Wheat identification method and device based on automatic sample collection and storage medium

Abstract

The invention discloses a wheat identification method and device based on automatic sample collection, and a storage medium. The method comprises the following steps: calculating a wheat feature enhanced vegetation index according to a first remote sensing image before wheat ripening and a second remote sensing image after wheat ripening in a to-be-identified area; extracting a first infrared band of the first remote sensing image, a second infrared band of the second remote sensing image and a green light band of a vegetation coverage area of the first remote sensing image; the kNDVI value, the kNDVI maximum value, the green light wave band maximum value and the green light wave band minimum value of the first remote sensing image and the kNDVI difference value of the first remote sensing image and the second remote sensing image are calculated; constructing a ground feature recognition condition according to the third information and the wheat feature enhanced vegetation index to obtain a training sample set; training a preset random forest model according to the training sample set to obtain a wheat recognition model; and inputting the combined remote sensing image into a wheat recognition model to obtain a wheat recognition result. According to the method, interference data are screened out, the adaptability is better, and manual annotation of a large amount of data is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of wheat identification, and in particular to a wheat identification method, device and storage medium based on automatic sample collection. Background Art

[0002] Satellite remote sensing technology is increasingly being used in agricultural monitoring, particularly for the cultivation and growth of major crops like wheat. Remote sensing technology, by acquiring surface reflectance spectral information and combining it with image processing and data analysis techniques, enables the monitoring and management of large areas of farmland. This technology offers advantages such as wide coverage and high timeliness, making it an indispensable tool in agricultural production.

[0003] Currently, satellite remote sensing wheat identification primarily relies on multispectral and hyperspectral imaging technologies. These technologies can identify wheat-growing areas and assess their growth status by capturing the spectral characteristics of wheat at different growth stages. Furthermore, combining them with geographic information systems (GIS) and big data analytics can further improve identification accuracy and efficiency.

[0004] Currently, the mainstream methods for wheat identification using satellite remote sensing, both domestically and internationally, include spectral analysis, machine learning, and deep learning. Spectral analysis utilizes the spectral reflectance characteristics of wheat across different wavelengths for identification, but is susceptible to interference from cloud cover and surface impurities. Machine learning uses trained models for classification, but is highly dependent on sample data and suffers from limited generalization. Deep learning, while building complex neural network models, offers high recognition accuracy but requires large amounts of labeled data and high-performance computing resources. Overall, these methods still face numerous challenges in practical application, such as difficulty in data acquisition, high algorithm complexity, and high computing resource requirements, urgently requiring further technological breakthroughs and optimization.

[0005] Therefore, a wheat identification method based on automatic sample collection was developed to solve the above problems. Summary of the Invention

[0006] The present invention proposes a wheat identification method, device and storage medium based on automatic sample collection to solve the problems that existing wheat identification is easily interfered by cloud cover and surface impurities, has a strong dependence on sample data, and requires a large amount of labeled data and high-performance computing resources.

[0007] The present invention achieves the above-mentioned purpose through the following technical solutions:

[0008] A wheat identification method based on automatic sample collection, comprising:

[0009] Acquiring first information, the first information including a first remote sensing image of the wheat in the area to be identified before it matures and a second remote sensing image of the wheat after it matures;

[0010] extracting second information, the second information including a first red light band and a first blue light band of the first remote sensing image and a second red light band and a second blue light band of the second remote sensing image, and calculating a wheat characteristic enhanced vegetation index based on the second information;

[0011] extracting a first infrared band of the first remote sensing image, a green band of the vegetation coverage area, and a second infrared band of the second remote sensing image;

[0012] Calculating third information based on the first infrared band, the first red band, the second infrared band, the second red band, the wheat characteristic enhanced vegetation index, and the green band, the third information including a kNDVI value, a maximum kNDVI value, a maximum value of the green band, and a minimum value of the green band of the first remote sensing image, and a kNDVI difference between the first remote sensing image and the second remote sensing image;

[0013] constructing a ground object recognition condition based on the third information and the wheat feature enhanced vegetation index, and automatically sampling the first remote sensing image and the second remote sensing image based on the ground object recognition condition to obtain a training sample set;

[0014] A preset random forest model is trained according to the training sample set to obtain a wheat recognition model;

[0015] Performing band combination on the first remote sensing image and the second remote sensing image to obtain a combined remote sensing image;

[0016] The combined remote sensing image is input into the wheat recognition model to obtain a wheat recognition result.

[0017] Specifically, the wheat characteristic enhanced vegetation index is calculated according to the second information. The calculation formula of the wheat characteristic enhanced vegetation index is as follows: ,

[0018] In this formula, represents the wheat characteristic enhanced vegetation index, represents the first red light band, represents the first blue light band, represents the second red light band, Indicates the second blue light band.

[0019] Specifically, calculating the third information according to the first infrared band, the first red band, the second infrared band, the second red band, the wheat feature enhanced vegetation index, and the green band includes:

[0020] Calculate the kNDVI value of the first remote sensing image according to the first infrared band and the first red band;

[0021] Calculate the kNDVI value of the second remote sensing image according to the second infrared band and the second red band;

[0022] Calculating a kNDVI difference between the first remote sensing image and the second remote sensing image according to the kNDVI value of the first remote sensing image and the kNDVI value of the second remote sensing image; Filter out the maximum value and the minimum value of the green light band according to the green light band;

[0023] The maximum kNDVI value of the first remote sensing image is screened out according to the kNDVI value of the first remote sensing image.

[0024] Specifically, constructing a ground object recognition condition based on the third information and the wheat feature enhanced vegetation index, and automatically sampling the first remote sensing image and the second remote sensing image based on the ground object recognition condition to obtain a training sample set, including:

[0025] Constructing ground feature recognition conditions, wherein the ground feature recognition conditions include wheat recognition conditions, other vegetation recognition conditions, water body or wetland recognition conditions, and other ground feature recognition conditions;

[0026] Marking pixels in the first remote sensing image and the second remote sensing image according to the ground object recognition condition to obtain a marked pixel set;

[0027] A training sample set is generated based on the labeled pixel set, wherein the training sample set includes pixel features and corresponding category labels, wherein the features include the third information of the pixel and the wheat feature enhanced vegetation index, and the category labels include wheat, other vegetation, water bodies or wetlands, and other land features.

[0028] Specifically, the conditions for ground feature recognition are constructed, including:

[0029] Construct wheat identification conditions, the wheat identification conditions are XMI< *XMI min And, at the same time *dkNDVI max , while kNDVI1>0 and BandG< *(BandG max -BandGmin )+BandG min ;

[0030] Construct other vegetation identification conditions, the other vegetation identification conditions are XMI> *XMI min and dkNDVI> *dkNDVI max And kNDVI1>0;

[0031] Establishing a water body or wetland identification condition, wherein the water body or wetland identification condition is kNDVI1<0;

[0032] Construct other ground feature recognition conditions, the other ground feature recognition conditions are XMI> *XMI min or dkNDVI< *dkNDVI max ;

[0033] Among them, XMI represents the wheat characteristic enhanced vegetation index, Indicates the first preset threshold, XMI min represents the minimum value of wheat characteristic enhanced vegetation index, dkNDVI represents the kNDVI difference between the first remote sensing image and the second remote sensing image, Indicates the second preset threshold, dkNDVI max represents the maximum kNDVI difference between the first remote sensing image and the second remote sensing image, kNDVI1 represents the kNDVI value of the first remote sensing image, and represents the third preset threshold, BandG represents the green light band, BandG max Indicates the maximum value of the green light band, BandG min Indicates the minimum value of the green light band.

[0034] Specifically, marking pixels in the first remote sensing image and the second remote sensing image according to the ground object recognition condition to obtain a marked pixel set includes:

[0035] Calculating fourth information, the fourth information including the third information and the wheat feature enhanced vegetation index of pixels corresponding to the same position in the first remote sensing image and the second remote sensing image;

[0036] Marking all pixels that meet the wheat recognition condition according to the fourth information to form a first pixel set;

[0037] Marking all pixels that meet other vegetation recognition conditions according to the fourth information to form a second pixel set;

[0038] Marking all pixels that meet the water body or wetland identification condition according to the fourth information to form a third pixel set;

[0039] Marking all pixels meeting other vegetation recognition conditions according to the fourth information to form a fourth pixel set;

[0040] The marked pixel set is constructed according to the first pixel set, the second pixel set, the third pixel set and the fourth pixel set.

[0041] Specifically, a preset random forest model is trained according to the training sample set to obtain a wheat recognition model, including:

[0042] Build a random forest model based on the random forest algorithm;

[0043] Constructing multiple decision trees based on the features and the category labels;

[0044] According to the random forest model, a subset is randomly extracted from the training sample set based on bootstrap sampling to train a plurality of decision trees;

[0045] The wheat recognition model is constructed based on a plurality of the decision trees.

[0046] The present invention also provides a wheat identification device based on automatic sample collection, comprising:

[0047] an acquisition module, the acquisition module being configured to acquire first information, the first information comprising a first remote sensing image of the wheat in the area to be identified before it matures and a second remote sensing image of the wheat after it matures;

[0048] a first extraction module, the first extraction module being configured to extract second information, the second information including a first red light band and a first blue light band of the first remote sensing image and a second red light band and a second blue light band of the second remote sensing image, and calculating a wheat characteristic enhanced vegetation index based on the second information;

[0049] a second extraction module, configured to extract a first infrared band of the first remote sensing image, a green band of the vegetation coverage area, and a second infrared band of the second remote sensing image;

[0050] A calculation module, wherein the calculation module is used to calculate third information based on the first infrared band, the first red band, the second infrared band, the second red band, the wheat characteristic enhanced vegetation index, and the green band, wherein the third information includes a kNDVI value, a maximum kNDVI value, a maximum value of the green band, and a minimum value of the green band of the first remote sensing image, and a kNDVI difference between the first remote sensing image and the second remote sensing image;

[0051] a sampling module, configured to automatically sample the first remote sensing image and the second remote sensing image according to the third information and the wheat feature enhanced vegetation index to obtain a training sample set;

[0052] A training module, wherein the training module is used to train a preset random forest model according to the training sample set to obtain a wheat recognition model;

[0053] a combining module, configured to combine the first remote sensing image and the second remote sensing image by performing band combination to obtain a combined remote sensing image;

[0054] The recognition module is used to input the combined remote sensing image into the wheat recognition model to obtain a wheat recognition result.

[0055] The present invention also provides a wheat identification device based on automatic sample collection, comprising:

[0056] memory for storing computer programs;

[0057] A processor is configured to implement the steps of the wheat identification method based on automatic sample collection when executing the computer program.

[0058] The present invention also provides a storage medium, which is a readable storage medium. A computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the wheat identification method based on automatic sample collection are implemented.

[0059] The beneficial effects of the present invention are:

[0060] The present invention proposes a wheat identification method based on automatic sample collection. By constructing ground feature identification conditions based on the third information and the wheat feature enhanced vegetation index, the first remote sensing image and the second remote sensing image are automatically sampled, thereby filtering out interference data, reducing sample data dependence, and having better adaptability. It can avoid manual labeling of large amounts of data while ensuring recognition accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 This is a flowchart of an implementation method of wheat identification based on automatic sample collection in an embodiment of the present application.

[0062] Figure 2 This is the multispectral remote sensing data of the month when wheat matures and the two months before.

[0063] Figure 3 This is the wheat enhanced vegetation index map in the embodiment of this application.

[0064] Figure 4This is the dkNDVI map of the wheat mature month and the previous two months in the area to be identified in the embodiment of this application.

[0065] Figure 5 This is the wheat recognition result in the examples of this application. DETAILED DESCRIPTION

[0066] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings of the embodiments of the present invention. It should be understood that the described embodiments are only a portion of the embodiments of the present invention, not all of them. Generally, the components of the embodiments of the present invention described and illustrated in the drawings herein may be arranged and designed in a variety of different configurations.

[0067] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort shall fall within the scope of protection of the present invention.

[0068] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings.

[0069] In the description of the present invention, it should be understood that the terms "upper", "lower", "inside", "outside", "left", "right", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, or are the orientations or positional relationships in which the inventive product is conventionally placed when in use, or are the orientations or positional relationships conventionally understood by those skilled in the art. These are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or component referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0070] Furthermore, the terms “first”, “second”, etc. are merely used for distinguishing descriptions and should not be understood as indicating or implying relative importance.

[0071] In the description of the present invention, it should also be noted that, unless otherwise expressly specified or limited, terms such as "disposed" and "connected" should be understood in a broad sense. For example, "connected" can mean a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can also mean internal communication between two components. Those skilled in the art will be able to understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0072] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0073] like Figure 1 As shown, this embodiment provides a wheat identification method based on automatic sample collection, including:

[0074] Acquiring first information, the first information including a first remote sensing image of the wheat in the area to be identified before it matures and a second remote sensing image of the wheat after it matures;

[0075] extracting second information, the second information including a first red light band of the first remote sensing image, a first blue light band of the first remote sensing image, a second red light band of the second remote sensing image, and a second blue light band of the second remote sensing image, and calculating a wheat characteristic enhanced vegetation index based on the second information;

[0076] extracting a first infrared band of the first remote sensing image, a second infrared band of the second remote sensing image, and a green light band of a vegetation coverage area of ​​the first remote sensing image;

[0077] Calculating third information based on the first infrared band, the first red band, the second infrared band, the second red band, the wheat characteristic enhanced vegetation index, and the green band, the third information including a kNDVI value, a maximum kNDVI value, a maximum value of the green band, and a minimum value of the green band of the first remote sensing image, and a kNDVI difference between the first remote sensing image and the second remote sensing image;

[0078] constructing a ground object recognition condition based on the third information and the wheat feature enhanced vegetation index, and automatically sampling the first remote sensing image and the second remote sensing image based on the ground object recognition condition to obtain a training sample set;

[0079] A preset random forest model is trained according to the training sample set to obtain a wheat recognition model;

[0080] Performing band combination on the first remote sensing image and the second remote sensing image to obtain a combined remote sensing image;

[0081] The combined remote sensing image is input into the wheat recognition model to obtain a wheat recognition result.

[0082] In this embodiment, the area to be identified is located in Suining City. Figure 2 As shown ( Figure 2In this example, Sentinel-2A multispectral data from March 21, 2024 and May 6, 2024 were acquired. Bands 2, 3, 4, and 8a in the Sentinel-2A multispectral data were selected as input for a wheat identification method based on automatic sample collection. Figure 5 Shown is the wheat recognition result of this embodiment.

[0083] In some embodiments, the wheat feature enhanced vegetation index is calculated based on the second information, and the calculation formula of the wheat feature enhanced vegetation index is as follows:

[0084]

[0085] In this formula, represents the wheat characteristic enhanced vegetation index, represents the first red light band, represents the first blue light band, represents the second red light band, Indicates the second blue light band. The wheat characteristic enhanced vegetation index is as follows Figure 3 shown.

[0086] In some embodiments, calculating third information based on the first infrared band, the first red band, the second infrared band, the second red band, the wheat feature enhanced vegetation index, and the green band includes:

[0087] Calculate the kNDVI value of the first remote sensing image according to the first infrared band and the first red band;

[0088] Calculate the kNDVI value of the second remote sensing image according to the second infrared band and the second red band;

[0089] The kNDVI difference between the first remote sensing image and the second remote sensing image is calculated based on the kNDVI value of the first remote sensing image and the kNDVI value of the second remote sensing image. The kNDVI difference is as follows: Figure 4 As shown;

[0090] The calculation formula of kNDVI value is as follows:

[0091]

[0092]

[0093] Where: NIR is the near-infrared band of the image, and Red is the red light band of the image.

[0094] Filter out the maximum value and the minimum value of the green light band according to the green light band;

[0095] The maximum kNDVI value of the first remote sensing image is screened out according to the kNDVI value of the first remote sensing image.

[0096] In some embodiments, constructing a ground object recognition condition based on the third information and the wheat feature enhanced vegetation index, and automatically sampling the first remote sensing image and the second remote sensing image based on the ground object recognition condition to obtain a training sample set includes:

[0097] Constructing ground feature recognition conditions, wherein the ground feature recognition conditions include wheat recognition conditions, other vegetation recognition conditions, water body or wetland recognition conditions, and other ground feature recognition conditions;

[0098] Marking pixels in the first remote sensing image and the second remote sensing image according to the ground object recognition condition to obtain a marked pixel set;

[0099] A training sample set is generated based on the labeled pixel set, wherein the training sample set includes pixel features and corresponding category labels, wherein the features include the third information of the pixel and the wheat feature enhanced vegetation index, and the category labels include wheat, other vegetation, water bodies or wetlands, and other land features.

[0100] In some embodiments, constructing the ground feature recognition conditions includes:

[0101] Construct wheat identification conditions, the wheat identification conditions are XMI< *XMI min And, at the same time *dkNDVI max , while kNDVI1>0 and BandG< *(BandG max -BandG min )+BandG min ;

[0102] Construct other vegetation identification conditions, the other vegetation identification conditions are XMI> *XMI min and dkNDVI> *dkNDVI max And kNDVI1>0;

[0103] Establishing a water body or wetland identification condition, wherein the water body or wetland identification condition is kNDVI1<0;

[0104] Construct other ground feature recognition conditions, the other ground feature recognition conditions are XMI> *XMI min or dkNDVI< *dkNDVI max ;

[0105] Among them, XMI represents the wheat characteristic enhanced vegetation index, Indicates the first preset threshold, XMI min represents the minimum value of wheat characteristic enhanced vegetation index, dkNDVI represents the kNDVI difference between the first remote sensing image and the second remote sensing image, Indicates the second preset threshold, dkNDVI max represents the maximum kNDVI difference between the first remote sensing image and the second remote sensing image, kNDVI1 represents the kNDVI value of the first remote sensing image, and represents the third preset threshold, BandG represents the green light band, BandG max Indicates the maximum value of the green light band, BandG min Indicates the minimum value of the green light band.

[0106] In this embodiment, and The default values ​​are all 0.1. The default value is 0.8, and in other embodiments, it can be adjusted appropriately according to the image quality.

[0107] In some embodiments, marking pixels in the first remote sensing image and the second remote sensing image according to the ground object recognition condition to obtain a marked pixel set includes:

[0108] Calculating fourth information, the fourth information including the third information and the wheat feature enhanced vegetation index of pixels corresponding to the same position in the first remote sensing image and the second remote sensing image;

[0109] Marking all pixels that meet the wheat recognition condition according to the fourth information to form a first pixel set;

[0110] Marking all pixels that meet other vegetation recognition conditions according to the fourth information to form a second pixel set;

[0111] Marking all pixels that meet the water body or wetland identification condition according to the fourth information to form a third pixel set;

[0112] Marking all pixels meeting other vegetation recognition conditions according to the fourth information to form a fourth pixel set;

[0113] The marked pixel set is constructed according to the first pixel set, the second pixel set, the third pixel set and the fourth pixel set.

[0114] In some embodiments, a preset random forest model is trained according to the training sample set to obtain a wheat recognition model, including:

[0115] Build a random forest model based on the random forest algorithm;

[0116] Constructing multiple decision trees based on the features and the category labels;

[0117] According to the random forest model, a subset is randomly extracted from the training sample set based on bootstrap sampling to train a plurality of decision trees;

[0118] The wheat recognition model is constructed based on a plurality of the decision trees.

[0119] This embodiment also provides a wheat identification device based on automatic sample collection, comprising:

[0120] an acquisition module, the acquisition module being configured to acquire first information, the first information comprising a first remote sensing image of the wheat in the area to be identified before it matures and a second remote sensing image of the wheat after it matures;

[0121] a first extraction module, the first extraction module being configured to extract second information, the second information including a first red light band of the first remote sensing image, a first blue light band of the first remote sensing image, a second red light band of the second remote sensing image, and a second blue light band of the second remote sensing image, and calculating a wheat characteristic enhanced vegetation index based on the second information;

[0122] a second extraction module, configured to extract a first infrared band of the first remote sensing image, a second infrared band of the second remote sensing image, and a green band of the vegetation coverage area of ​​the first remote sensing image;

[0123] A calculation module, wherein the calculation module is used to calculate third information based on the first infrared band, the first red band, the second infrared band, the second red band, the wheat characteristic enhanced vegetation index, and the green band, wherein the third information includes a kNDVI value, a maximum kNDVI value, a maximum value of the green band, and a minimum value of the green band of the first remote sensing image, and a kNDVI difference between the first remote sensing image and the second remote sensing image;

[0124] a sampling module, constructing a ground object recognition condition according to the third information and the wheat feature enhanced vegetation index, and automatically sampling the first remote sensing image and the second remote sensing image according to the ground object recognition condition to obtain a training sample set;

[0125] A training module, wherein the training module is used to train a preset random forest model according to the training sample set to obtain a wheat recognition model;

[0126] a combining module, configured to combine the first remote sensing image and the second remote sensing image by performing band combination to obtain a combined remote sensing image;

[0127] The recognition module is used to input the combined remote sensing image into the wheat recognition model to obtain a wheat recognition result.

[0128] This embodiment also provides a wheat identification device based on automatic sample collection, comprising:

[0129] memory for storing computer programs;

[0130] A processor is configured to implement the steps of the wheat identification method based on automatic sample collection when executing the computer program.

[0131] This embodiment further provides a storage medium, which is a readable storage medium. A computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the wheat identification method based on automatic sample collection are implemented.

[0132] The present invention proposes a wheat identification method based on automatic sample collection. By constructing ground feature identification conditions based on the third information and the wheat feature enhanced vegetation index, the first remote sensing image and the second remote sensing image are automatically sampled, thereby filtering out interference data, reducing sample data dependence, and having better adaptability. It can avoid manual labeling of large amounts of data while ensuring recognition accuracy.

[0133] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A wheat identification method based on automatic sample collection, characterized in that: include: Acquire first information, wherein the first information includes a first remote sensing image of the wheat in the to-be-identified area before it matures and a second remote sensing image of the wheat after it matures; extracting second information, where the second information includes a first red light band and a first blue light band of the first remote sensing image and a second red light band and a second blue light band of the second remote sensing image, and calculating a wheat characteristic enhanced vegetation index according to the second information; Extracting a first infrared band of the first remote sensing image, a green light band of the vegetation coverage area, and a second infrared band of the second remote sensing image; Calculate third information according to the first infrared band, the first red band, the second infrared band, the second red band, the wheat characteristic enhanced vegetation index and the green band, wherein the third information includes a kNDVI value, a maximum kNDVI value, a maximum value of a green band and a minimum value of a green band of the first remote sensing image, and a kNDVI difference between the first remote sensing image and the second remote sensing image; constructing a ground object recognition condition according to the third information and the wheat characteristic enhanced vegetation index, and automatically sampling the first remote sensing image and the second remote sensing image according to the ground object recognition condition to obtain a training sample set; A preset random forest model is trained according to the training sample set to obtain a wheat recognition model; Combining the first remote sensing image and the second remote sensing image by band combining to obtain a combined remote sensing image; The combined remote sensing image is input into the wheat recognition model to obtain a wheat recognition result.

2. A wheat identification method based on automatic sample collection according to claim 1, characterized in that: The wheat characteristic enhanced vegetation index is calculated according to the second information. The calculation formula of the wheat characteristic enhanced vegetation index is as follows:

3. In this formula, represents the wheat characteristic enhanced vegetation index, represents the first red light band, represents the first blue light band, represents the second red light band, Indicates the second blue light band.

4. A wheat identification method based on automatic sample collection according to claim 1, characterized in that: Calculating third information according to the first infrared band, the first red band, the second infrared band, the second red band, the wheat characteristic enhanced vegetation index and the green band includes: Calculate the kNDVI value of the first remote sensing image according to the first infrared band and the first red band; Calculate the kNDVI value of the second remote sensing image according to the second infrared band and the second red band; Calculating a kNDVI difference between the first remote sensing image and the second remote sensing image according to the kNDVI value of the first remote sensing image and the kNDVI value of the second remote sensing image; Filter out the maximum value and the minimum value of the green light band according to the green light band; The kNDVI maximum value of the first remote sensing image is screened out according to the kNDVI value of the first remote sensing image.

5. The wheat identification method based on automatic sample collection according to claim 1, characterized in that: Constructing a ground object recognition condition according to the third information and the wheat feature enhanced vegetation index, and automatically sampling the first remote sensing image and the second remote sensing image according to the ground object recognition condition to obtain a training sample set, including: Constructing ground object recognition conditions, wherein the ground object recognition conditions include wheat recognition conditions, other vegetation recognition conditions, water body or wetland recognition conditions and other ground object recognition conditions; Marking pixels in the first remote sensing image and the second remote sensing image according to the ground object recognition condition to obtain a marked pixel set; A training sample set is generated based on the labeled pixel set, the training sample set including pixel features and corresponding category labels, the features including the third information of the pixels and the wheat feature enhanced vegetation index, and the category labels including wheat, other vegetation, water bodies or wetlands and other landforms.

6. A wheat identification method based on automatic sample collection according to claim 4, characterized in that: Construct the conditions for object recognition, including: Construct wheat identification conditions, the wheat identification conditions are XMI< *XMI min And, at the same time dkNDVI> *dkNDVI max , while kNDVI1>0 and BandG< *(BandG max -BandG min )+BandG min ; Construct other vegetation identification conditions, the other vegetation identification conditions are XMI> *XMI min And dkNDVI> *dkNDVI max And kNDVI1>0; Establishing a water body or wetland identification condition, wherein the water body or wetland identification condition is kNDVI1<0; Construct other ground object recognition conditions, the other ground object recognition conditions are XMI> *XMI min or dkNDVI< *dkNDVI max ; Among them, XMI represents the wheat characteristic enhanced vegetation index, Indicates the first preset threshold, XMI min represents the minimum value of wheat characteristic enhanced vegetation index, dkNDVI represents the kNDVI difference between the first remote sensing image and the second remote sensing image, Indicates the second preset threshold, dkNDVI max represents the maximum kNDVI difference between the first remote sensing image and the second remote sensing image, kNDVI1 represents the kNDVI value of the first remote sensing image, and represents the third preset threshold, BandG represents the green light band, BandG max Indicates the maximum value of the green light band, BandG min Indicates the minimum value of the green light band.

7. A wheat identification method based on automatic sample collection according to claim 4, characterized in that: Marking pixels in the first remote sensing image and the second remote sensing image according to the ground object recognition condition to obtain a marked pixel set includes: Calculating fourth information, wherein the fourth information includes the third information and the wheat characteristic enhanced vegetation index of the first remote sensing image and the second remote sensing image at pixels corresponding to the same position; Marking all the pixels meeting the wheat recognition condition according to the fourth information to form a first pixel set; Marking all the pixels meeting other vegetation recognition conditions according to the fourth information to form a second pixel set; Marking all the pixels that meet the water body or wetland identification condition according to the fourth information to form a third pixel set; Marking all the pixels meeting other vegetation recognition conditions according to the fourth information to form a fourth pixel set; The marked pixel set is formed according to the first pixel set, the second pixel set, the third pixel set and the fourth pixel set.

8. A wheat identification method based on automatic sample collection according to claim 4, characterized in that: The preset random forest model is trained according to the training sample set to obtain a wheat recognition model, including: Build a random forest model based on the random forest algorithm; Constructing multiple decision trees according to the features and the category labels; According to the random forest model, a subset is randomly selected from the training sample set based on bootstrap sampling to train a plurality of decision trees; The wheat recognition model is constructed based on a plurality of the decision trees.

9. A wheat identification device based on automatic sample collection, characterized in that: include: Memory for storing computer programs; A processor, configured to implement the steps of a wheat identification method based on automatic sample collection as claimed in any one of claims 1 to 7 when executing the computer program.

10. A storage medium, characterized in that: The storage medium is a readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the wheat identification method based on automatic sample collection as described in any one of claims 1 to 7 are implemented.

Images