of plots of each plot type on the ground and calculate the fourth crop leaf area index;

[0083] Using the third crop leaf area index and the fourth crop leaf area index to establish a second training label library;

[0084] Determining a second mapping relationship between the third crop leaf area index and the fourth crop leaf area index, and establishing a second mapping relationship model according to the second mapping relationship;

[0085] The second mapping relationship model is trained using the second training label library to obtain a second crop leaf area index conversion model;

[0086] The third crop leaf area index is input into a transformation model using the second crop leaf area index to generate a transformed fifth crop leaf area index.

[0087] In the actual application process, optionally, the first mapping relationship model and the second mapping relationship model are both neural network models based on the attention mechanism. The use of drones to collect crop spectra of plots of various types of plots near the ground is not limited to the use of drones and the installation of instruments, and the instruments for collecting crop spectra of plots of various types of plots on the ground are not limited to LAI2000 or ASD. The first crop leaf area index is inverted using satellite data, the second crop leaf area index is calculated using the method of collecting near the ground using drones, and the second crop leaf area index is calculated using the method of collecting on the ground, so as to realize the conversion of satellite data to near-ground data and the conversion of near-ground data to ground data. According to different crop types, two conversions are carried out type by type to realize the correction of crop leaf area index inverted from satellite data, overcome the problems of insufficient inversion accuracy of satellite data in a large range, inversion of near-ground crop leaf area index and small inversion range of ground crop leaf area index, realize the correction of crop leaf area index inverted from satellite data at the crop level, and realize large-scale, high-precision and fine-scale crop leaf area index inversion.

[0088] Optionally, determine the land type of each plot based on the farmland remote sensing image, including:

[0089] Obtain farmland remote sensing image database and historical farmland remote sensing images in the database;

[0090] Extracting the spatial features of the historical farmland remote sensing image, and determining the land plot type corresponding to the spatial features of the historical farmland remote sensing image;

[0091] Constructing a spatial convolutional neural network model according to the spatial characteristics of the historical farmland remote sensing images and the land plot type;

[0092] Training the spatial convolutional neural network model to obtain a land parcel type extraction model;

[0093] The spatial features of the current farmland remote sensing image of each plot are extracted as input to the plot type extraction model to obtain the plot type of each plot.

[0094] In actual application, farmland remote sensing images first rely on naturally existing or artificially constructed linear features such as terrain, road networks and water networks; secondly, the plots corresponding to regular cultivated land and terraces are extracted based on edge features, and the plots corresponding to sloping cultivated land are extracted using texture models.

[0095] The land parcel type extraction model can be described as follows:

[0096]

[0097] Among them, X i，jIndicates the sample number, H l-1 represents the pooling layer, b l represents the regularization term, W l Represents the weight information after each training.

[0098] Optionally, the spatial features of the farmland remote sensing image include edge features and texture features; the land plot types include cultivated land, regular terraces and sloping cultivated land.

[0099] In actual application, regular cultivated land refers to farmland located in a flat area with a slope of less than 3 degrees and with obvious long and short sides; terraced fields refer to farmland located in a hillside area with a slope of 3-10 degrees and with long sides; sloping cultivated land refers to farmland that may exist in hillside areas or mountainous areas with a slope of more than 3 degrees and has no obvious boundaries.

[0100] Optionally, a spatial distribution map of the plots is constructed based on the plot type, including:

[0101] According to the type of the plot, construct a first plot spatial distribution map;

[0102] Performing a patch fusion process on the spatial distribution map of the first plot to obtain a spatial distribution map of the second plot;

[0103] The spatial distribution map of the second plot is smoothed, the spatial distribution map of the second plot is reconstructed, and the spatial distribution map of the third plot is determined.

[0104] In actual application, optionally, after performing patch fusion processing on the spatial distribution map of the first plot to obtain the spatial distribution map of the second plot, a value of an integer multiple (such as four times) of the spatial resolution of the remote sensing image is used to perform global smoothing on the second plot spatial distribution map, and the farmland distribution map of the entire area is reconstructed to obtain the spatial distribution map of the third plot.

[0105] Optionally, identifying the crop types in the plot spatial distribution map and constructing the crop spatial distribution map includes:

[0106] Obtain the temporal characteristics of crops in each plot in the plot spatial distribution map;

[0107] The crop type data of each plot in the spatial distribution map of the plots are collected on the ground, and the crop type data collected on the ground are used as samples to build a crop type discrimination neural network model;

[0108] Using satellite data to obtain and calculate the temporal characteristics of crops in each plot;

[0109] Input the time series features into the crop type discrimination neural network model to determine the crop type of each plot;

[0110] Determine the spatial distribution map of target crops based on the target crop types of all plots.

[0111] In the actual application process, the crop type discrimination neural network model is constructed based on the multi-layer LSTM unit. The time series characteristics of the crop are input into the time series neural network model to realize the recognition of crop types and the division of plots according to crop types to obtain the crop spatial distribution map of all areas corresponding to the spatial distribution map. By constructing a crop type discrimination neural network model, the crop type of each plot is determined using the crop's time series characteristics, and the subsequent crop leaf area index acquisition model is constructed at the level of each type of crop, avoiding the disadvantage of limited accuracy of constructing a unified leaf area index inversion model, which is conducive to the realization of fine-scale crop leaf area index inversion.

[0112] Optionally, the temporal characteristics of crops include, but are not limited to, the normalized difference vegetation index (NDVI), the enhanced vegetation index (EVI), the land surface water index (LSWI) and the greenness normalized difference vegetation index (GNDVI) at the plot scale.

[0113] Optionally, each time series feature is calculated as follows:

[0114]

[0115]

[0116]

[0117]

[0118] Among them, B is the reflectivity of the blue light band, G is the reflectivity of the green light band, R is the reflectivity of the red light band, NIR is the reflectivity of the near infrared band, and SWIRI is the reflectivity of the short-wave infrared band.

[0119] Example 2

[0120] Based on the same principle as the method shown in Example 1 of the present invention, as shown in the attached Figure 2 As shown, an embodiment of the present invention also provides an intelligent extraction device for leaf area index, including a first acquisition unit, a first determination unit, a first construction unit, a second construction unit, a first calculation processing unit, a second calculation processing unit, a first training label library establishment unit, a first mapping relationship model establishment unit, a first model training unit and a first generation unit:

[0121] The first acquisition unit is used to acquire farmland remote sensing images and collect time series remote sensing data;

[0122] The first determination unit is used to determine the land type of each land plot according to the farmland remote sensing image;

[0123] A first construction unit is used to construct a spatial distribution map of the plots according to the plot types;

[0124] The second construction unit is used to identify the target crop type in the spatial distribution map of the plot and construct the spatial distribution map of the target crop;

[0125] A first calculation processing unit is used to invert the leaf area index of the first crop using satellite data according to the spatial distribution map of the target crop;

[0126] A second calculation processing unit is used to collect crop spectra of different crops near the ground using a drone to calculate a leaf area index of a second crop;

[0127] A first training label library establishing unit, used to establish a first training label library using the first crop leaf area index and the second crop leaf area index;

[0128] A first mapping relationship model establishing unit, used for determining a first mapping relationship between a first crop leaf area index and a second crop leaf area index, and establishing a first mapping relationship model according to the first mapping relationship;

[0129] A first model training unit is used to train the first mapping relationship model using the first training label library to obtain a first crop leaf area index conversion model;

[0130] The first generating unit is used to input the first crop leaf area index into a transformation model using the first crop leaf area index to generate a transformed third crop leaf area index.

[0131] Optionally, the intelligent extraction device for crop leaf area index further includes:

[0132] a third calculation processing unit, for collecting crop spectra of plots of various plot types on the ground, and calculating a fourth crop leaf area index;

[0133] A second training label library establishing unit, using the third crop leaf area index and the fourth crop leaf area index to establish a second training label library;

[0134] A second mapping relationship model establishing unit is used to determine a second mapping relationship between the third crop leaf area index and the fourth crop leaf area index, and establish a second mapping relationship model according to the second mapping relationship;

[0135] A second model training unit is used to train the second mapping relationship model using a second training label library to obtain a second crop leaf area index conversion model;

[0136] The second generating unit is used for inputting the third crop leaf area index into the second crop leaf area index conversion model to generate a converted fifth crop leaf area index.

[0137] Optionally, the first determining unit includes:

[0138] The second acquisition unit is used to acquire a farmland remote sensing image database and acquire historical farmland remote sensing images in the database;

[0139] The first determination subunit is used to extract the spatial features of the historical farmland remote sensing images and determine the land plot type corresponding to the spatial features of the historical farmland remote sensing images;

[0140] A model building unit is used to build a spatial convolutional neural network model based on the spatial characteristics and plot types of farmland remote sensing images;

[0141] The third model training unit is used to train the spatial convolutional neural network model to obtain a land parcel type extraction model;

[0142] The second determination subunit is used to extract the spatial features of the farmland remote sensing images of each plot as the input of the plot type extraction model to obtain the plot type of each plot.

[0143] Optionally, the first building block includes:

[0144] A first construction unit subunit is used to construct a first land parcel spatial distribution map according to the land parcel type;

[0145] A first image processing unit is used to perform a patch fusion process on the first land parcel spatial distribution map to obtain a second land parcel spatial distribution map;

[0146] The second image processing unit is used to smooth the second farmland data spatial distribution map, reconstruct the second farmland data spatial distribution map, and determine the third farmland data spatial distribution map.

[0147] Optionally, the second building block includes:

[0148] A third acquisition unit is used to acquire the time series characteristics of crops in each plot in the plot spatial distribution map;

[0149] The second construction unit subunit is used to collect crop type data of each plot in the plot spatial distribution map on the ground, and use the crop type data collected on the ground as samples to build a crop type discrimination neural network model;

[0150] A fourth acquisition unit, configured to acquire and calculate the time series characteristics of crops in each plot using satellite data;

[0151] The second determination unit is used to input the time series features into the crop type discrimination neural network model to determine the crop type of each plot;

[0152] The third determination unit is used to determine the target crop spatial distribution map according to the target crop types of all plots.

[0153] Optionally, the spatial features of the farmland remote sensing image include edge features and texture features; the land plot types include cultivated land, regular terraces and sloping cultivated land.

[0154] Optionally, the temporal characteristics of crops include plot-scale normalized difference vegetation index, enhanced vegetation index, land surface water index and greenness normalized difference vegetation index.

[0155] Example 3

[0156] Based on the same principle as the method shown in the embodiment of the present invention, an electronic device is also provided in the embodiment of the present invention, as shown in the attached Figure 3 As shown, the electronic device may include but is not limited to:

[0157] A processor and a memory; the memory is used to store computer operation instructions; the processor is used to execute the method shown in any embodiment of the present invention by calling the computer operation instructions.

[0158] In an optional embodiment, an electronic device is provided. Figure 3 The electronic device 30 shown includes: a processor 310 and a memory 330. The processor 310 and the memory 330 are connected, for example, via a bus 320.

[0159] Optionally, the electronic device 30 may further include a transceiver 340, which may be used for data interaction between the electronic device and other electronic devices, such as data transmission and / or data reception. It should be noted that in actual applications, the transceiver 340 is not limited to one, and the structure of the electronic device 30 does not constitute a limitation on the embodiments of the present invention.

[0160] The processor 310 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of the present invention. The processor 310 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0161] The bus 320 may include a path to transmit information between the above components. The bus 320 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 320 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0162] The memory 330 can be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to this.

[0163] The memory 330 is used to store application code (computer program) for executing the solution of the present invention, and is controlled to be executed by the processor 310. The processor 310 is used to execute the application code stored in the memory 330 to implement the contents shown in the above method embodiment.

[0164] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent extraction method for crop leaf area index, characterized in that: The steps include: Acquire farmland remote sensing images and collect time series remote sensing data; Determining the land type of each land parcel according to the farmland remote sensing image; Constructing a spatial distribution map of the plots according to the plot types and the plot spatial forms; Identify the target crop type in the spatial distribution map of the plot, and construct a spatial distribution map of the target crop; According to the spatial distribution map of target crops, the leaf area index of the first crop is inverted using satellite data; Use drones to collect crop spectra of different crops near the ground and calculate the leaf area index of the second crop; Establishing a first training label library using the first crop leaf area index and the second crop leaf area index; Determine a first mapping relationship between the first crop leaf area index and the second crop leaf area index, and establish a first mapping relationship model according to the first mapping relationship; Using the first training label library to train the first mapping relationship model to obtain a first crop leaf area index conversion model; The first crop leaf area index inverted from the satellite data is input into the first crop leaf area index conversion model to generate a converted third crop leaf area index.

2. The intelligent extraction method of crop leaf area index according to claim 1, characterized in that: Also includes: collecting crop spectra of each plot type on the ground, and calculating a fourth crop leaf area index; Establishing a second training label library using the third crop leaf area index and the fourth crop leaf area index; Determine a second mapping relationship between the third crop leaf area index and the fourth crop leaf area index, and establish a second mapping relationship model according to the second mapping relationship; Using the second training label library to train the second mapping relationship model to obtain a second crop leaf area index conversion model; The third crop leaf area index is input into the transformation model using the second crop leaf area index to generate a transformed fifth crop leaf area index.

3. The intelligent extraction method of crop leaf area index according to claim 1, characterized in that: Determining the land type of each land plot according to the farmland remote sensing image includes: Obtain farmland remote sensing image database and historical farmland remote sensing images in the database; Extracting the spatial features of the historical farmland remote sensing image, and determining the land plot type corresponding to the spatial features of the historical farmland remote sensing image; Constructing a spatial convolutional neural network model according to the spatial characteristics of the historical farmland remote sensing images and the land plot type; Training the spatial convolutional neural network model to obtain a land parcel type extraction model; The spatial features of the current farmland remote sensing image of each plot are extracted as input to the plot type extraction model to obtain the plot type of each plot.

4. The intelligent extraction method of crop leaf area index according to claim 3, characterized in that: The spatial features of the farmland remote sensing image include edge features and texture features; the land plot types include cultivated land, regular terraces and sloping cultivated land.

5. The intelligent extraction method of crop leaf area index according to claim 1, characterized in that: The step of constructing a spatial distribution map of the land parcels according to the land parcel types and land parcel spatial forms includes: Constructing a first land parcel spatial distribution map according to the land parcel type and the land parcel spatial morphology; Performing a patch fusion process on the first land parcel spatial distribution map to obtain a second land parcel spatial distribution map; The second land parcel spatial distribution map is smoothed, the second land parcel spatial distribution map is reconstructed, and the third land parcel spatial distribution map is determined.

6. The intelligent extraction method of crop leaf area index according to claim 1, characterized in that: The identifying the target crop type in the spatial distribution map of the plot and constructing the spatial distribution map of the target crop includes: Acquire the time series characteristics of crops in each plot in the plot spatial distribution map; Collecting crop type data of each plot in the plot spatial distribution map on the ground, and using the crop type data collected on the ground as samples to construct a crop type discrimination neural network model; Using satellite data to invert the temporal characteristics of crops in each plot; Inputting the time series features into the crop type discrimination neural network model to determine the crop type of each plot; According to the target crop types of all plots, the spatial distribution map of the target crop is determined.

7. The intelligent extraction method of crop leaf area index according to claim 6, characterized in that: The temporal characteristics of the crops include the normalized difference vegetation index, enhanced vegetation index, land surface water index and greenness normalized difference vegetation index at the plot scale.

8. An intelligent extraction device for crop leaf area index, characterized in that: It includes a first acquisition unit, a first determination unit, a first construction unit, a second construction unit, a first calculation processing unit, a second calculation processing unit, a first training label library establishment unit, a first mapping relationship model establishment unit, a first model training unit and a first generation unit: The first acquisition unit is used to acquire farmland remote sensing images and collect time series remote sensing data; The first determination unit is used to determine the land type of each land parcel according to the farmland remote sensing image; The first construction unit is used to construct a spatial distribution map of the land parcels according to the land parcel types and land parcel spatial forms; The second construction unit is used to identify the target crop type in the spatial distribution map of the plot and construct the spatial distribution map of the target crop; The first calculation processing unit is used to invert the leaf area index of the first crop using satellite data according to the spatial distribution map of the target crop; The second calculation processing unit is used to collect crop spectra of different crops near the ground using a drone to calculate a second crop leaf area index; The first training label library establishing unit is used to establish a first training label library using the first crop leaf area index and the second crop leaf area index; The first mapping relationship model establishing unit is used to determine a first mapping relationship between the first crop leaf area index and the second crop leaf area index, and establish a first mapping relationship model according to the first mapping relationship; The first model training unit is used to train the first mapping relationship model using the first training label library to obtain a first crop leaf area index conversion model; The first generating unit is used to input the first crop leaf area index into the first crop leaf area index conversion model to generate a converted third crop leaf area index.

9. An electronic device, characterized in that: include: Processor and memory; The memory is used to store computer operation instructions; The processor is used to execute the method according to any one of claims 1 to 7 by calling the computer operation instructions.

Images