Pest monitoring system for agriculture

Abstract

The invention discloses a pest monitoring system for agriculture. The pest monitoring system comprises a data acquisition module, a data communication transmission module, a data processing module and a data monitoring module, the data acquisition module is used for acquiring crop images; the data communication transmission module is used for transmitting the crop image to the data processing module; the data processing module is used for judging the crop image to obtain the category of the pest and sending the category information of the pest to the data monitoring module; and the data monitoring module is used for obtaining the number of the pests of each category based on the category information of the pests so as to complete monitoring of the pests. According to the invention, automatic monitoring and early warning of pests are realized, all-weather real-time monitoring is realized, the monitoring cost is effectively reduced, and the monitoring efficiency is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of pest risk analysis and identification, and in particular relates to a pest monitoring system for agriculture. Background Art

[0002] The healthy and stable development of agriculture is related to people's livelihood. Therefore, any country has invested a lot of manpower, material and financial resources in agricultural production to continuously inject momentum into agricultural development.

[0003] In agricultural production, pest control is a significant factor influencing crop quality and yield. Accurately identifying and counting pests is paramount for pest forecasting. Traditional pest identification and counting methods rely on manual identification, field surveys, and trapping. However, due to complex and unstable field environments, manual identification and counting suffer from low recognition rates, poor technical accuracy, high labor intensity, and non-real-time performance. These methods no longer meet the current monitoring requirements for serious pest outbreaks in farmland. Field surveys are time-consuming and labor-intensive, and the data collection, recording, and reporting process involves numerous steps, resulting in a high workload for monitoring personnel and significant subjective influences. The timeliness of data application also hinders accurate pest forecasting and falls short of practical needs. With the development of computer technology, microelectronics technology, and the internet, their applications in agriculture are increasing, bringing numerous benefits to agricultural development. Therefore, a pest monitoring system that operates 24 / 7, offers high stability, provides real-time online monitoring, and is easily shareable is urgently needed. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention proposes a pest monitoring system for agriculture, which realizes automatic monitoring and early warning of pests, realizes all-weather real-time monitoring, effectively reduces monitoring costs and improves monitoring efficiency.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A pest monitoring system for agriculture, comprising: a data acquisition module, a data communication transmission module, a data processing module and a data monitoring module;

[0007] The data acquisition module is used to collect crop images;

[0008] The data communication transmission module is used to transmit the crop image to the data processing module;

[0009] The data processing module is used to judge the crop image, obtain the category of the pest, and send the category information of the pest to the data monitoring module;

[0010] The data monitoring module is used to obtain the number of pests of each category based on the category information of the pests and complete the monitoring of the pests.

[0011] Preferably, the data acquisition module adopts a vertical pole and is distributed in a matrix in the farmland, and includes: a camera, a photovoltaic panel, a battery and a communication module;

[0012] The camera is used to capture images of crops;

[0013] The photovoltaic panel is used to provide power for the monitoring system;

[0014] The battery is used to store electrical energy.

[0015] Preferably, the data communication transmission module is used to transmit the captured crop images to the data processing module using a wifi module.

[0016] Preferably, the data processing module includes: an image pre-processing unit and a recognition unit;

[0017] The image preprocessing unit is used to preprocess the crop image to obtain a preprocessed image;

[0018] The recognition unit is used to use a deep learning model to preliminarily determine the category of the pest based on the preprocessed image.

[0019] Preferably, the image preprocessing unit includes: a shadow elimination subunit, a color space conversion subunit, a leaf region segmentation subunit, a leaf region RGB extraction subunit and an image size adjustment subunit;

[0020] The shadow removal subunit uses a brightness elevation model to perform shadow area division and brightness equalization on the RGB image of the crop image to obtain a shadow-free RGB image of the crop;

[0021] The color space conversion subunit is used to convert the shadow-free crop RGB image into an HSV image;

[0022] The leaf region segmentation subunit is configured to perform leaf region segmentation based on different components of the HSV image to obtain a leaf region image;

[0023] The leaf area RGB extraction subunit is used to extract the RGB image of the leaf area to obtain the leaf area RGB image;

[0024] The image size adjustment unit is used to unify the size of the leaf area RGB image to obtain a preprocessed image.

[0025] Preferably, the recognition unit includes: a feature extraction subunit, a training subunit and an identification subunit;

[0026] The feature extraction subunit is used to extract multiple types of features of harmful organisms in crops based on image processing technology;

[0027] The training subunit is used to fuse the multiple types of features and use them as input data for training and detection of the BP neural network;

[0028] The identification subunit is used to identify new harmful organisms using the trained BP network classifier;

[0029] The multi-category features include three types of features: color, shape and texture; the BP neural network is regarded as a nonlinear mapping from the multi-category feature parameters of the input pests to the output corresponding pest categories.

[0030] Preferably, the data monitoring module includes: an alarm unit and a prediction unit;

[0031] The prediction unit is used to calculate the proportion of each type of pest according to the category of pests in each crop image, obtain the number of pests of each category based on the proportion, and predict the future trend of pests and diseases based on the number;

[0032] The alarm unit is used to forecast pests and diseases after trend prediction.

[0033] Preferably, the prediction unit includes: a model building subunit, a database subunit, a statistics subunit and a trend prediction subunit;

[0034] The model building subunit is used to build an LSTM neural network;

[0035] The database subunit is used to store the category of harmful organisms in each crop image;

[0036] The statistical subunit is used to count the proportion of each type of pests according to the category of the pests in each crop image, and obtain the number of pests of each category based on the proportion;

[0037] The trend prediction subunit is used to train the LSTM neural network according to the number of pests of each category, and predict the future trend of pests and diseases based on the trained LSTM neural network.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] The present invention uses the number of various types of harmful organisms counted in real time over a period of time as training data to train the network model in the trend prediction subunit, and uses the trained network model to predict the future development trend of various types of pests and diseases, thereby predicting the risk of pests and diseases in advance according to the number limit values of various types of harmful organisms and issuing an alarm, so that after obtaining the harmful organism information, a corresponding prevention and control plan can be formulated, and the prevention and control plan can be sent to the executor. After the executor executes the prevention and control plan, the harmful organism information is collected again, and the prevention and control effect is evaluated based on the collected harmful organism information. This allows a subsequent supervision process to be set up after the prevention and control plan is issued, which not only supervises the work of the executor, but also enables timely understanding of the effect of the prevention and control plan so as to adjust the prevention and control plan in a timely manner. The present invention realizes all-weather real-time monitoring, effectively reduces monitoring costs, improves monitoring efficiency, and has a wide range of promotion and application value in the field of agricultural technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0041] Figure 1 The figure is an overall framework diagram of a pest monitoring system for agriculture in an embodiment of the present invention. DETAILED DESCRIPTION

[0042] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0043] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0044] Example 1

[0045] like Figure 1 As shown, a pest monitoring system for agriculture includes: a data acquisition module, a data communication transmission module, a data processing module and a data monitoring module;

[0046] The data acquisition module is used to collect crop images;

[0047] The data communication transmission module is used to transmit the crop images to the data processing module;

[0048] The data processing module is used to judge the crop images, obtain the category of the pests, and send the category information of the pests to the data monitoring module;

[0049] The data monitoring module is used to obtain the number of pests of each category based on the category information of the pests and complete the monitoring of the pests.

[0050] In this embodiment, the data acquisition module adopts poles and is distributed in a matrix on the farmland, including: cameras, photovoltaic panels, batteries and communication modules;

[0051] The camera is used to capture images of crops;

[0052] Photovoltaic panels are used to provide power for the monitoring system;

[0053] Batteries are used to store electrical energy.

[0054] In this embodiment, the data communication transmission module is used to transmit the captured crop images to the data processing module using the WiFi module.

[0055] In this embodiment, the data processing module includes: an image pre-processing unit and a recognition unit;

[0056] The image preprocessing unit is used to preprocess the crop image to obtain a preprocessed image;

[0057] The recognition unit is used to use a deep learning model to preliminarily determine the category of pests based on preprocessed images.

[0058] In this embodiment, the image preprocessing unit includes: a shadow removal subunit, a color space conversion subunit, a leaf region segmentation subunit, a leaf region RGB extraction subunit, and an image size adjustment subunit;

[0059] The shadow elimination sub-unit uses the brightness elevation model to divide the shadow area of the RGB image of the crop image and balance the brightness to obtain a shadow-free RGB image of the crop;

[0060] The color space conversion subunit is used to convert the shadow-free crop RGB image into an HSV image;

[0061] The leaf region segmentation subunit is used to segment the leaf region based on different components of the HSV image to obtain a leaf region image;

[0062] The leaf area RGB extraction subunit is used to extract the RGB image of the leaf area to obtain the leaf area RGB image;

[0063] The image size adjustment unit is used to unify the size of the RGB image of the leaf area to obtain a preprocessed image.

[0064] In this embodiment, the recognition unit includes: a feature extraction subunit, a training subunit and an identification subunit;

[0065] The feature extraction subunit is used to extract multiple types of features of harmful organisms in crops based on image processing technology;

[0066] The training subunit is used to fuse multiple types of features and use them as input data for BP neural network training and detection;

[0067] The identification subunit is used to identify new pests using the trained BP network classifier;

[0068] The multi-category features include three types of features: color, shape and texture; the BP neural network is regarded as a nonlinear mapping from the multi-category feature parameters of the input pests to the output corresponding pest categories.

[0069] Specifically, multi-feature extraction includes extracting three types of features of pests, namely color, morphology and texture through preprocessing images; color features have the advantages of rotation, scale and translation invariance; morphological features include shape parameters of pests, which are related to the human visual perception system; texture features are regional features that reflect the spatial distribution of pixels, and are also another important feature that supplements the color and morphological features of pests; color feature extraction is to extract the first-order moment, second-order moment and third-order moment of the R, G, B, H, S color components under the RGB and HSV color space models of the image; morphological feature extraction includes the extraction of basic shape parameters such as area, perimeter, length and width; and the extraction of dimensionless geometric morphological features such as dispersion, density, aspect ratio, circularity and elongation calculated through basic shape parameters; texture feature extraction, based on the gray-level co-occurrence matrix, extracts contrast, energy, homogeneity, correlation, mean, standard deviation, smoothness, third-order moment, consistency and texture features.

[0070] The specific steps of BP neural network design include: 1) determining the number of input nodes; 2) determining the number of hidden layer nodes; 3) determining the number of output nodes; 4) selecting the learning rate; 5) selecting the expected error.

[0071] BP network classifier training, randomly extracting 80% of the pre-processed image samples as training samples, and training the BP network classifier; using normalization processing when inputting data to speed up the convergence speed of BP neural network learning.

[0072] In this embodiment, the data monitoring module includes: an alarm unit and a prediction unit;

[0073] The prediction unit is used to calculate the proportion of each type of pest according to the category of pests in each crop image, obtain the number of pests of each category based on the proportion, and predict the future trend of pests based on the number;

[0074] The alarm unit is used to forecast pests and diseases after trend prediction.

[0075] In this embodiment, the prediction unit includes: a model building subunit, a database subunit, a statistics subunit and a trend prediction subunit;

[0076] The model building subunit is used to build an LSTM neural network;

[0077] The database subunit is used to store the category of pests in each crop image;

[0078] The statistical subunit is used to count the proportion of each type of pests according to the category of the pests in each crop image, and obtain the number of pests of each category based on the proportion;

[0079] The trend prediction subunit is used to train the LSTM neural network according to the number of each category of pests, and predict the future trend of pests and diseases based on the trained LSTM neural network.

[0080] Specifically, the trend prediction subunit can predict pest and disease trends in the short, medium, or long term. Short-term refers to 4-48 hours, medium-term refers to 2-7 days, and long-term refers to 7-15 days. Medium-term and long-term trend predictions are not limited to the aforementioned ranges; they can be set to larger ranges. However, the longer the forecast timeframe, the greater the corresponding error.

[0081] An LSTM neural network consists of an input layer, hidden layers, and an output layer. The input layer receives training data; the hidden layers iteratively learn the short-range and long-range semantic features of time series data; and the output layer outputs predictions. LSTM network parameters include the learning rate, number of iterations, and stepsize. The key network parameter, stepsize, ranges from 1 to 24, depending on the environment, the size of the training data, and actual conditions and requirements.

[0082] When stepsize=1, the labeling processing method is to use the training data at the n+xth moment as the label of the training data at the nth moment; when stepsize=2, the labeling processing method is to use the training data at the n+xth moment as the label of the training data at the nth and n-1th moments; when stepsize=3, the labeling processing method is to use the training data at the n+xth moment as the label of the environmental training data at the nth, n-1 and n-2th moments, and so on. Where x is the prediction step size parameter, which can be any integer greater than or equal to 0.

[0083] The value of the prediction step parameter x is related to short-term, medium-term and long-term predictions. If it is a short-term prediction, the value of x should be smaller, if it is a long-term prediction, the value of x should be larger, and if it is a medium-term prediction, the value of x should be between the above two. By adjusting the prediction step parameter x, the development trend of the predicted pests and diseases corresponding to the near or far future moment can be obtained. For example, when the value of x is 3 (the time span between the nth and n+1th moments is set to 4 hours), the trend prediction model can obtain the development trend of the predicted pests and diseases 12 hours later; when the value of x is 24, the trend prediction model can obtain the development trend of the predicted pests and diseases 96 hours later (that is, 4 days later); when the value of x is 72, the trend prediction model can obtain the development trend of the predicted pests and diseases 12 days later.

[0084] The present invention uses the number of various types of harmful organisms counted in real time over a period of time as training data to train the network model in the trend prediction subunit, and uses the trained network model to predict the future development trend of various types of pests and diseases, thereby predicting the risk of pests and diseases in advance according to the number limit values of various types of harmful organisms and issuing an alarm, so that after obtaining the harmful organism information, a corresponding prevention and control plan can be formulated, and the prevention and control plan can be sent to the executor. After the executor executes the prevention and control plan, the harmful organism information is collected again, and the prevention and control effect is evaluated based on the collected harmful organism information. This allows a subsequent supervision process to be set up after the prevention and control plan is issued, which not only supervises the work of the executor, but also enables timely understanding of the effect of the prevention and control plan so as to adjust the prevention and control plan in a timely manner. The present invention realizes all-weather real-time monitoring, effectively reduces monitoring costs, improves monitoring efficiency, and has a wide range of promotion and application value in the field of agricultural technology.

[0085] The embodiments described above are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by persons skilled in the art should fall within the scope of protection defined by the claims of the present invention.

Claims

1. A pest monitoring system for agriculture, characterized in that: include: Data acquisition module, data communication transmission module, data processing module and data monitoring module; The data acquisition module is used to collect crop images; The data communication transmission module is used to transmit the crop image to the data processing module; The data processing module is used to judge the crop image, obtain the category of the pest, and send the category information of the pest to the data monitoring module; The data monitoring module is used to obtain the number of pests of each category based on the category information of the pests and complete the monitoring of the pests.

2. The agricultural pest monitoring system according to claim 1, characterized in that: The data acquisition module adopts a pole and is distributed in a matrix on the farmland, including: a camera, a photovoltaic panel, a battery and a communication module; The camera is used to capture images of crops; The photovoltaic panel is used to provide power for the monitoring system; The battery is used to store electrical energy.

3. The agricultural pest monitoring system according to claim 1, characterized in that: The data communication transmission module is used to transmit the captured crop images to the data processing module using a wifi module.

4. The agricultural pest monitoring system according to claim 1, characterized in that: The data processing module includes: an image preprocessing unit and a recognition unit; The image preprocessing unit is used to preprocess the crop image to obtain a preprocessed image; The recognition unit is used to use a deep learning model to preliminarily determine the category of the pest based on the preprocessed image.

5. The agricultural pest monitoring system according to claim 4, characterized in that: The image preprocessing unit includes: a shadow elimination subunit, a color space conversion subunit, a leaf area segmentation subunit, a leaf area RGB extraction subunit and an image size adjustment subunit; The shadow removal subunit uses a brightness elevation model to perform shadow area division and brightness equalization on the RGB image of the crop image to obtain a shadow-free RGB image of the crop; The color space conversion subunit is used to convert the shadow-free crop RGB image into an HSV image; The leaf region segmentation subunit is configured to perform leaf region segmentation based on different components of the HSV image to obtain a leaf region image; The leaf area RGB extraction subunit is used to extract the RGB image of the leaf area to obtain the leaf area RGB image; The image size adjustment unit is used to unify the size of the leaf area RGB image to obtain a preprocessed image.

6. The agricultural pest monitoring system according to claim 4, characterized in that: The recognition unit includes: a feature extraction subunit, a training subunit and an identification subunit; The feature extraction subunit is used to extract multiple types of features of harmful organisms in crops based on image processing technology; The training subunit is used to fuse the multiple types of features and use them as input data for training and detection of the BP neural network; The identification subunit is used to identify new harmful organisms using the trained BP network classifier; The multi-category features include three types of features: color, shape and texture; the BP neural network is regarded as a nonlinear mapping from the multi-category feature parameters of the input pests to the output corresponding pest categories.

7. The agricultural pest monitoring system according to claim 1, characterized in that: The data monitoring module includes: an alarm unit and a prediction unit; The prediction unit is used to calculate the proportion of each type of pest according to the category of the pests in each crop image, obtain the number of pests of each category based on the proportion, and predict the future trend of pests and diseases based on the number; The alarm unit is used to forecast pests and diseases after trend prediction.

8. The agricultural pest monitoring system according to claim 7, characterized in that: The prediction unit includes: a model building subunit, a database subunit, a statistics subunit and a trend prediction subunit; The model building subunit is used to build an LSTM neural network; The database subunit is used to store the category of harmful organisms in each crop image; The statistical subunit is used to count the proportion of each type of pests according to the category of the pests in each crop image, and obtain the number of pests of each category based on the proportion; The trend prediction subunit is used to train the LSTM neural network according to the number of pests of each category, and predict the future trend of pests and diseases based on the trained LSTM neural network.

Images