A mulberry tree growth monitoring system and method based on image processing

Abstract

The present invention relates to a mulberry tree growth monitoring system and method based on image processing, which comprises the following steps: acquiring data of each growth stage of mulberry trees, constructing mulberry tree data samples and marking them, inputting them into a mulberry tree growth analysis model for training; acquiring real-time images of mulberry trees and transmitting them to the mulberry tree growth analysis model, and outputting growth analysis results of mulberry trees; matching optimal growth curves, environmental parameters and fertilization data, controlling environmental parameters, and controlling fertilization time, type and amount by a fertilization control module; acquiring images at a specified time point and transmitting them to a mulberry tree growth analysis model for analysis and drawing an actual growth curve of the mulberry tree, adjusting environmental parameters, fertilization time and fertilization amount according to specified rules; and picking mulberry leaves based on the corresponding mulberry leaf maturity data acquired at a specified time point. The system realizes growth monitoring of mulberry trees and acquisition of mulberry leaf maturity, provides a data basis for environmental parameter and fertilization control, and picks mulberry leaves of corresponding maturity for silkworms of corresponding age to eat.

Description

Technical Field

[0001] The invention belongs to the technical field of image processing, and in particular relates to a mulberry tree growth monitoring system and method based on image processing. Background Art

[0002] The mulberry tree is a deciduous tree or shrub that has many uses. It can be used as livestock feed or made into products such as mulberry leaf tea. The fruit of the mulberry tree is called mulberry, which can be eaten raw or used to make juice, wine, etc. In addition, the leaves, roots, fruits and branches of the mulberry tree can be used as medicine. The bark fiber is fine and can be used as a raw material for textiles and papermaking. The wood is hard and suitable for making furniture, musical instruments and carvings. my country is the largest mulberry planting country. Mulberry leaves occupy a major position in the sericulture industry, so the most important use of mulberry leaves is as the main feed for silkworms.

[0003] However, in the process of silkworm rearing, silkworms only need to feed on mulberry leaves in the larval stage. However, silkworms are also divided into multiple instars in the larval stage. The larvae will undergo four moltings, and each time they molt, they will enter the next instar (1st instar, 2nd instar, 3rd instar, 4th instar), and finally become 5th instar larvae. In the larval stage of silkworms, silkworms of different instars need to eat mulberry leaves of different maturity due to the different development of their mouthparts. Now, silkworm breeders judge the growth degree of mulberry leaves based on experience, and then pick mulberry leaves of corresponding maturity for silkworms to feed. On the one hand, this method is likely to cause silkworms of different ages to be unable to eat mulberry leaves of matching maturity, which will lead to the inability of silkworms to grow normally and quickly. On the other hand, in the growth management process of mulberry trees, different environmental parameters and fertilizer application amounts of mulberry trees at different growth stages can enable mulberry trees to grow rapidly at each stage and increase the yield of mulberry leaves.

[0004] Therefore, how to monitor the growth of mulberry trees, provide a data basis for the control of environmental parameters and fertilization, and provide the maturity of mulberry leaves through mulberry tree growth monitoring, so as to pick mulberry leaves of corresponding maturity for silkworms of corresponding age to eat and promote the rapid growth of silkworms, is a technical problem that urgently needs to be solved. Summary of the invention

[0005] The purpose of the present invention is to provide a mulberry tree growth monitoring system and method based on image processing, so as to realize the growth monitoring of mulberry trees, provide a data basis for the control of environmental parameters and fertilization, and provide the maturity of mulberry leaves through the growth monitoring of mulberry trees, so as to pick mulberry leaves of corresponding maturity for silkworms of corresponding age to eat, thereby promoting the rapid growth of silkworms.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is as follows:

[0007] In a first aspect, a mulberry tree growth monitoring method based on image processing is provided, comprising the following steps:

[0008] S1: Based on big data, image data of each growth stage of mulberry trees, optimal growth curves, and corresponding environmental parameters and fertilization data are obtained, and data preprocessing is performed on the obtained data, and mulberry tree data samples are constructed based on the preprocessed data;

[0009] S2: labeling the mulberry tree data samples and creating a mulberry tree growth analysis model, inputting the labeled mulberry tree data samples into the mulberry tree growth analysis model to train the model, performing performance verification on the trained mulberry tree growth analysis model, and judging whether the verification result meets the specified requirements, if so, executing step S3, if not, performing parameter optimization on the trained mulberry tree growth analysis model until the model meets the specified requirements;

[0010] S3: performing real-time image acquisition of the mulberry tree through the image acquisition module, transmitting the acquired image data to the mulberry tree growth analysis model for analysis, and outputting the growth analysis result of the mulberry tree, wherein the growth analysis result includes the current growth stage of the mulberry tree and the corresponding maturity of each leaf of the current mulberry tree;

[0011] S4: The mulberry tree growth analysis model matches the optimal growth curve, environmental parameters and fertilization data according to the current growth stage of the mulberry tree, and transmits the environmental parameters and fertilization data to the environmental parameter control module and the fertilization control module, wherein the environmental parameter control module controls the environmental parameters, and the fertilization control module controls the fertilization time, type and amount of fertilization;

[0012] S5: collecting images of the mulberry tree at a specified time point, transmitting the collected image data to the mulberry tree growth analysis model for analysis and obtaining the maturity of the mulberry leaves, and drawing an actual growth curve of the mulberry tree, comparing the actual growth curve of the mulberry tree with the optimal growth curve of the mulberry tree, and judging whether the difference between the actual growth curve of the mulberry tree and the optimal growth curve is within a preset range, if not, executing step S6, and if yes, executing step S7;

[0013] S6: adjusting the environmental parameters, fertilization time and fertilization amount in step S4 according to the specified rules, and executing through the environmental parameter control module and the fertilization control module based on the adjusted environmental parameters, fertilization time and fertilization amount;

[0014] S7: picking mulberry leaves based on the corresponding mulberry leaf maturity data obtained at the specified time point, and storing them in corresponding positions according to the maturity of the mulberry leaves.

[0015] Preferably, the specific process of data preprocessing in step S1 is as follows:

[0016] S11: removing duplicate data, noise data and invalid data from the acquired data;

[0017] S12: Convert the color image in the image data in each item of data into a grayscale image, and perform geometric transformation on the grayscale image.

[0018] Preferably, the specific process of the image geometric transformation in step S12 is:

[0019] S121: performing scaling processing on all images so that all images are scaled to a uniform specified size;

[0020] S122: All images are translated to a specified position, and all images are rotated to a specified direction by image rotation.

[0021] Preferably, the image scaling process in step S121 adopts grayscale image interpolation processing, and the specific process is as follows:

[0022] S1211: Mapping each pixel point in the target image to be interpolated to a corresponding position in the original image based on the scaling ratio specified by the image;

[0023] S1212: determining in the original image a specified number of neighboring pixel points corresponding to each pixel point in the target image, wherein the specified number of neighboring pixel points are located around the target pixel point in the original image;

[0024] S1213: Calculate the pixel value of the target position by setting a corresponding weight coefficient according to the determined designated number of neighboring pixel points. The calculation formula is as follows:

[0025] P( x ,y)=S( j , k )* w 1+ S( j , k+ 1)* w 2 + S( j+ 1, k +1)* w 3+ S( j+ 1, k )* w 4;

[0026] Among them, P( x , y) is the target position ( x , y) pixel value, S( j , k )、S( j , k+ 1) S( j+ 1, k +1)、S( j+ 1, k ) are the target positions ( x, y) neighboring pixel values ​​in four directions, w 1. w 2. w 3. w 4 are weight coefficients corresponding to the neighboring points in the four directions respectively.

[0027] Preferably, the specific process of labeling the mulberry tree data sample in step S2 is as follows:

[0028] S21: performing preliminary image segmentation on the image data of the mulberry tree data to segment the leaf area and the branch area in the image;

[0029] S22: extracting features from the segmented leaf area, extracting color features, size features, and texture features of the leaf, and marking them at corresponding positions;

[0030] S23: Perform feature extraction on the segmented branch area, extract the color features, size features and texture features of the branches, and mark them at corresponding positions.

[0031] Preferably, the mulberry tree growth analysis model in step S2 is a neural network model, including an input layer, a hidden layer and an output layer, the hidden layer is provided with 5 convolutional layers and a pooling layer and 3 fully connected layers, the output layer is a softmax output layer, two adjacent layers are connected by a maxpool layer, and each hidden layer adopts a ReLU activation unit, the convolutional layer is formed by stacking two 3x3 convolutional kernels, and the fully connected layer is a 1x1 convolutional kernel.

[0032] Preferably, the training and verification process of the mulberry tree growth analysis model is as follows:

[0033] S24: First, create a basic model and train the basic model by random initialization;

[0034] S25: Initializing the mulberry tree growth analysis model using the weights of the basic model, and preprocessing the image using a ReLU activation unit using RELU as an activation function;

[0035] S26: randomly select a training scale within a specified range to crop the preprocessed image;

[0036] S27: Use the validation set to evaluate the performance of the mulberry tree growth analysis model and adjust the hyperparameters according to the specified rules.

[0037] Preferably, the specific process of the mulberry tree growth analysis model in step S3 is as follows:

[0038] S31: The convolution layer and pooling layer in the hidden layer of the mulberry tree growth analysis model extract features from the image input by the input layer;

[0039] S32: convert the features output by the convolution layer and the pooling layer into feature vectors for output, and process them using global average pooling;

[0040] S33: construct a parameterized function, input the extracted image features into the parameterized function, calculate the output probability of the input variables through the parameterized function, classify the extracted image features based on the output probability, and obtain the growth stage of the mulberry tree and the maturity of the mulberry leaves.

[0041] In the second aspect, a mulberry tree growth monitoring system based on image processing is provided, including a data acquisition module, a data preprocessing module, a data annotation module, a model creation module, a mulberry tree growth analysis model, a growth curve drawing module, an environmental parameter adjustment module, a fertilization adjustment module and a mulberry leaf picking module.

[0042] The data acquisition module is linked to the data preprocessing module, the data preprocessing module is connected to the data annotation module, the data annotation module and the model creation module are both connected to the mulberry tree growth analysis model, the mulberry tree growth analysis model is connected to the growth curve drawing module, the environmental parameter adjustment module, the fertilization adjustment module and the mulberry leaf picking module.

[0043] The data acquisition module is used to acquire image data of each growth stage of the mulberry tree, the optimal growth curve, and corresponding environmental parameters and fertilization data based on big data, and to perform real-time image acquisition of the mulberry tree through the image acquisition module;

[0044] The data preprocessing module is used to perform data preprocessing on the acquired data and construct a mulberry tree data sample based on the preprocessed data;

[0045] The data labeling module is used to label the mulberry tree data sample;

[0046] The model creation module is used to create a mulberry tree growth analysis model, and input the labeled mulberry tree data samples into the mulberry tree growth analysis model to train the model;

[0047] The mulberry tree growth analysis model is used to analyze based on the collected image data, output the growth analysis results of the mulberry tree, the growth analysis results include the current growth stage of the mulberry tree and the corresponding maturity of each leaf of the current mulberry tree, match the optimal growth curve, environmental parameters and fertilization data according to the current growth stage of the mulberry tree, and transmit the environmental parameters and fertilization data to the environmental parameter control module and the fertilization control module;

[0048] The growth curve drawing module is used to draw the actual growth curve of the mulberry tree, compare the actual growth curve of the mulberry tree with the optimal growth curve of the mulberry tree, and determine whether the difference between the actual growth curve of the mulberry tree and the optimal growth curve is within a preset range;

[0049] The environmental parameter adjustment module is used to perform environmental parameter adjustment control based on the adjusted environmental parameters;

[0050] The fertilization adjustment module is used to execute through the fertilization control module based on the adjusted fertilization time and fertilization amount;

[0051] The mulberry leaf picking module is used to pick mulberry leaves based on the corresponding mulberry leaf maturity data obtained at the specified time point.

[0052] The beneficial effects of the present invention include:

[0053] The mulberry tree growth monitoring system and method based on image processing provided by the present invention obtain various data of various growth stages of mulberry trees to construct mulberry tree data samples and mark them, input them into a mulberry tree growth analysis model for training; perform real-time image acquisition on mulberry trees and transmit them to the mulberry tree growth analysis model, and output the growth analysis results of mulberry trees; match the best growth curve, environmental parameters and fertilization data according to the current growth stage of mulberry trees, control the environmental parameters through the environmental parameter control module, and control the fertilization time, type of fertilization and amount of fertilization by the fertilization control module; acquire images at a specified time point, transmit them to the mulberry tree growth analysis model for analysis, draw the actual growth curve of mulberry trees, and compare them with the best growth curve; and adjust the environmental parameters, fertilization time and fertilization amount according to specified rules; and pick mulberry leaves based on the corresponding mulberry leaf maturity data obtained at a specified time point. The growth monitoring of mulberry trees is realized, and a data basis is provided for the control of environmental parameters and fertilization. The maturity of mulberry leaves is provided through the growth monitoring of mulberry trees, so that mulberry leaves of corresponding maturity can be picked for silkworms of corresponding age to eat, thereby promoting the rapid growth of silkworms.

[0054] First, by removing duplicate data, noise data and invalid data from the acquired data, the color images in the image data of each data are converted into grayscale images, and the grayscale images are scaled to a uniform specified size, all images are translated to a specified position, and all images are rotated to a specified direction through geometric transformation processing, which provides an effective data basis for subsequent model training and can improve the training efficiency of the model.

[0055] Secondly, by mapping each pixel point in the target image to be interpolated to the corresponding position in the original image based on the scaling ratio specified by the image, a specified number of neighboring pixels corresponding to each pixel point in the target image are determined in the original image, and the specified neighboring pixels are located around the target pixel point in the original image; the process of calculating the pixel value of the target position by setting the corresponding weight coefficient according to the determined specified number of neighboring pixels realizes the generation of new pixel values ​​for the specified position of the image in the scaling process, so that the image of the mulberry tree has a higher resolution, and improves the accuracy of the subsequent model based on the image analysis of the growth stage of the mulberry tree to obtain the growth curve and the results of the maturity of the mulberry leaves.

[0056] Third, by performing preliminary image segmentation on the image data of the mulberry tree data, the leaf area and branch area in the image are segmented out; feature extraction is performed on the segmented leaf area, and the color features, size features, and texture features of the leaves are extracted, and marked at the corresponding positions; the process of feature extraction and marking of the segmented branch area can effectively reduce the amount of labeled samples and meet the training requirements of the model with a small amount of data samples.

[0057] Fourth, the input layer, hidden layer and output layer are set through the mulberry tree growth analysis model. The hidden layer is composed of 5 convolutional layers, pooling layers and 3 fully connected layers. The output layer is a softmax output layer. The two adjacent layers are connected by the maxpool layer, and each hidden layer uses a ReLU activation unit. The convolution layer is composed of two 3x3 convolution kernels stacked together, and the fully connected layer is a 1x1 convolution kernel, so that the mulberry tree growth analysis model can accurately extract the key features of each data and improve the accuracy of the model output results.

[0058] Fifth, a basic model is created and trained by random initialization. The weights of the basic model are used to initialize the mulberry tree growth analysis model, and the ReLU activation unit is used to preprocess the image using RELU as the activation function; the preprocessed image is cropped by randomly selecting a training scale within a specified range; the performance of the mulberry tree growth analysis model is evaluated using a validation set, and the hyperparameters are adjusted according to specified rules, so that the model has good generalization ability and robustness.

[0059] Sixth, the convolution layer and pooling layer in the hidden layer of the mulberry tree growth analysis model are used to extract features of the image input by the input layer. The features output by the convolution layer and the pooling layer are converted into feature vectors for output, and processed using global average pooling. The output probability of the input variables is calculated through a parameterized function, and the extracted image features are classified based on the output probability, thereby achieving accurate analysis of the growth stage of the mulberry tree and the maturity of the mulberry leaves. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 The figure is a schematic diagram of the process of the mulberry tree growth monitoring method based on image processing of the present invention.

[0061] Figure 2 The figure is a schematic diagram of the process of analyzing the mulberry tree growth analysis model of the present invention.

[0062] Figure 3 It is a schematic diagram of the architecture of the mulberry tree growth monitoring system based on image processing of the present invention. DETAILED DESCRIPTION

[0063] The following is combined with Figure 1~Figure 3 The present invention is further described in detail:

[0064] Example 1

[0065] See attached Figure 1 As shown, a mulberry tree growth monitoring method based on image processing comprises the following steps:

[0066] S1: Based on big data, image data of mulberry trees at various growth stages, optimal growth curves, and corresponding environmental parameters and fertilization data are obtained, and data preprocessing is performed on the acquired data, and mulberry tree data samples are constructed based on the preprocessed data. Through the data preprocessing process, noise and outliers in the data can be removed, reducing the risk of overfitting in the subsequent model training process, thereby improving the generalization ability of the model.

[0067] S2: labeling the mulberry tree data samples and creating a mulberry tree growth analysis model, inputting the labeled mulberry tree data samples into the mulberry tree growth analysis model to train the model, performing performance verification on the trained mulberry tree growth analysis model, and judging whether the verification result meets the specified requirements, if so, executing step S3, if not, performing parameter optimization on the trained mulberry tree growth analysis model until the model meets the specified requirements;

[0068] S3: performing real-time image acquisition of the mulberry tree through the image acquisition module, transmitting the acquired image data to the mulberry tree growth analysis model for analysis, and outputting the growth analysis result of the mulberry tree, wherein the growth analysis result includes the current growth stage of the mulberry tree and the corresponding maturity of each leaf of the current mulberry tree;

[0069] S4: The mulberry tree growth analysis model matches the optimal growth curve, environmental parameters and fertilization data according to the current growth stage of the mulberry tree, and transmits the environmental parameters and fertilization data to the environmental parameter control module and the fertilization control module, wherein the environmental parameter control module controls the environmental parameters, and the fertilization control module controls the fertilization time, type and amount of fertilization;

[0070] S5: collecting images of the mulberry tree at a specified time point, transmitting the collected image data to the mulberry tree growth analysis model for analysis and obtaining the maturity of the mulberry leaves, and drawing an actual growth curve of the mulberry tree, comparing the actual growth curve of the mulberry tree with the optimal growth curve of the mulberry tree, and judging whether the difference between the actual growth curve of the mulberry tree and the optimal growth curve is within a preset range, if not, executing step S6, and if yes, executing step S7;

[0071] S6: adjusting the environmental parameters, fertilization time and fertilization amount in step S4 according to the specified rules, and executing through the environmental parameter control module and the fertilization control module based on the adjusted environmental parameters, fertilization time and fertilization amount;

[0072] S7: picking mulberry leaves based on the corresponding mulberry leaf maturity data obtained at the specified time point, and storing them in corresponding positions according to the maturity of the mulberry leaves.

[0073] The above process constructs mulberry data samples and annotates them by acquiring various data of each growth stage of mulberry trees, and inputs them into the mulberry growth analysis model for training; real-time image acquisition of mulberry trees is transmitted to the mulberry growth analysis model, and the growth analysis results of mulberry trees are output; the optimal growth curve, environmental parameters and fertilization data are matched according to the current growth stage of mulberry trees, and the environmental parameters are controlled by the environmental parameter control module, and the fertilization control module controls the fertilization time, type and amount of fertilization; images are acquired at a specified time point, and transmitted to the mulberry growth analysis model for analysis to draw the actual growth curve of mulberry trees; and the environmental parameters, fertilization time and fertilization amount are adjusted according to the specified rules; mulberry leaves are picked at a specified time point based on the corresponding mulberry leaf maturity data obtained. Real-time growth monitoring of mulberry trees is realized, and based on comparison with the optimal growth curve, an effective data basis is provided for the control of environmental parameters and fertilization, so that mulberry trees can grow with the optimal growth curve or grow close to the optimal growth curve, increase the yield of mulberry leaves, and reduce the cost of silkworm breeding. By monitoring the growth of mulberry trees, the maturity of mulberry leaves can be provided so that mulberry leaves of corresponding maturity can be picked for silkworms of corresponding ages to eat. Silkworms of corresponding ages can eat mulberry leaves of matching maturity, allowing silkworms to eat more mulberry leaves, thereby promoting the rapid growth of silkworms while reducing the waste of mulberry leaves.

[0074] In this embodiment, the specific process of data preprocessing in step S1 is as follows:

[0075] S11: Remove duplicate data, noise data, and invalid data from the acquired data, effectively reducing the risk of overfitting in the subsequent model training process, thereby improving the generalization ability of the model;

[0076] S12: Convert the color image in the image data of each item of data into a grayscale image, and perform geometric transformation on the grayscale image, thereby simplifying the image matrix and improving the computational efficiency of the model.

[0077] The specific process of the image geometric transformation in step S12 is:

[0078] S121: performing scaling processing on all images so that all images are scaled to a uniform specified size;

[0079] S122: All images are translated to a specified position, and all images are rotated to a specified direction by image rotation. The above-mentioned scaling and rotation of images increase the diversity of data, so that the model can accurately recognize images of different sizes, directions and positions, which can effectively improve the model's adaptability to various situations in practical applications.

[0080] The image scaling process in step S121 adopts grayscale image interpolation processing, and the specific process is as follows:

[0081] S1211: Mapping each pixel point in the target image to be interpolated to a corresponding position in the original image based on the scaling ratio specified by the image;

[0082] S1212: determining in the original image a specified number of neighboring pixel points corresponding to each pixel point in the target image, wherein the specified number of neighboring pixel points are located around the target pixel point in the original image;

[0083] S1213: Calculate the pixel value of the target position by setting a corresponding weight coefficient according to the determined designated number of neighboring pixel points. The calculation formula is as follows:

[0084] P( x ,y)=S( j , k )* w 1+ S( j , k+ 1)* w 2 + S( j+ 1, k +1)* w 3+ S( j+ 1, k )* w 4;

[0085] Among them, P( x , y) is the target position ( x , y) pixel value, S( j , k )、S( j , k+ 1) S(j+ 1, k +1)、S( j+ 1, k ) are the target positions ( x , y) neighboring pixel values ​​in four directions, w 1. w 2. w 3. w 4 are weight coefficients corresponding to the neighboring points in the four directions respectively.

[0086] Based on the zoom ratio specified by the image, each pixel point in the target image to be interpolated is mapped to the corresponding position in the original image, and a specified number of neighboring pixel points corresponding to each pixel point in the target image are determined in the original image, and the specified neighboring pixel points are located around the target pixel point in the original image; the process of setting the corresponding weight coefficient according to the determined specified number of neighboring pixel points calculates the pixel value of the target position, thereby realizing the generation of new pixel values ​​for the specified position of the image in the zoom process, so that the image of the mulberry tree has a higher resolution, and improves the accuracy of the subsequent model based on the image analysis of the growth stage of the mulberry tree to obtain the growth curve and the results of the maturity of the mulberry leaves.

[0087] Example 2

[0088] Based on Example 1, the specific process of labeling the mulberry tree data sample in step S2 is as follows:

[0089] S21: performing preliminary image segmentation on the image data of the mulberry tree data to segment the leaf area and the branch area in the image;

[0090] S22: extracting features from the segmented leaf area, extracting color features, size features, and texture features of the leaf, and marking them at corresponding positions;

[0091] S23: Perform feature extraction on the segmented branch area, extract the color features, size features and texture features of the branches, and mark them at corresponding positions.

[0092] By performing preliminary image segmentation on the image data of mulberry tree data, the leaf area and branch area in the image are segmented out; feature extraction is performed on the segmented leaf area to extract the color features, size features, and texture features of the leaves, and mark them at the corresponding positions; the process of feature extraction and marking the segmented branch area effectively reduces the amount of labeled samples and meets the training requirements of the model with a small amount of data samples.

[0093] Example 3

[0094] On the basis of Example 1 or Example 2, the mulberry tree growth analysis model in step S2 is a neural network model, including an input layer, a hidden layer and an output layer, wherein the hidden layer is provided with 5 convolutional layers and a pooling layer and 3 fully connected layers, and the output layer is a softmax output layer, and two adjacent layers are connected by a maxpool layer, and each hidden layer adopts a ReLU activation unit, and the convolutional layer is formed by stacking two 3x3 convolutional kernels, and the fully connected layer is a 1x1 convolutional kernel. The above-mentioned model structure of the mulberry tree growth analysis model enables accurate extraction of key features in various data and improves the accuracy of the model output results.

[0095] The training and verification process of the mulberry tree growth analysis model is as follows:

[0096] S24: First, create a basic model and train the basic model by random initialization;

[0097] S25: Initializing the mulberry tree growth analysis model using the weights of the basic model, and preprocessing the image using a ReLU activation unit using RELU as an activation function;

[0098] S26: randomly select a training scale within a specified range to crop the preprocessed image;

[0099] S27: Use the validation set to evaluate the performance of the mulberry tree growth analysis model and adjust the hyperparameters according to the specified rules.

[0100] In the above process, a basic model is created and trained by random initialization. The weight of the basic model is used to initialize the mulberry tree growth analysis model, and the ReLU activation unit is used to preprocess the image using RELU as the activation function; the training scale is randomly selected within a specified range to crop the preprocessed image; the performance of the mulberry tree growth analysis model is evaluated using a validation set, and the hyperparameters are adjusted according to specified rules, so that the model has good generalization ability and robustness.

[0101] In this embodiment, see Figure 2 The specific process of the mulberry tree growth analysis model in step S3 is as follows:

[0102] S31: The convolution layer and pooling layer in the hidden layer of the mulberry tree growth analysis model extract features from the image input by the input layer;

[0103] S32: convert the features output by the convolution layer and the pooling layer into feature vectors for output, and process them using global average pooling;

[0104] S33: construct a parameterized function, input the extracted image features into the parameterized function, calculate the output probability of the input variables through the parameterized function, classify the extracted image features based on the output probability, and obtain the growth stage of the mulberry tree and the maturity of the mulberry leaves.

[0105] The convolution layer and pooling layer in the hidden layer of the mulberry tree growth analysis model are used to extract features of the image inputted into the input layer. The features outputted by the convolution layer and the pooling layer are converted into feature vectors for output and processed using global average pooling. The output probability of the input variables is calculated through a parameterized function, and the extracted image features are classified based on the output probability, thus achieving accurate analysis of the growth stage of mulberry trees and the maturity of mulberry leaves.

[0106] In the second aspect, a mulberry tree growth monitoring system based on image processing is provided. Figure 3 , including data acquisition module, data preprocessing module, data annotation module, model creation module, mulberry tree growth analysis model, growth curve drawing module, environmental parameter adjustment module, fertilization adjustment module and mulberry leaf picking module.

[0107] The data acquisition module is linked to the data preprocessing module, the data preprocessing module is connected to the data annotation module, the data annotation module and the model creation module are both connected to the mulberry tree growth analysis model, the mulberry tree growth analysis model is connected to the growth curve drawing module, the environmental parameter adjustment module, the fertilization adjustment module and the mulberry leaf picking module.

[0108] The data acquisition module is used to acquire image data of each growth stage of mulberry trees, optimal growth curves, and corresponding environmental parameters and fertilization data based on big data, and to perform real-time image acquisition of mulberry trees through the image acquisition module. The data preprocessing module is used to perform data preprocessing on the acquired data, and to construct mulberry tree data samples based on the preprocessed data; the data annotation module is used to annotate the mulberry tree data samples; the model creation module is used to create a mulberry tree growth analysis model, and input the annotated mulberry tree data samples into the mulberry tree growth analysis model to train the model.

[0109] The mulberry tree growth analysis model is used to analyze based on the collected image data and output the growth analysis results of the mulberry tree, wherein the growth analysis results include the current growth stage of the mulberry tree and the corresponding maturity of each leaf of the current mulberry tree, match the optimal growth curve, environmental parameters and fertilization data according to the current growth stage of the mulberry tree, and transmit the environmental parameters and fertilization data to the environmental parameter control module and the fertilization control module; the growth curve drawing module is used to draw the actual growth curve of the mulberry tree, compare the actual growth curve of the mulberry tree with the optimal growth curve of the mulberry tree, and judge whether the difference between the actual growth curve of the mulberry tree and the optimal growth curve is within a preset range; the environmental parameter adjustment module is used to adjust and control the environmental parameters based on the adjusted environmental parameters; the fertilization adjustment module is used to execute through the fertilization control module based on the adjusted fertilization time and fertilization amount; the mulberry leaf picking module is used to pick mulberry leaves at the specified time point based on the corresponding mulberry leaf maturity data obtained.

[0110] In summary, the mulberry growth monitoring system and method based on image processing provided by the present invention, by removing duplicate data, noise data and invalid data in the acquired data, converting to grayscale image, scaling processing and geometric transformation processing, provides an effective data basis for subsequent model training, and can improve the training efficiency of the model. By mapping each pixel point in the target image to be interpolated to the corresponding position in the original image, and setting the corresponding weight coefficient according to the determined designated number of adjacent pixels to calculate the pixel value of the target position, it is realized that a new pixel value is generated for the designated position of the image of the scaling process, so that the image of the mulberry tree has a higher resolution, and improves the accuracy of the subsequent model based on the image analysis of the growth stage of the mulberry tree to obtain the growth curve and the results of the maturity of the mulberry leaves.

[0111] By performing preliminary image segmentation, feature extraction and labeling on the image data of mulberry tree data, and extracting and labeling features on the segmented branch area, the amount of labeled samples can be effectively reduced, and the training requirements of the model can be met with a small amount of data samples. The input layer, hidden layer and output layer are set through the mulberry tree growth analysis model. The hidden layer is set with 5 convolutional layers, pooling layers and 3 fully connected layers. The output layer is a softmax output layer. The adjacent two layers are connected through the maxpool layer, and each hidden layer uses a ReLU activation unit. The convolution layer is stacked with two 3x3 convolution kernels, and the fully connected layer is a 1x1 convolution kernel, so that the mulberry tree growth analysis model can accurately extract the key features of each data and improve the accuracy of the model output results. A basic model is created and trained by random initialization. The weight of the basic model is used to initialize the mulberry growth analysis model, and the ReLU activation unit is used to preprocess the image using RELU as the activation function; the training scale is randomly selected within a specified range to crop the preprocessed image; the performance of the mulberry growth analysis model is evaluated using a validation set, and the hyperparameters are adjusted according to the specified rules, so that the model has good generalization ability and robustness. The image input to the input layer is extracted by the convolution layer and the pooling layer in the hidden layer of the mulberry growth analysis model, the features output by the convolution layer and the pooling layer are converted into feature vectors for output, and processed by global average pooling, the output probability of the input variable is calculated by a parameterized function, and the extracted image features are classified based on the output probability, so as to achieve accurate analysis of the growth stage of the mulberry tree and the maturity of the mulberry leaves.

Claims

1. A mulberry tree growth monitoring method based on image processing, characterized in that: The following steps are involved: S1: Based on big data, image data of each growth stage of mulberry trees, optimal growth curves, and corresponding environmental parameters and fertilization data are obtained, and data preprocessing is performed on the obtained data, and mulberry tree data samples are constructed based on the preprocessed data; S2: labeling the mulberry tree data samples and creating a mulberry tree growth analysis model, inputting the labeled mulberry tree data samples into the mulberry tree growth analysis model to train the model, performing performance verification on the trained mulberry tree growth analysis model, and judging whether the verification result meets the specified requirements, if so, executing step S3, if not, performing parameter optimization on the trained mulberry tree growth analysis model until the model meets the specified requirements; S3: performing real-time image acquisition of the mulberry tree through the image acquisition module, transmitting the acquired image data to the mulberry tree growth analysis model for analysis, and outputting the growth analysis result of the mulberry tree, wherein the growth analysis result includes the current growth stage of the mulberry tree and the corresponding maturity of each leaf of the current mulberry tree; S4: The mulberry tree growth analysis model matches the optimal growth curve, environmental parameters and fertilization data according to the current growth stage of the mulberry tree, and transmits the environmental parameters and fertilization data to the environmental parameter control module and the fertilization control module, wherein the environmental parameter control module controls the environmental parameters, and the fertilization control module controls the fertilization time, type and amount of fertilization; S5: collecting images of the mulberry tree at a specified time point, transmitting the collected image data to the mulberry tree growth analysis model for analysis and obtaining the maturity of the mulberry leaves, and drawing an actual growth curve of the mulberry tree, comparing the actual growth curve of the mulberry tree with the optimal growth curve of the mulberry tree, and judging whether the difference between the actual growth curve of the mulberry tree and the optimal growth curve is within a preset range, if not, executing step S6, and if yes, executing step S7; S6: adjusting the environmental parameters, fertilization time and fertilization amount in step S4 according to the specified rules, and executing through the environmental parameter control module and the fertilization control module based on the adjusted environmental parameters, fertilization time and fertilization amount; S7: picking mulberry leaves based on the corresponding mulberry leaf maturity data obtained at the specified time point, and storing them in corresponding positions according to the maturity of the mulberry leaves.

2. A mulberry tree growth monitoring method based on image processing according to claim 1, characterized in that: The specific process of data preprocessing in step S1 is as follows: S11: removing duplicate data, noise data and invalid data from the acquired data; S12: Convert the color image in the image data in each item of data into a grayscale image, and perform geometric transformation on the grayscale image.

3. A mulberry tree growth monitoring method based on image processing according to claim 2, characterized in that: The specific process of the image geometric transformation in step S12 is: S121: performing scaling processing on all images so that all images are scaled to a uniform specified size; S122: All images are translated to a specified position, and all images are rotated to a specified direction by image rotation.

4. A mulberry tree growth monitoring method based on image processing according to claim 3, characterized in that: The image scaling process in step S121 adopts grayscale image interpolation processing, and the specific process is as follows: S1211: Mapping each pixel point in the target image to be interpolated to a corresponding position in the original image based on the scaling ratio specified by the image; S1212: determining in the original image a specified number of neighboring pixel points corresponding to each pixel point in the target image, wherein the specified number of neighboring pixel points are located around the target pixel point in the original image; S1213: Calculate the pixel value of the target position by setting a corresponding weight coefficient according to the determined designated number of neighboring pixel points. The calculation formula is as follows: P( x ,y)=S( j , k )* w 1+ S( j , k+ 1)* w 2 + S( j+ 1, k +1)* w 3+ S( j+ 1, k )* w 4; Among them, P( x , y) is the target position ( x , y) pixel value, S( j , k )、S( j , k+ 1) S( j+ 1, k +1)、S( j+ 1, k ) are the target positions ( x , y) neighboring pixel values ​​in four directions, w 1. w 2. w 3. w 4 are weight coefficients corresponding to the neighboring points in the four directions respectively.

5. The mulberry tree growth monitoring method based on image processing according to claim 1, characterized in that: The specific process of labeling the mulberry tree data sample in step S2 is as follows: S21: performing preliminary image segmentation on the image data of the mulberry tree data to segment the leaf area and the branch area in the image; S22: extracting features from the segmented leaf area, extracting color features, size features, and texture features of the leaf, and marking them at corresponding positions; S23: Perform feature extraction on the segmented branch area, extract the color features, size features and texture features of the branches, and mark them at corresponding positions.

6. The mulberry tree growth monitoring method based on image processing according to claim 1, characterized in that: The mulberry tree growth analysis model in step S2 is a neural network model, including an input layer, a hidden layer and an output layer. The hidden layer is provided with 5 convolutional layers and a pooling layer and 3 fully connected layers. The output layer is a softmax output layer. Two adjacent layers are connected by a maxpool layer, and each hidden layer adopts a ReLU activation unit. The convolutional layer is formed by stacking two 3x3 convolutional kernels, and the fully connected layer is a 1x1 convolutional kernel.

7. A mulberry tree growth monitoring method based on image processing according to claim 6, characterized in that: The training and verification process of the mulberry tree growth analysis model is as follows: S24: First, create a basic model and train the basic model by random initialization; S25: Initializing the mulberry tree growth analysis model using the weights of the basic model, and preprocessing the image using a ReLU activation unit using RELU as an activation function; S26: randomly select a training scale within a specified range to crop the preprocessed image; S27: Use the validation set to evaluate the performance of the mulberry tree growth analysis model and adjust the hyperparameters according to the specified rules.

8. The mulberry tree growth monitoring method based on image processing according to claim 6, characterized in that: The specific process of the mulberry tree growth analysis model in step S3 is as follows: S31: The convolution layer and pooling layer in the hidden layer of the mulberry tree growth analysis model extract features from the image input by the input layer; S32: convert the features output by the convolution layer and the pooling layer into feature vectors for output, and process them using global average pooling; S33: construct a parameterized function, input the extracted image features into the parameterized function, calculate the output probability of the input variables through the parameterized function, classify the extracted image features based on the output probability, and obtain the growth stage of the mulberry tree and the maturity of the mulberry leaves.

9. A mulberry tree growth monitoring system based on image processing, used to implement a mulberry tree growth monitoring method based on image processing according to any one of claims 1 to 8, characterized in that: It includes data acquisition module, data preprocessing module, data annotation module, model creation module, mulberry tree growth analysis model, growth curve drawing module, environmental parameter adjustment module, fertilization adjustment module and mulberry leaf picking module; The data acquisition module is linked to the data preprocessing module, the data preprocessing module is connected to the data annotation module, the data annotation module and the model creation module are both connected to the mulberry tree growth analysis model, the mulberry tree growth analysis model is connected to the growth curve drawing module, the environmental parameter adjustment module, the fertilization adjustment module and the mulberry leaf picking module; The data acquisition module is used to acquire image data of each growth stage of the mulberry tree, the optimal growth curve, and corresponding environmental parameters and fertilization data based on big data, and to perform real-time image acquisition of the mulberry tree through the image acquisition module; The data preprocessing module is used to perform data preprocessing on the acquired data and construct a mulberry tree data sample based on the preprocessed data; The data labeling module is used to label the mulberry tree data sample; The model creation module is used to create a mulberry tree growth analysis model, and input the labeled mulberry tree data samples into the mulberry tree growth analysis model to train the model; The mulberry tree growth analysis model is used to analyze based on the collected image data, output the growth analysis results of the mulberry tree, the growth analysis results include the current growth stage of the mulberry tree and the corresponding maturity of each leaf of the current mulberry tree, match the optimal growth curve, environmental parameters and fertilization data according to the current growth stage of the mulberry tree, and transmit the environmental parameters and fertilization data to the environmental parameter control module and the fertilization control module; The growth curve drawing module is used to draw the actual growth curve of the mulberry tree, compare the actual growth curve of the mulberry tree with the optimal growth curve of the mulberry tree, and determine whether the difference between the actual growth curve of the mulberry tree and the optimal growth curve is within a preset range; The environmental parameter adjustment module is used to perform environmental parameter adjustment control based on the adjusted environmental parameters; The fertilization adjustment module is used to execute through the fertilization control module based on the adjusted fertilization time and fertilization amount; The mulberry leaf picking module is used to pick mulberry leaves based on the corresponding mulberry leaf maturity data obtained at the specified time point.

Images