Land natural scene non-grain crop intelligent identification method and system

Abstract

The application relates to the technical field of non-grain intelligent identification, and discloses a land natural scene non-grain crop intelligent identification method and system, which comprises a collection subsystem, a planning subsystem and a processing subsystem; the collection subsystem comprises a tower; the tower is used for carrying a collection device and providing a communication network; the planning subsystem comprises an intelligent planning module and a collection planning module; the intelligent planning module is used for automatically planning a fixed-point monitoring point, and the tower is arranged at the fixed-point monitoring point; the collection planning module is used for planning a view adjustment scheme of a camera and controlling the action of the camera according to the scheme; the processing subsystem comprises a feature extraction module, an environment analysis module and a judgment module; the environment analysis module is used for confirming an influence type according to environment information of a region to be identified; the feature extraction module is used for extracting non-grain crop features; and the judgment module is used for judging whether a non-grain phenomenon exists in a region corresponding to the features. The application can accurately identify the non-grain phenomenon of cultivated land and has high identification efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of non-grain intelligent identification, and particularly relates to a land natural scene non-grain crop intelligent identification method and system. BACKGROUND

[0002] Intelligent identification of non-grain crops in cultivated areas helps to effectively protect agricultural land.

[0003] Cultivated land is an important basis for the safety of agricultural products and the foundation of agricultural production. However, at present, it is not possible to effectively supervise cultivated land and timely feedback the phenomenon of "non-grain" of cultivated land.

[0004] From the perspective of monitoring, "non-agricultural" and "non-grain" pay attention to the change of cultivated land. For a long time, the change of cultivated land is mainly detected by manual visual interpretation and field investigation, which is low in timeliness and difficult to meet the demand of large-scale, high-frequency and rapid change monitoring of cultivated land protection in the new period. With the development of remote sensing big data, cloud computing, mobile Internet, Internet of Things and artificial intelligence technology, it is possible to monitor the strong and timely change of cultivated land. Specifically, some existing intelligent supervision schemes for land also use unmanned aerial vehicles to patrol the land and use remote sensing image analysis to realize intelligent analysis and judgment of "non-grain". However, such schemes have the following problems:

[0005] (1) The resolution of remote sensing images is low, so that the accuracy of subsequent image color extraction based on remote sensing images is difficult to guarantee, the feature recognition is not accurate, and the supervision precision is not high.

[0006] (2) The patrol cycle of the unmanned aerial vehicle is limited, and the supervision efficiency and cost are high.

[0007] (3) The existing image analysis method is not accurate enough for "non-grain" identification, so that the supervision precision is not high. SUMMARY

[0008] The present application aims to provide a land natural scene non-grain crop intelligent identification method and system to solve the technical problems of low precision and low efficiency of existing cultivated land supervision, and to achieve high "non-grain" identification precision and efficiency.

[0009] To achieve the above purpose, the present application provides the following scheme:

[0010] Scheme one

[0011] The land natural scene non-grain crop intelligent identification system comprises a collection subsystem, a planning subsystem and a processing subsystem.

[0012] The collection subsystem includes a tower; the tower is used to carry a collection device and to provide a communication network; the collection device includes a camera and an environment monitoring device; the camera is used to collect a region image of a region to be identified; the environment monitoring device is used to collect environment information of the region to be identified;

[0013] The planning subsystem includes an intelligent planning module and a collection planning module; the intelligent planning module is used to confirm the region to be identified and to automatically plan a fixed-point monitoring point according to the region to be identified, and the tower is arranged at the fixed-point monitoring point; the collection planning module is used to plan a field-of-view adjustment scheme of the camera and to control the camera to act according to the field-of-view adjustment scheme;

[0014] The processing subsystem includes a feature extraction module, an environment analysis module and a judgment module; the environment analysis module is used to confirm an influence type according to the environment information of the region to be identified, and the influence type has a corresponding relationship with preset common features of non-food crops; the feature extraction module is used to extract the preset common features of non-food crops from the region image, and when the features are extracted, the preset common features matched to the influence type are further subjected to secondary feature extraction, and verification features are obtained; and the judgment module is used to judge whether the region corresponding to the features has a non-food phenomenon according to the number of preset common features and the number of verification features.

[0015] Scheme two

[0016] The land natural scene non-food crop intelligent identification method adopts the land natural scene non-food crop intelligent identification system as described in scheme one.

[0017] The working principle and advantages of the present application are that:

[0018] In the information collection stage, the tower carries the collection device to collect the information of the region to be identified, which breaks through the conventional unmanned aerial vehicle inspection mode, and compared with the conventional unmanned aerial vehicle remote sensing image collection method, the relative fixed-point camera monitoring of the present scheme can obtain higher image collection accuracy, fundamentally solves the problem of low image resolution in the conventional scheme, and can realize continuous monitoring of the region to be identified, and there is no inspection blank. The functional integration design of the tower carrying the collection device realizes the carrying of the collection device by using the structure of the tower, saves the structure setting cost, and also uses the tower to build a stable communication network. In actual application, many lands prone to non-food have poor communication, which actually has a great impact on the collection and transmission efficiency of land information. While improving the collection system, the present scheme naturally optimizes the information transmission environment, which helps to improve the identification and supervision efficiency.

[0019] And, the monitoring point positions in the scheme are obtained according to the to-be-identified region, and the scene monitoring is customized, which is helpful to realize comprehensive monitoring of the to-be-identified region. In addition, the setting of the visual field adjustment scheme of the camera can actually enable the visual fields of the cameras to compensate each other, and the cameras at different positions of the to-be-identified region can interact with each other, so that less cameras can be used to realize more comprehensive monitoring, which is helpful to reduce the identification and supervision cost.

[0020] In the feature extraction stage, unlike the conventional simple non-grain feature or grain feature identification, the scheme sets two types of feature extraction mechanisms (preset common features and verification features), the feature analysis is more detailed, the verification features are further verification of the preset common features, and the feature identification is more reliable and accurate. The conventional scheme does not think of two types of extraction, and cannot achieve two types of feature extraction, which is affected by the basic image acquisition accuracy. Even if the conventional scheme performs secondary feature extraction, the extraction accuracy is not high, and the feature recognition effect is improved very little. However, the present scheme does not have this problem, and the acquisition device can acquire sufficient standard images as the basis for identification.

[0021] In addition, the secondary feature extraction method of the present scheme is special, and environmental information is introduced to assist in identifying the second type of feature (verification feature). The present scheme can confirm the influence type according to the environmental information of the to-be-identified region, and the influence type has a corresponding relationship with the preset common features of non-grain crops. In the actual plant growth process, environmental factors will directly affect the growth state and phenotype characteristics of various plants, making it easy for some non-grain characteristics to be confused with grain characteristics. In addition, in the actual image acquisition scene, environmental factors will directly affect the image acquisition effect, which is easy to affect the feature recognition and extraction accuracy. The present scheme pays attention to the above actual scene application problems, and sets up a corresponding relationship by collecting environmental information, sets up double identification of verification features, checks features that are easy to confuse, are easily affected by the environment, and are prone to identification errors, and finally determines the non-grain phenomenon based on the number of preset common features and the number of verification features. It can reduce environmental influence error and achieve high identification accuracy to effectively monitor the "non-grain" phenomenon. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 It is a system structure schematic diagram of the embodiment one of the land natural scene non-grain crop intelligent identification method and system of the present application.

[0023] Figure 2 It is a part flow schematic diagram of the method of the embodiment one of the land natural scene non-grain crop intelligent identification method and system of the present application.

[0024] Figure 3This is a schematic diagram of a steel tower, representing an embodiment of the intelligent identification method and system for non-grain crops in the natural landscape of the present invention. Detailed Implementation

[0025] The following detailed explanation illustrates the specific implementation methods:

[0026] Example 1

[0027] The basic implementation examples are as follows: Figure 1 As shown:

[0028] The intelligent identification system for non-grain crops in natural landscapes includes a data collection subsystem, a planning subsystem, and a processing subsystem.

[0029] The data acquisition subsystem includes a tower. The tower is used to mount the data acquisition device and to provide a communication network; the data acquisition device includes a camera and environmental monitoring equipment; the camera is used to acquire regional images of the area to be identified; the environmental monitoring equipment is used to acquire environmental information of the area to be identified.

[0030] Specifically, as shown in the attached document Figure 3 As shown, the tower is a miniaturized micro-station tower. Furthermore, in this embodiment, the tower is equipped with a 360° rotating track, on which a camera is mounted, allowing it to move 360° around the tower body. Specifically, the rotating track is located near the top of the tower to obtain a wider field of view for image acquisition; the camera is an ultra-high-definition camera. Environmental monitoring equipment is installed on the upper part of the tower. In this embodiment, the environmental information collected by the environmental monitoring equipment includes weather, temperature, humidity, and light intensity, among other diverse information.

[0031] The planning subsystem includes an intelligent planning module and a data acquisition planning module. The intelligent planning module identifies the area to be identified and automatically plans fixed-point monitoring points based on that area, with the tower located at each monitoring point. The data acquisition planning module plans the camera's field-of-view adjustment scheme and controls the camera's actions according to the scheme.

[0032] Specifically, the planning strategy of the intelligent planning module includes the following sub-steps:

[0033] S1: Call an online map to obtain the geospatial coverage information of the area to be identified; the geospatial coverage information includes the location (latitude and longitude) of the area to be identified, the total area of ​​the area, the boundary line of the area, the altitude of the area, etc.

[0034] S2: Divide the land elevation surface according to the geospatial coverage information; in this embodiment, the elevation difference is greater than [tower height - (1~1.5)] m as the dividing benchmark for the land elevation surface.

[0035] S3: divide each land height surface into a basic land block; the size of the basic land block matches the field of view of the pre-loaded camera. The field of view here refers to the area that can be monitored by the pre-loaded camera.

[0036] S4: take the center point of each basic land block as a fixed monitoring point. The fixed monitoring point obtained by this selection fully considers the land basic features of the to-be-identified region, especially the elevation features, and can comprehensively monitor the land at different levels. For some small part "non-grain" behaviors, it can also complete sufficient information collection to facilitate subsequent accurate identification of small "non-grain" phenomena.

[0037] When the collection planning module plans the field of view adjustment scheme, first, a plurality of cameras in all cameras are randomly selected to adjust the monitoring view angle at a certain rate, and the motion of the plurality of cameras is taken as a reference to adjust the monitoring view angle of the cameras near the plurality of cameras to make the cameras at different points in the to-be-identified region have the interaction with each other, so that fewer cameras can be used to achieve more comprehensive monitoring, which helps to reduce the identification supervision cost.

[0038] The processing subsystem includes a preprocessing module, a feature extraction module, an environment analysis module, and a judgment module.

[0039] The preprocessing module is used for preprocessing the collected regional image; the regional image includes a multispectral image and a multi-view image.

[0040] Specifically, the preprocessing step includes:

[0041] Step 1: delete the fuzzy image in the regional image to simplify the system data processing workload;

[0042] Step 2: divide the input color image into plants and backgrounds by using the K-means algorithm; so as to intuitively analyze the plants;

[0043] Step 3: for the multispectral image, enhance the contrast of the plant part by processing each channel of R, G, and B;

[0044] Step 4: for the multi-view image, uniformly mark the multi-view images of the same plant (or the same region) to facilitate subsequent overall extraction;

[0045] Step 5: for the regional image, set a category mark for the obviously different types of plant images; and calculate the centroid and major axis of each plant, and rotate the plant to make its main axis horizontal, so that the direction of the plant is normal, which is more convenient for subsequent feature extraction.

[0046] The environment analysis module is configured to determine the influence type according to the environment information of the region to be identified, the influence type being in a corresponding relationship with preset common characteristics of the non-food crops. Specifically, the preset common characteristics include stem height, stem diameter, leaf length, leaf width, leaf angle, leaf area, stem color, leaf color, etc. The environment information is also in a corresponding relationship with the influence type, and is preset in the environment analysis module in the form of a comparison table. For example, weak light and heavy rain correspond to a class I negative influence, and strong light and sunny weather correspond to a class I positive influence. The corresponding relationship between the influence type and the preset common characteristics is also preset in the environment analysis module in the form of a comparison table. For example, a class I negative influence corresponds to a negative influence on leaf angle and leaf area, and a class I positive influence corresponds to a positive influence on leaf color, stem color and leaf angle, etc. The corresponding relationship changes according to different shade-tolerant plants and sun-loving plants.

[0047] The feature extraction module is configured to extract the preset common characteristics of the non-food crops from the region image, and when extracting the characteristics, the preset common characteristics matched to the influence type are further subjected to secondary feature extraction, and verification characteristics are obtained. The determination module is configured to determine whether the region corresponding to the characteristics has a non-food phenomenon according to the number of preset common characteristics and the number of verification characteristics.

[0048] When extracting the preset common characteristics, the feature extraction module extracts according to a first extraction strategy. The first extraction strategy includes calculating non-food characteristic parameter values corresponding to each grid pixel of the multispectral image according to the multispectral image, and substituting the calculated non-food characteristic parameter values into a non-food crop identification model. The non-food crops identify each grid pixel of the multispectral image, and the identification result is converted into the number of confirmed preset common characteristics. For the identification result, the present scheme does not focus on the type of non-food crops, but directly confirms the number of non-food characteristics, which can more efficiently complete the non-food determination.

[0049] When extracting the verification characteristics, the feature extraction module extracts according to a second extraction strategy. The second extraction strategy includes extracting a multi-view image corresponding to the preset common characteristics matched to the influence type according to the preset common characteristics matched to the influence type, obtaining point cloud data according to the multi-view image, fitting plant parameters according to the point cloud data, and determining the verification characteristics according to the comparison similarity. Specifically, in the present embodiment, if the comparison similarity is greater than 70%, it is determined that there is one verification characteristic. This setting is relatively strict for the identification of verification characteristics, which helps to improve the accuracy of subsequent non-food determination.

[0050] For example, when the preset common feature of the impact type is the leaf area of a plant, the multi-view image corresponding to the leaf area of the plant is extracted (specifically, a multi-view image containing the leaf feature of the plant in the image, and the image requires that the complete leaf feature can be observed by combination). Then, the point cloud data is obtained according to the multi-view image. Specifically, a conventional multi-view reconstruction method can be used to estimate the position of the 3D point and the camera pose by using the corresponding set of image feature points (extracted from the image) of multiple images with different angles, and then the point cloud data is obtained. After obtaining the point cloud data, the plant parameters are fitted by using the point cloud data. Specifically, the plant parameters fitted by the point cloud are the leaf length, the leaf width and the leaf area (i.e., the leaf area). Specifically, the distance between any two points on the edge of the leaf surface can be calculated according to the leaf surface fitted by the point cloud, and the edge with the maximum distance parallel to the median line of the leaf surface is taken as the leaf length, and the edge with the maximum distance parallel to the leaf length is taken as the leaf width. The leaf area can be calculated according to the leaf length and the leaf width, for example, by integrating the leaf area.

[0051] After obtaining the plant parameters, the plant parameters are compared with the preset plant feature library, and the verification features are determined according to the comparison similarity. The preset plant feature library in this embodiment records the basic phenotype parameter range of different types of plants (specifically, non-food crop plants in this embodiment). If the calculated plant parameters all belong to the basic phenotype parameter range, the comparison similarity is 100%. If the calculated plant parameters include items that do not belong to the basic phenotype parameter range, the comparison similarity is determined according to the ratio of the plant parameter items belonging to the basic phenotype parameter range to the total plant parameter items.

[0052] The determination module includes the following steps when determining whether the feature corresponding region exists non-grain phenomenon:

[0053] Step 1: comparing the number of preset common features identified with the first preset threshold value;

[0054] Step 2: comparing the number of verification features identified with the second preset threshold value;

[0055] Step 3: if the number of identified preset common features is greater than the first preset threshold value, it is determined that the feature corresponding region exists non-grain phenomenon; if the number of identified preset common features is less than the first preset threshold value, and the number of identified verification features is greater than the second preset threshold value, it is determined that the feature corresponding region exists non-grain phenomenon.

[0056] The first preset threshold and the second preset threshold are both dynamic thresholds, and the fluctuation of the dynamic threshold changes with the influence degree of the environmental information. Specifically, the basic value of the first preset threshold and the second preset threshold is set according to the actual judgment requirement. The influence degree of the environmental information here mainly refers to the severity of the collection environment. When the collection environment is obviously severe weather such as rainstorm and typhoon, the preset threshold value is automatically adjusted by 8% to 15% with the change of the severity; when the collection environment is normal weather, the basic value is maintained.

[0057] As shown in the accompanying Figure 2 The embodiment also provides a land natural scene non-food crop intelligent identification method, which adopts the land natural scene non-food crop intelligent identification system.

[0058] The land natural scene non-food crop intelligent identification method and system provided by the embodiment break through the new collection subsystem, can achieve high image collection accuracy and comprehensive land area monitoring effect, and cooperates with the iron tower to build a stable information communication environment, effectively improves the land information collection accuracy and efficiency from the collection stage, and the collection is no longer limited by the unmanned aerial vehicle inspection cycle, and the collection cost is reduced. Cooperate with the planning subsystem, can realize the customized supervision of the cultivated land, customize the monitoring and collection scheme for different types of cultivated land, and through the setting of the planning strategy, for the small "non-food" phenomenon, the scheme can also accurately identify, and the monitoring meticulousness and comprehensiveness are improved. In addition, the scheme takes into account the external environmental information and special collection scene, greatly improves the analysis meticulousness, and through the setting of the double feature extraction analysis, high judgment accuracy can be achieved.

[0059] Embodiment two

[0060] The land natural scene non-food crop intelligent identification system adjusts the iron tower structure and the processing subsystem on the basis of the embodiment one.

[0061] The processing subsystem further includes an alarm module. The alarm module is used to convey alarm information to the corresponding supervisor or supervision center as soon as the judgment module confirms the existence of the non-food phenomenon; the alarm information conveying mode includes information pop-up window, voice notification, short message notification and the like. Optionally, the alarm module also distinguishes the alarm level according to the size of the non-food area when conveying the alarm information, for example: when the non-food area accounts for 0 to 20% of the overall area, the information pop-up window is used for notification; when the non-food area accounts for 20 to 50% of the overall area, the short message and the information pop-up window are used for simultaneous notification; when the non-food area accounts for more than 50% of the overall area, the voice telephone and the information pop-up window are used for simultaneous notification. In this way, different alarms are triggered according to different situations, the alarm is more accurate, and the non-food supervision effect is better.

[0062] And the size of the non-grain area can be confirmed by the determination module. Specifically, after confirming that the feature corresponding area has the non-grain phenomenon, the determination module confirms the area of the feature corresponding area, compares the area with the area of the overall region to be identified, and then obtains the proportion of the non-grain area.

[0063] The lower part of the tower body of the iron tower is further provided with a one-key alarm. The one-key alarm is used for manually conveying alarm information. If a patrol personnel or other personnel finds that there is a non-grain phenomenon in the cultivated land, the one-key alarm can be used to complete the rapid feedback of information.

[0064] The land natural scene non-grain crop intelligent identification method and system provided by the embodiment can intelligently alarm when the non-grain phenomenon occurs, which helps the supervisors to timely control the cultivated land, and the alarm additionally arranged on the iron tower helps to realize the real-time feedback of the first-line information.

[0065] Embodiment three

[0066] The land natural scene non-grain crop intelligent identification system adjusts the planning subsystem and the iron tower structure on the basis of the embodiment one.

[0067] The planning subsystem further comprises an intelligent on-off module. The intelligent on-off module is used for controlling the power on-off of the acquisition device according to an on-off strategy. The on-off strategy comprises that if the determination module does not determine the non-grain phenomenon within a preset period (the preset period is set to one month in this embodiment), the power of the acquisition device is turned off for a certain period of time (the time is also set to one month). The iron tower is correspondingly provided with a power on-off controller, which is electrically connected with the acquisition device and is in communication connection with the intelligent on-off module.

[0068] The land natural scene non-grain crop intelligent identification method and system provided by the embodiment can intelligently turn on and off the power of the acquisition device. In the case of basically no non-grain phenomenon, keeping the power off for a certain period of time helps to save energy consumption.

[0069] Embodiment four

[0070] The land natural scene non-grain crop intelligent identification system adjusts the preprocessing module and the feature extraction module on the basis of the embodiment one.

[0071] The preprocessing module is used for preprocessing the acquired regional image. The preprocessing step further comprises,

[0072] Step 6: For the regional image, the images of the same plant (or the same region) at different time stages are specifically marked. In this embodiment, the time interval difference of 14 days is taken as a time stage.

[0073] The feature extraction module can also extract the verification features according to a third extraction strategy.

[0074] The third extraction strategy includes: extracting the image with specific marks corresponding to the preset common features of the influence type according to the matching; confirming the feature difference value of two adjacent time stages in the image with specific marks; determining the growth vigor (growth direction, growth speed, etc.) of the plant corresponding to the image according to the feature difference value; comparing the growth vigor with the preset plant feature library, and determining the verification features according to the comparison similarity. Specifically, in the embodiment, if the comparison similarity is greater than 80%, it is determined as a verification feature.

[0075] The method and system for intelligent identification of non-grain crops in natural land scenes provided by the embodiment provide a new verification feature extraction strategy, which starts from the growth vigor of the plant. This method is more effective for distinguishing some grain crops and non-grain crops that are extremely similar in appearance, and can achieve more accurate identification and determination of non-grain.

[0076] The above is only an embodiment of the present application, and the common knowledge of the specific structure and characteristics in the scheme is not described in detail. The ordinary skilled person in the art knows all the ordinary technical knowledge in the field of the application before the application date or the priority date, can know all the prior art in the field, and has the ability to apply conventional experimental means before that date. The ordinary skilled person in the art can improve and implement the present scheme based on their own ability under the guidance of the present application. Some typical known structures or known methods should not be an obstacle for the ordinary skilled person in the art to implement the present application. It should be noted that for those skilled in the art, without departing from the structure of the present application, a number of modifications and improvements can be made, which should be considered as the protection scope of the present application. These will not affect the effect and practicality of the present application.

Claims

1. A smart identification system for non-grain crops in natural landscape settings, characterized in that, It includes a data acquisition subsystem, a planning subsystem, and a processing subsystem; The data acquisition subsystem includes a tower; the tower is used to mount the data acquisition device and to provide a communication network; the data acquisition device includes a camera and environmental monitoring equipment; the camera is used to acquire regional images of the area to be identified; the environmental monitoring equipment is used to acquire environmental information of the area to be identified; The planning subsystem includes an intelligent planning module and a data acquisition planning module. The intelligent planning module is used to identify the area to be identified and automatically plan fixed monitoring points based on the area to be identified, with the iron tower located at the fixed monitoring point. The data acquisition planning module is used to plan the field of view adjustment scheme of the camera and control the camera's actions according to the field of view adjustment scheme. The processing subsystem includes a preprocessing module, a feature extraction module, an environmental analysis module, and a decision module; the preprocessing module is used to preprocess the acquired regional images; the regional images include multispectral images and multi-view images; The environmental analysis module is used to confirm the impact type based on the environmental information of the area to be identified. The impact type corresponds to a preset common feature of non-grain crops. The feature extraction module is used to extract the preset common feature of non-grain crops from the area image. When extracting features, the preset common feature that matches the impact type is further extracted to obtain the verification feature. The judgment module is used to determine whether there is a non-grainization phenomenon in the area corresponding to the feature based on the number of preset common features and the number of verification features. The preset common features include stem height, stem diameter, leaf length, leaf width, leaf tilt angle, leaf area, stem color, and leaf color; the influence types include Type I negative influence and Type I positive influence; Type I negative influence corresponds to negative influence on leaf tilt angle and leaf area; Type I positive influence corresponds to positive influence on leaf color, stem color, and leaf tilt angle. When extracting preset common features, the feature extraction module extracts them according to a first extraction strategy. The first extraction strategy includes: calculating the non-grain feature parameter values ​​corresponding to each grid cell of the multispectral image; and substituting the calculated non-grain feature parameter values ​​into the non-grain crop recognition model; the non-grain crop recognizes each grid cell of the multispectral image, and the recognition result is output as the number of confirmed preset common features. When extracting verification features, the feature extraction module extracts them according to a second extraction strategy. The second extraction strategy includes: extracting corresponding multi-view images based on preset common features that match the influence type; obtaining point cloud data based on the multi-view images; fitting plant parameters based on the point cloud data; comparing the plant parameters with a preset plant feature library; and determining the verification features based on the comparison similarity.

2. The intelligent identification system for non-grain crops in natural landscape scenarios according to claim 1, characterized in that, The tower is a miniaturized micro-station tower; the upper part of the tower is equipped with a 360° rotating track, and the camera is installed on the rotating track and can move 360° around the tower body through the rotating track.

3. The intelligent identification system for non-grain crops in natural landscape scenarios according to claim 1, characterized in that, The environmental information includes weather, temperature, humidity, and light intensity.

4. The intelligent identification system for non-grain crops in natural landscape scenarios according to claim 1, characterized in that, When determining whether a non-grain-producing phenomenon exists in the region corresponding to a feature, the determination module includes the following steps: Step 1: Compare the number of pre-defined common features obtained from the identification with the first pre-defined threshold; Step 2: Compare the number of verification features obtained from the identification with the second preset threshold; Step 3: If the number of preset common features identified is greater than the first preset threshold, it is determined that the region corresponding to the feature has a non-grain phenomenon; if the number of preset common features identified is less than the first preset threshold, and the number of verification features identified is greater than the second preset threshold, it is determined that the region corresponding to the feature has a non-grain phenomenon.

5. The intelligent identification system for non-grain crops in natural landscape scenarios according to claim 4, characterized in that, Both the first preset threshold and the second preset threshold are dynamic thresholds; and the fluctuation of the dynamic threshold varies with the degree of influence of environmental information.

6. The intelligent identification system for non-grain crops in natural landscape scenarios according to claim 1, characterized in that, The planning strategy of the intelligent planning module includes the following sub-steps: S1: Call an online map to obtain the geospatial coverage information of the area to be identified; the geospatial coverage information includes the location of the area to be identified, the total area of ​​the area, the boundary line of the area, and the altitude of the area. S2: Based on geospatial coverage information, divide the land elevation surface; and use an elevation difference greater than [tower height - (1~1.5)] m as the benchmark for dividing the land elevation surface; S3: Divide each land elevation surface into basic land blocks; the size of the basic land blocks matches the field of view of the pre-loaded camera; S4: Use the center point of each basic land parcel as the fixed monitoring point.

7. A method for intelligent identification of non-grain crops in natural landscape scenes, characterized in that: The system employs an intelligent identification system for non-grain crops in natural landscape settings as described in any one of claims 1-6.

Images