A smart agricultural system based on big data

Abstract

The present invention relates to the field of smart agriculture technology, and in particular to a smart agriculture system based on big data. The present invention is provided with an inspection module, a regular distribution recognition module, an image extraction module and a model analysis module. The inspection module is used to collect images of a target area, and specific continuous texture segments are identified based on chromaticity. The distribution regularity characterization coefficient is calculated according to the number of specific continuous texture segments and the angle difference between the specific continuous texture segments, and the regular distribution category of the target area image is divided. According to the regular distribution category, the texture difference analysis and chromaticity analysis are performed on the target area image, and abnormal local images are identified and transmitted to the model analysis module, and the pest and disease characteristics in the abnormal local images transmitted by the image extraction module are identified. The present invention can accurately and quickly identify pest and disease characteristics, avoid repeated traversal, avoid waste of computing power, and improve the computing speed.

Description

Technical Field

[0001] The present invention relates to the technical field of smart agriculture, and in particular to a smart agriculture system based on big data. Background Art

[0002] Smart agriculture is a major development direction under the background of new rural areas. As an agricultural form that deeply integrates modern information technology with agricultural production, smart agriculture is of great significance to promoting agricultural modernization. It uses information technologies such as the Internet of Things, big data, spatial information, artificial intelligence, and blockchain, combined with modern biotechnology and planting and breeding technology, and through the online, data-based, and networked agricultural production factors such as agricultural resources, environment, and facilities and equipment, it realizes comprehensive perception, quantitative decision-making, intelligent control, precise investment, and professional services for agricultural production and management, making agricultural production more convenient and management more efficient.

[0003] Chinese patent publication number: CN118366105A, discloses a mung bean pest intelligent monitoring system, including image acquisition equipment, image processing module, data transmission module, data analysis module and data display equipment. The data analysis module includes a mung bean weevil trapping image analysis module, a mung bean leaf pest analysis module and a mung bean pest comprehensive analysis module, and proposes a mung bean weevil trapping analysis formula, a mung bean leaf pest analysis formula and a mung bean pest comprehensive analysis formula; through image processing and data analysis technology, the degree of combined pests and diseases of mung bean weevil, anthracnose and leaf spot of mung bean plants in the field is monitored in real time, and the deep learning algorithm is used to identify the characteristics of mung bean weevil and its biting parts, the characteristics of anthracnose and leaf spot leaf diseases through field images, and the mung bean weevil, anthracnose and leaf spot disease conditions are automatically analyzed.

[0004] In the prior art, when surveying and monitoring large areas of farmland, computer image recognition methods are mostly used to analyze the characteristics of farmland pests and diseases. However, photographing large areas of farmland will generate massive images. Identifying pests and diseases in these images will slow down the computer and cause a waste of computing power. Based on the above situation, the present invention considers dividing the target area into regular distributions, identifying areas with abnormal tendencies and analyzing only these areas, which effectively improves the calculation speed, avoids waste of computing power, and improves analysis efficiency. Summary of the invention

[0005] To this end, the present invention provides a rural environmental supervision system based on a cloud platform to solve the problem that when surveying and monitoring large areas of farmland, the computer identifies pests and diseases from the massive images obtained, which will cause the computer to run slowly and cause a waste of computing power.

[0006] To achieve the above objectives, the present invention provides a rural environmental supervision system based on a cloud platform, which includes:

[0007] An inspection module, comprising a mobile inspection unit for moving in a target area and an image acquisition device arranged on the mobile inspection unit for acquiring an image of the target area;

[0008] A regular distribution recognition module, which is connected to the inspection module, is used to recognize specific continuous texture segments based on chromaticity, calculate the distribution regularity characterization coefficient according to the number of specific continuous texture segments and the angle difference between each specific continuous texture segment, and classify the regular distribution category of the target area image;

[0009] An image extraction module, which is connected to the regular distribution recognition module and includes a texture difference analysis unit, a chromaticity difference analysis unit and a scheduling unit,

[0010] The scheduling unit is used to call the texture difference analysis unit and the chromaticity difference analysis unit based on the division result of the regular distribution recognition module;

[0011] The texture difference analysis unit is used to identify abnormal sub-texture segments based on the curvature characteristics of the sub-texture segments and the adjacent sub-texture segments in the continuous texture segments in the target area image, determine the abnormal local image based on the center coordinates of the abnormal sub-texture segments and transmit it to the model analysis module;

[0012] The chromaticity difference analysis unit is used to determine the abnormal local image based on the chromaticity difference between each sub-region and the adjacent sub-region in the target region image and transmit it to the model analysis module;

[0013] A model analysis module, which is connected to the image extraction module and is used to identify the characteristics of pests and diseases in the abnormal local image transmitted by the image extraction module.

[0014] Furthermore, the regular distribution recognition module is used to recognize specific continuous texture segments based on chromaticity, including:

[0015] Used to identify all continuous texture segments in the target area image;

[0016] To determine the chromaticity of all continuous texture segments, and calculate the difference between the chromaticity corresponding to each continuous texture segment and the average chromaticity of the target area image;

[0017] for identifying a continuous texture segment that meets the identification conditions as a specific continuous texture segment;

[0018] The recognition condition includes that the difference value corresponding to the continuous texture segment is greater than a predetermined difference threshold.

[0019] Furthermore, the regular distribution recognition module calculates the distribution regularity characterization coefficient according to formula (1):

[0020]

[0021] In formula (1), D represents the distribution law characterization coefficient, M represents the number of specific continuous texture segments, M0 represents the number of reference specific continuous texture segments, A represents the angle difference value between each specific continuous texture segment, A0 represents the angle difference value between reference specific continuous texture segments, α represents the influence coefficient of the number, and β represents the influence coefficient of the angle difference value.

[0022] Furthermore, the regular distribution recognition module divides and determines the regular distribution category of the target area image based on the distribution regularity characterization coefficient, wherein:

[0023] If the distribution law representation coefficient is greater than or equal to the reference distribution law representation coefficient, the target area image is determined to be a dominant law category.

[0024] If the distribution law representation coefficient is less than the reference distribution law representation coefficient, the target area image is determined to be a hidden law category.

[0025] Further, the scheduling unit calls the texture difference analysis unit and the chromaticity difference analysis unit based on the division result of the regular distribution recognition module, including:

[0026] If the target area image is of the dominant regularity category, calling the texture difference analysis unit and the chromaticity difference analysis unit to analyze the target area image;

[0027] If the target area image is of an implicit regularity category, a chromaticity difference analysis unit is called to analyze the target area image.

[0028] Furthermore, the texture difference analysis unit is used to identify abnormal sub-texture segments, including:

[0029] for dividing the continuous texture segment in the target area image into a plurality of sub-texture segments;

[0030] It is used to calculate the difference between the curvature features of the sub-texture segment and the adjacent sub-texture segment, and identify the sub-texture segment whose difference is greater than a predetermined curvature difference threshold as an abnormal sub-texture segment.

[0031] Further, the texture difference analysis unit determines the abnormal local image based on the center coordinates of the abnormal sub-texture segment, including:

[0032] To obtain the center coordinates of the abnormal sub-texture segment and set it as the center of the circle;

[0033] to obtain the length of the abnormal sub-texture segment and determine the radius according to the length;

[0034] A circle is selected based on the center and radius, and an image in the circle is determined as an abnormal local image.

[0035] Further, the chromaticity difference analysis unit is used to determine the abnormal local image including:

[0036] for identifying the chromaticity difference between each sub-region and an adjacent sub-region in the target region image;

[0037] Used to determine the chromaticity difference boundary point;

[0038] Used to connect the boundary points to determine the closed area;

[0039] The image within the closed area is used to determine an abnormal local image.

[0040] Furthermore, it also includes an abnormal storage module, which is connected to the image extraction module and is used to store the abnormal local image identified by the image extraction module and the corresponding target area image.

[0041] Furthermore, it also includes a display module for displaying the data extracted by the image extraction module.

[0042] Compared with the prior art, the present invention is provided with an inspection module, a regular distribution recognition module, an image extraction module and a model analysis module, and the inspection module is used to collect the target area image, and the specific continuous texture segment is recognized based on the chromaticity, and the distribution regularity characterization coefficient is calculated according to the number of the specific continuous texture segments and the angle difference between the specific continuous texture segments, and the regular distribution category of the target area image is divided, and the texture difference analysis and chromaticity analysis are performed on the target area image according to the regular distribution category, and the abnormal local image is identified and transmitted to the model analysis module, and the pest and disease characteristics in the abnormal local image transmitted by the image extraction module are identified. Through the above method, the present invention is suitable for the analysis of large-area farmland images, can accurately and quickly identify the pest and disease characteristics, avoid repeated traversal, avoid waste of computing power, and improve the calculation speed and analysis efficiency.

[0043] In particular, the present invention recognizes massive images based on chromaticity. In actual situations, all images are usually transmitted to a computer for characteristic analysis of farmland pests and diseases. However, for large areas of farmland, there will be many pictures without abnormal pests and diseases. The computer uses a pre-trained model to comprehensively analyze these images, resulting in a waste of computing power. Therefore, the present invention considers calculating the distribution law characterization coefficient based on the number of specific continuous texture segments and the angle difference between the specific continuous texture segments. In actual situations, due to crop planting habits, crops in more farmlands are equally spaced and arranged in a regular pattern, which is reflected in aerial images as more regular textures. Therefore, the present invention considers calculating the distribution law characterization coefficient, identifying target area images with more obvious texture patterns, and then dividing regular distribution categories. Subsequently, different analysis methods are used for different categories of target area images for targeted analysis to avoid repeated traversal and waste of computing power, thereby improving analysis efficiency while ensuring reliability.

[0044] In particular, the present invention considers using texture difference analysis and chromaticity difference analysis under the explicit regularity category. The data representation of the texture features in the target area image under the explicit regularity category is relatively strong. Based on the texture features, abnormal local images that may contain pests and diseases can be preliminarily identified. In actual situations, pests and diseases may affect the growth of crops, but may not be reflected in the color features, which may cause the crops to fall over or grow poorly, which will be reflected in the target area image and cause the continuous texture segments to bend. For target area images with strong texture regularity, texture features can be included to preliminarily determine the presence of pests and diseases in the current area. At the same time, the target area image is comprehensively determined and identified in combination with chromaticity analysis for abnormal local images. Only the pre-trained model is used to determine the pest and disease situation for the abnormal local images, and then targeted analysis is performed to avoid repeated traversal and waste of computing power, thereby improving analysis efficiency while ensuring reliability.

[0045] In particular, the present invention only considers chromaticity difference analysis under the implicit regularity category. In actual situations, the texture features in the target area images of this type have poor regularity and weak data representation. After the pest and disease area is preliminarily screened by chromaticity features, a pre-trained model is used to identify the pest and disease characteristics, and then targeted analysis is performed to avoid repeated traversal and waste of computing power, thereby improving analysis efficiency while ensuring reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A schematic diagram of the structure of a smart agriculture system based on big data according to an embodiment of the invention;

[0047] Figure 2 A schematic diagram of the structure of an image extraction module according to an embodiment of the invention;

[0048] Figure 3A logical block diagram of the distribution categories of the division rules of an embodiment of the invention;

[0049] Figure 4 This is a logic block diagram of a scheduling unit calling object according to an embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0051] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0052] It should be noted that in the description of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected" and "connected" should be understood in a broad sense, for example, it can be 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, or it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0053] See also Figures 1 to 4 As shown, Figure 1 The structure diagram of the smart agriculture system based on big data of the embodiment of the invention is as follows: Figure 2 This is a schematic diagram of the structure of an image extraction module according to an embodiment of the invention. Figure 3 This is a logic block diagram of the distribution categories of the division rules of the embodiment of the invention. Figure 4 This is a logic block diagram of a dispatch unit calling an object in an embodiment of the invention. The cloud platform-based rural environment supervision system of the present invention includes:

[0054] An inspection module, comprising a mobile inspection unit for moving in a target area and an image acquisition device arranged on the mobile inspection unit for acquiring an image of the target area;

[0055] A regular distribution recognition module, which is connected to the inspection module, is used to recognize specific continuous texture segments based on chromaticity, calculate the distribution regularity characterization coefficient according to the number of specific continuous texture segments and the angle difference between each specific continuous texture segment, and classify the regular distribution category of the target area image;

[0056] An image extraction module, which is connected to the regular distribution recognition module and includes a texture difference analysis unit, a chromaticity difference analysis unit and a scheduling unit,

[0057] The scheduling unit is used to call the texture difference analysis unit and the chromaticity difference analysis unit based on the division result of the regular distribution recognition module;

[0058] The texture difference analysis unit is used to identify abnormal sub-texture segments based on the curvature characteristics of the sub-texture segments and the adjacent sub-texture segments in the continuous texture segments in the target area image, determine the abnormal local image based on the center coordinates of the abnormal sub-texture segments and transmit it to the model analysis module;

[0059] The chromaticity difference analysis unit is used to determine the abnormal local image based on the chromaticity difference between each sub-region and the adjacent sub-region in the target region image and transmit it to the model analysis module;

[0060] A model analysis module, which is connected to the image extraction module and is used to identify the characteristics of pests and diseases in the abnormal local image transmitted by the image extraction module.

[0061] Specifically, there is no limitation on the equipment selection of the mobile inspection unit, which can be a drone, as long as it can perform mobile inspections of farmland. This will not be elaborated here.

[0062] Specifically, there is no limitation on the choice of image acquisition equipment, which can be a high-speed camera as long as it can capture images of farmland. In particular, there is no limitation on the location of the image acquisition equipment, and the image acquisition equipment can be set at any part of the mobile inspection unit. Relevant personnel in this field can make adjustments based on needs and actual conditions, which will not be elaborated here.

[0063] Specifically, there is no limitation on the specific structures of the regular distribution recognition module, the image extraction module and the model analysis module, and they themselves or each unit therein can be composed of logical components, and the logical components include a field programmable processor, a computer or a microprocessor in a computer.

[0064] Specifically, there is no limitation on the way in which the model analysis module identifies the characteristics of pests and diseases. A pre-trained model can be built into the model analysis module. For model training, a deep learning model suitable for image recognition tasks can be selected, such as convolutional neural network (CNN), especially YOLO (You On ly Look Once) series algorithms. In addition, a number of images containing pest and disease characteristics are constructed as training samples, and the selected deep learning model is trained until the model performance is qualified to realize the relevant functions of identifying pest and disease characteristics. This will not be repeated.

[0065] Of course, the existing open source pest and disease identification model can also be used to identify the pest and disease characteristics in the image, which will not be elaborated here.

[0066] The regular distribution recognition module is used to recognize specific continuous texture segments based on chromaticity, including:

[0067] Used to identify all continuous texture segments in the target area image;

[0068] to extract the average chromaticity of all continuous texture segments, and respectively calculate the difference between the chromaticity corresponding to each continuous texture segment and the average chromaticity;

[0069] for identifying a continuous texture segment that meets the identification conditions as a specific continuous texture segment;

[0070] The recognition condition includes that the difference value corresponding to the continuous texture segment is greater than a predetermined difference threshold.

[0071] Specifically, when there are multiple angles between specific continuous texture segments in an image, the angle difference between the specific continuous texture segments can be the average value of the multiple angle differences, as long as it can represent the angle difference of the target area image. The larger the angle difference, the more irregular the specific continuous texture segments in the target area image.

[0072] Specifically, the predetermined difference threshold G0 is determined according to the average chromaticity value G, and G=0.3×G0 is set.

[0073] Specifically, the regular distribution recognition module calculates the distribution regularity characterization coefficient according to formula (1):

[0074]

[0075] In formula (1), D represents the distribution law characterization coefficient, M represents the number of specific continuous texture segments, M0 represents the number of reference specific continuous texture segments, A represents the angle difference value between specific continuous texture segments, A0 represents the angle difference value between reference specific continuous texture segments, α represents the influence coefficient of the number, and β represents the influence coefficient of the angle difference value.

[0076] The number of reference specific continuous texture segments is obtained by pre-statistics. Technicians in related fields can screen several different target area images, determine the mean number ΔM of specific continuous texture segments, and set M0=m×ΔM, where m represents the precision coefficient, 1.2<m<1.4.

[0077] The angle difference value between the reference specific continuous texture segments is obtained by pre-statistical analysis. Technicians in related fields can screen several different target area images, determine the angle difference value ΔA between the specific continuous texture segments, and set M0 = ΔA × g, where g represents the offset coefficient, 1.15 < g < 1.3.

[0078] In this embodiment, α is 0.61 and β is 0.39.

[0079] Specifically, the present invention recognizes massive images based on chromaticity. In actual situations, all images are usually transmitted to a computer for characteristic analysis of farmland pests and diseases. However, for large areas of farmland, there will be many pictures without abnormal pests and diseases. The computer uses a pre-trained model to perform a comprehensive analysis of these images, resulting in a waste of computing power. Therefore, the present invention considers calculating the distribution law characterization coefficient, identifying the target area images with more obvious texture patterns, and then dividing the regular distribution categories. Different analysis methods are used for targeted analysis of target area images of different categories to avoid repeated traversal and waste of computing power, thereby improving analysis efficiency while ensuring reliability.

[0080] Specifically, the regular distribution recognition module determines the regular distribution category of the target area image based on the distribution regularity characterization coefficient, wherein:

[0081] If the distribution law representation coefficient is greater than or equal to the reference distribution law representation coefficient, the target area image is determined to be a dominant law category;

[0082] If the distribution law representation coefficient is less than the reference distribution law representation coefficient, the target area image is determined to be a hidden law category.

[0083] Specifically, the benchmark distribution law characterization coefficient D0 is selected in the interval [0.95, 1.15].

[0084] Specifically, the scheduling unit calls the texture difference analysis unit and the chromaticity difference analysis unit based on the division result of the regular distribution recognition module, including:

[0085] If the target area image is of the dominant regularity category, calling the texture difference analysis unit and the chromaticity difference analysis unit to analyze the target area image;

[0086] If the target area image is of an implicit regularity category, a chromaticity difference analysis unit is called to analyze the target area image.

[0087] Preferably, the present invention considers using texture difference analysis and chromaticity difference analysis under the explicit rule category, preliminarily identifies abnormal local images that may contain pests and diseases based on texture features, and comprehensively determines and identifies abnormal local images in the target area in combination with chromaticity analysis. Only the pre-trained model is used to determine the pest and disease situation for the abnormal local images. Under the implicit rule category, only chromaticity difference analysis is considered, and the chromaticity features are used to preliminarily screen the pest and disease areas, and then the pre-trained model is used to identify the pest and disease characteristics, thereby performing targeted analysis to avoid repeated traversal and waste of computing power, thereby improving analysis efficiency while ensuring reliability.

[0088] Specifically, the texture difference analysis unit is used to identify abnormal sub-texture segments, including:

[0089] for dividing the continuous texture segment in the target area image into a plurality of sub-texture segments;

[0090] It is used to calculate the difference between the curvature features of the sub-texture segment and the adjacent sub-texture segment, and identify the sub-texture segment whose difference is greater than a predetermined curvature difference threshold as an abnormal sub-texture segment.

[0091] It can be understood that the curvature feature is the average curvature of the sub-texture segment, the difference threshold is determined based on the average curvature S0 of the adjacent sub-texture segment, and the difference threshold is set to be selected within the interval [0.2S0, 0.4S0].

[0092] Specifically, the texture difference analysis unit determines the abnormal local image based on the center coordinates of the abnormal sub-texture segment, including:

[0093] To obtain the center coordinates of the abnormal sub-texture segment and set it as the center of the circle;

[0094] to obtain the length of the abnormal sub-texture segment and determine the radius according to the length;

[0095] A circle is selected based on the center and radius, and an image in the circle is determined as an abnormal local image.

[0096] Preferably, in the embodiment, the radius length may be set to 0.7 times the length of the abnormal sub-texture segment, and those skilled in the art may also adjust the radius length according to actual conditions.

[0097] Specifically, there may be multiple abnormal sub-texture segments in the target area image. It is necessary to determine the center coordinates and abnormal local image of each abnormal sub-texture segment for pest and disease feature recognition to ensure the integrity and efficiency of regional pest and disease identification.

[0098] Specifically, the chromaticity difference analysis unit is used to determine the abnormal local image including:

[0099] for identifying the chromaticity difference between each sub-region and an adjacent sub-region in the target region image;

[0100] Used to determine the chromaticity difference boundary point;

[0101] Used to connect the boundary points to determine the closed area;

[0102] It is used to determine abnormal local images from the images within the closed area.

[0103] The chromaticity difference boundary point is a midpoint of a local image whose chromaticity difference with an adjacent local image is greater than a predetermined chromaticity difference threshold.

[0104] The chromaticity difference threshold Se0 is determined based on the chromaticity mean value Se of the adjacent local images, and Se0=Se×s is set, where s represents the difference ratio parameter, and 0.25<s<0.5.

[0105] Specifically, the chromaticity difference between each local image and the adjacent local image is determined by a computer. The determination method is not limited and can be based on the CIE1976L*a*b* color difference formula and the CMC(l:c) color difference formula. This is an existing method and will not be repeated here.

[0106] Specifically, it also includes an abnormal storage module, which is connected to the image extraction module and is used to store the abnormal local image identified by the image extraction module and the corresponding target area image.

[0107] It is understandable that there is no limitation on the storage method of the storage module, and it can be hard disk storage, cloud storage, etc. This is the existing technology and will not be elaborated.

[0108] Specifically, it also includes a display module for displaying the data extracted by the image extraction module and the characteristics of pests and diseases.

[0109] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. A smart agriculture system based on big data, characterized in that: include: An inspection module, comprising a mobile inspection unit for moving in a target area and an image acquisition device arranged on the mobile inspection unit for acquiring an image of the target area; A regular distribution recognition module, which is connected to the inspection module, is used to recognize specific continuous texture segments based on chromaticity, calculate the distribution regularity characterization coefficient according to the number of specific continuous texture segments and the angle difference between each specific continuous texture segment, and classify the regular distribution category of the target area image; An image extraction module, which is connected to the regular distribution recognition module and includes a texture difference analysis unit, a chromaticity difference analysis unit and a scheduling unit, The scheduling unit is used to call the texture difference analysis unit and the chromaticity difference analysis unit based on the division result of the regular distribution recognition module; The texture difference analysis unit is used to identify abnormal sub-texture segments based on the curvature characteristics of the sub-texture segments and the adjacent sub-texture segments in the continuous texture segments in the target area image, determine the abnormal local image based on the center coordinates of the abnormal sub-texture segments and transmit it to the model analysis module; The chromaticity difference analysis unit is used to determine the abnormal local image based on the chromaticity difference between each sub-region and the adjacent sub-region in the target region image and transmit it to the model analysis module; A model analysis module, which is connected to the image extraction module and is used to identify the features of pests and diseases in the abnormal local image transmitted by the image extraction module; The regular distribution recognition module is used to recognize specific continuous texture segments based on chromaticity, including: Used to identify all continuous texture segments in the target area image; To determine the chromaticity of all continuous texture segments, and calculate the difference between the chromaticity corresponding to each continuous texture segment and the average chromaticity of the target area image; for identifying a continuous texture segment that meets the identification conditions as a specific continuous texture segment; The recognition condition includes that the difference value corresponding to the continuous texture segment is greater than a predetermined difference threshold.

2. The smart agriculture system based on big data according to claim 1, characterized in that: The regular distribution recognition module calculates the distribution regularity characterization coefficient according to formula (1): In formula (1), D represents the distribution law characterization coefficient, M represents the number of specific continuous texture segments, M0 represents the number of reference specific continuous texture segments, A represents the angle difference value between each specific continuous texture segment, A0 represents the angle difference value between reference specific continuous texture segments, α represents the influence coefficient of the number, and β represents the influence coefficient of the angle difference value.

3. The smart agriculture system based on big data according to claim 1 is characterized in that: The regular distribution recognition module determines the regular distribution category of the target area image based on the distribution regularity characterization coefficient, wherein: If the distribution law representation coefficient is greater than or equal to the reference distribution law representation coefficient, the target area image is determined to be a dominant law category; If the distribution law representation coefficient is less than the reference distribution law representation coefficient, the target area image is determined to be a hidden law category.

4. The smart agriculture system based on big data according to claim 3 is characterized in that: The scheduling unit calls the texture difference analysis unit and the chromaticity difference analysis unit based on the division result of the regular distribution recognition module, including: If the target area image is of the dominant regularity category, calling the texture difference analysis unit and the chromaticity difference analysis unit to analyze the target area image; If the target area image is of an implicit regularity category, a chromaticity difference analysis unit is called to analyze the target area image.

5. The smart agriculture system based on big data according to claim 1, characterized in that: The texture difference analysis unit is used to identify abnormal sub-texture segments, including: for dividing the continuous texture segment in the target area image into a plurality of sub-texture segments; It is used to calculate the difference between the curvature features of the sub-texture segment and the adjacent sub-texture segment, and identify the sub-texture segment whose difference is greater than a predetermined curvature difference threshold as an abnormal sub-texture segment.

6. The smart agriculture system based on big data according to claim 5 is characterized in that: The texture difference analysis unit determines an abnormal local image based on the center coordinates of the abnormal sub-texture segment, including: To obtain the center coordinates of the abnormal sub-texture segment and set it as the center of the circle; to obtain the length of the abnormal sub-texture segment and determine the radius according to the length; A circle is selected based on the center and radius, and an image in the circle is determined as an abnormal local image.

7. The smart agriculture system based on big data according to claim 1, characterized in that: The chromaticity difference analysis unit is used to determine abnormal local images including: for identifying the chromaticity difference between each sub-region and an adjacent sub-region in the target region image; Used to determine the chromaticity difference boundary point; Used to connect the boundary points to determine the closed area; The image within the closed area is used to determine an abnormal local image.

8. The smart agriculture system based on big data according to claim 1, characterized in that: It also includes an abnormal storage module, which is connected to the image extraction module and is used to store the abnormal local image identified by the image extraction module and the corresponding target area image.

9. The smart agriculture system based on big data according to claim 1, characterized in that: It also includes a display module for displaying the data extracted by the image extraction module.

Images