CN115438934A

CN115438934A - Crop growth environment monitoring method and system based on block chain

Info

Publication number: CN115438934A
Application number: CN202211013461.3A
Authority: CN
Inventors: 王卓薇; 黄焕洲; 程良伦; 赵艮平
Original assignee: Guangdong University of Technology
Current assignee: Guangdong University of Technology
Priority date: 2022-08-23
Filing date: 2022-08-23
Publication date: 2022-12-06

Abstract

The invention discloses a crop growth environment monitoring method and system based on a block chain, wherein the method comprises the following steps: acquiring plot image data and constructing a plot change detection model; obtaining crop yield remote sensing data and constructing a crop yield estimation model; acquiring crop disease and insect pest image data and constructing a disease and insect pest detection model; obtaining crop sowing data and growth data and constructing a crop traceability model based on the alliance block chain; and integrating a plot change detection model, a crop yield estimation model, a pest and disease detection model and a crop traceability model based on a block chain technology to obtain a crop growth environment monitoring model. The system comprises: the system comprises a plot detection module, a yield estimation module, a pest detection module, a crop source tracing module and a monitoring module. By using the method and the device, the problems of inaccurate data of an industrial chain, more prominent agricultural disasters, poor agricultural production accuracy and incapability of tracing agricultural products can be solved.

Description

Crop growth environment monitoring method and system based on block chain

Technical Field

The invention relates to the field of crop growth environment monitoring, in particular to a crop growth environment monitoring method and system based on a block chain.

Background

With the explosive development of information technology, the penetration of the Internet subverts the traditional agricultural mode, and big data, sensors, the Internet of things and cloud computing change the traditional manual working mode and the extensive agricultural production mode, so that the traditional agriculture is advanced to intensification, precision, intellectualization and datamation; the new generation internet technology provides important support for the innovation of digital agriculture, however, with the improvement of agriculture industrialization and scale level, a series of problems are gradually exposed: the problems that data of the whole agricultural industry chain is inaccurate, agricultural disasters are prominent, precision of agricultural production is poor, agricultural products cannot be traced, food safety cannot be guaranteed and the like are closely related to the growth environment of crops, so that monitoring of the growth environment of the crops and taking of corresponding precautionary measures are necessary when necessary.

However, traditional crop growth environment monitoring still depends on manual regulation and control, and a manager cannot timely know environment information under certain conditions, so that accurate judgment cannot be made, especially in some economic crop planting fields, regulation and control cannot be timely performed if the growth environment information of crops cannot be accurately obtained, and final output is affected.

Disclosure of Invention

In order to solve the technical problems, the invention aims to provide a crop growth environment monitoring method and system based on a block chain, which can solve the problems of inaccurate industrial chain data, prominent agricultural disasters, poor agricultural production accuracy and incapability of tracing agricultural products.

The first technical scheme adopted by the invention is as follows: a crop growth environment monitoring method based on a block chain comprises the following steps:

acquiring plot image data and constructing a plot change detection model;

obtaining crop yield remote sensing data and constructing a crop yield estimation model;

acquiring crop pest image data and constructing a pest detection model;

acquiring crop sowing data and growth data and constructing a crop traceability model based on the alliance block chain;

and integrating a plot change detection model, a crop yield estimation model, a pest and disease detection model and a crop traceability model based on a block chain technology to obtain a crop growth environment monitoring model.

Further, the step of acquiring the image data of the land parcel and constructing the land parcel change detection model specifically includes:

acquiring plot image data and constructing a plot database;

selecting two to-be-fused land images from a land database and respectively calculating corresponding information entropies;

performing wavelet decomposition on the two to-be-fused land image and the corresponding information entropy to obtain a multi-resolution structure;

on a multi-resolution structure, wavelet coefficients corresponding to the information entropies of two to-be-fused land images in the horizontal, vertical and diagonal directions are compared, and the coefficient of one to-be-fused land image is selected as a reconstruction coefficient of the to-be-fused land image on the resolution;

based on the wavelet coefficient, a fused image is obtained through wavelet reconstruction of a two-dimensional image;

identifying the changed position in the fusion image and outputting a binary image;

and reselecting the plot image data to execute circular training to obtain a plot change detection model.

Further, the step of obtaining crop yield remote sensing data and constructing a crop yield estimation model specifically includes:

obtaining crop yield remote sensing data and constructing a crop yield database;

selecting crop yield remote sensing data from a crop yield database and carrying out data fusion;

determining nodes of a decision tree in a random forest by using a least square fitting method;

training a decision tree by using crop yield remote sensing data, and calculating the average reduced impurity degree of each feature as a feature selection value;

training an artificial neural network by using the value selected by the characteristics and the corresponding crop yield to obtain the yield of each position and output a yield color chart;

and reselecting crop yield remote sensing data to execute cycle training to obtain a crop yield estimation model.

Further, the step of obtaining crop disease and insect pest image data and constructing a disease and insect pest detection model specifically comprises:

acquiring a crop disease and insect pest picture and constructing a disease and insect pest database;

selecting training data from a pest database, and carrying out pretreatment and image enhancement to obtain image-enhanced pretreatment data;

and training the Resnet 50 model of the residual error network by utilizing the preprocessed data after image enhancement, and outputting a disease and insect category distribution result by taking focal loss as a loss function to obtain a disease and insect detection model.

Further, the step of obtaining crop seeding data and growth data and constructing a crop traceability model based on the alliance block chain specifically includes:

acquiring crop sowing data and growth data and constructing a crop tracing source database;

and constructing a seeding chain according to the seeding data, constructing a growth chain according to the growth data, and integrating the seeding chain and the growth chain based on the alliance block chain to obtain a crop traceability model.

Further, the seeding chain is used for recording the seeding condition, and specifically comprises:

before sowing, recording the source of crop seeds and detecting and analyzing a report as the initial position of a sowing chain;

in the sowing process, real-time digital description is carried out on the sowing stage by recording the application amount of the fertilizer and the usage record of the fertilizer and the pesticide;

and linking the seeding conditions before and during seeding after seeding.

Further, the growth chain includes that the environment is accurate to be monitored, the record of weeding deinsectization, growth reap three node of record, specifically includes:

the environment accurate monitoring predicts the future weather condition through an antagonistic neural network, and selects the optimal growth harvesting time;

the record of weeding and deinsectization is used for recording the medicament, the dosage and the time used by weeding and deinsectization;

the growing harvest record is used to record the number of crops harvested and their corresponding harvest time.

Further, the method also comprises a data acquisition step, and specifically comprises the following steps:

collecting first type data with an unmanned aerial vehicle through a hyperspectral technology;

collecting second-class data through a satellite remote sensing technology;

collecting third type data through a fixed sensor and a camera;

integrating the first type of data, the second type of data and the third type of data to obtain integrated data;

and dividing the integrated data to obtain weather data, land image data, crop yield remote sensing data, crop pest image data, crop sowing data and growth data.

The second technical scheme adopted by the invention is as follows: a block chain based crop growth environment monitoring system comprising:

the land parcel detection module is used for acquiring land parcel image data and constructing a land parcel change detection model;

the yield estimation module is used for acquiring crop yield remote sensing data and constructing a crop yield estimation model;

the pest detection module is used for acquiring crop pest image data and constructing a pest detection model;

the crop tracing module is used for acquiring crop sowing data and growth data and constructing a crop tracing model based on the alliance block chain;

and the monitoring module integrates a plot change detection model, a crop yield estimation model, a pest and disease detection model and a crop traceability model based on a block chain technology to obtain a crop growth environment monitoring model.

The method and the system have the beneficial effects that: the crop growth environment monitoring system is complete in function, safe, reliable and simple to operate, first the first kind of data is collected by a hyperspectral technology and an unmanned aerial vehicle, the second kind of data is collected by a satellite remote sensing technology, and the third kind of data is collected by a fixed sensor and a camera, so that a high-precision multi-dimensional monitoring system is realized, the cost of traditional manual monitoring and statistics is reduced, and the problem of poor precision of agricultural production is solved; secondly, by providing a crop growth environment monitoring model, a grower can be ensured to accurately know information such as plot change, pest and disease damage conditions, crop yield and agricultural product traceability in the crop production process, and the problems that agricultural disasters are more prominent, agricultural products cannot be traced and the like in agricultural modernization are solved; meanwhile, the authenticity and the validity of data are ensured by utilizing a block chain technology.

Drawings

FIG. 1 is a flow chart of the steps of a method for monitoring the growth environment of a crop based on a blockchain according to the present invention;

FIG. 2 is a block diagram of a crop growth environment monitoring system based on a block chain according to the present invention;

FIG. 3 is a schematic diagram of a first image of a parcel to be fused in accordance with an embodiment of the present invention;

FIG. 4 is a schematic diagram of a second tile image to be merged according to an embodiment of the invention;

FIG. 5 is a schematic diagram of a land parcel change detection result according to an embodiment of the present invention;

FIG. 6 is a schematic illustration of remote crop yield data according to an embodiment of the present invention;

FIG. 7 is a diagram illustrating yield estimation results according to an embodiment of the present invention;

FIG. 8 is a schematic diagram of the result of pest detection according to the embodiment of the present invention;

FIG. 9 is a schematic flow chart of building a federation blockchain according to a specific embodiment of the present invention;

FIG. 10 is a schematic diagram of a traceability process of agricultural products according to an embodiment of the present invention;

fig. 11 is a schematic diagram of the agricultural product traceability result according to the embodiment of the present invention.

Detailed Description

The invention is described in further detail below with reference to the figures and the specific embodiments. The step numbers in the following embodiments are provided only for convenience of illustration, the order between the steps is not limited at all, and the execution order of each step in the embodiments can be adapted according to the understanding of those skilled in the art.

Referring to fig. 1, the invention provides a crop growth environment monitoring method based on a block chain, which comprises the following steps:

s1, collecting data;

the method comprises the steps of firstly acquiring first-class data, second-class data and third-class data, integrating and dividing the first-class data, the second-class data and the third-class data, and finally obtaining weather data, land image data, crop yield remote sensing data, crop pest image data, crop seeding data and growth data required by a construction model.

The first type of data is also called as 'empty' base class data, and the acquisition of the first type of data is mainly realized by a hyperspectral technology and an unmanned aerial vehicle; not only can observation be more flexible by using the unmanned aerial vehicle remote sensing technology, but also the resolution ratio is greatly improved, the precision of acquired information is higher, and the monitoring and application requirements of modern agriculture are met.

The second kind of data is also called data of a 'sky' base class, and the acquisition of the data is mainly realized by a satellite remote sensing technology; remote sensing provides an important technical means for quickly, accurately and dynamically acquiring space variation parameters required by precision agriculture by periodically collecting earth surface information in different electromagnetic spectrum sections.

The third kind of data is also called as data of ground base class, and the data is most widely collected and used and is mainly realized by a fixed sensor and a camera; the data such as air pressure, temperature and humidity, soil, illumination, weather, water quality and the like can be acquired.

S2, acquiring land image data and constructing a land change detection model;

s2.1, acquiring the image data of the land parcel and constructing a land parcel database;

s2.2, selecting land parcel image data from the land parcel database, and constructing a land parcel change detection model based on an improved image fusion algorithm.

Specifically, two land parcel images F (i, j) and G (i, j) to be fused are selected from a land parcel database;

calculating the local information entropy of each point by using a 3*3 area for F (i, j) and G (i, j) point by point to obtain X (i, j) and Y (i, j);

respectively carrying out wavelet decomposition on X (i, j) and Y (i, j) as well as F (i, j) and G (i, j) to obtain respective multi-resolution structures, wherein the multi-resolution structure is a structure consisting of a plurality of resolutions;

at each resolution, wavelet coefficients corresponding to X (i, j) and Y (i, j) in the horizontal, vertical and diagonal directions are compared, and the coefficient of F (i, j) or G (i, j) is selected as a reconstruction coefficient of the fused image H at the resolution, wherein the fusion criterion is as follows:

wherein k =1,2,3 represents the horizontal high frequency, vertical high frequency and diagonal high frequency component coefficients respectively, and l is the wavelet decomposition series;

obtaining a fusion image H through wavelet reconstruction of a two-dimensional image by utilizing the wavelet coefficient obtained by the formula, identifying a changed position in the fusion image H and outputting a binary image;

When the method is applied, as shown in fig. 3, 4 and 5, a first image of a land parcel to be fused and a second image of a land parcel to be fused are input into a land parcel change detection model, the first image of the land parcel and the second image of the land parcel are subjected to matrix reconstruction based on an improved image fusion algorithm, a fusion image is generated, a changed position is identified, a binary image is output, and therefore the change of the land parcel is detected.

S3, obtaining crop yield remote sensing data and constructing a crop yield estimation model to obtain the crop yield estimation model;

s3.1, obtaining crop yield remote sensing data and constructing a crop yield database;

s3.2, selecting crop yield remote sensing data from the crop yield database and carrying out data fusion;

in the field of remote sensing imaging detection, the multispectral imaging technology can acquire images of different wave bands of an observation scene, and the laser radar can acquire accurate three-dimensional space information of the observation scene; the data fusion is to fuse multispectral data and laser radar data in the crop yield remote sensing data so as to eliminate heterogeneity among different modes of data.

S3.3, performing feature screening on the crop yield remote sensing data subjected to data fusion by using a random forest average impurity degree reduction method;

specifically, a random forest is composed of a plurality of decision trees, each node in the decision trees is a condition about a certain characteristic, and the purpose is to divide crop yield data into two parts according to different response variables; then determining a node by using a least square fitting method; and finally, training a decision tree by using the crop yield remote sensing data after data fusion, calculating the impurity degree reduced by each feature, calculating the average reduced impurity degree of each feature for a decision tree forest, taking the average reduced impurity degree of each feature as a feature selection value, and obtaining the feature importance of each feature according to the feature selection value, thereby realizing the feature screening of the crop yield remote sensing data.

S3.4, training an artificial neural network by using the value selected by the characteristics and the corresponding crop yield to obtain the yield of each position and output a yield color chart;

and S3.5, reselecting the crop yield remote sensing data to execute cycle training to obtain a crop yield estimation model.

Wherein the corresponding crop yield can be obtained from the remote sensing data of the crop yield.

In application, as shown in fig. 6 and 7, the remote sensing data of crop yield is input into the crop yield estimation model, the crop yield at each position is estimated, and the yield color chart is used as an output result according to the crop yield.

S4, acquiring crop disease and insect pest image data and constructing a disease and insect pest detection model to obtain a disease and insect pest detection model;

s4.1, obtaining a crop disease and insect pest picture and constructing a disease and insect pest database;

s4.2, selecting training data from the pest and disease damage database, and carrying out pretreatment and image enhancement to obtain pretreatment data after image enhancement;

specifically, the data preprocessing mainly adopts noise addition, and aims to enable the detection of plant diseases and insect pests to be closer to the real image recognition process, so that the overfitting problem is effectively inhibited. The probability density function of gaussian noise is:

in the above formula, E is the sample mean, σ is the standard deviation of t, and then gaussian noise with a mean of 0 and a standard deviation of 0.05 is added to the crop pest image.

And S4.3, training the residual error network Resnet 50 model by utilizing the preprocessed data after image enhancement, and outputting a pest type distribution result by taking focal loss as a loss function to obtain a pest detection model.

Specifically, in the residual error network Resnet 50 model adopted in the preferred scheme of the embodiment, two continuous 3*3 convolutions are used for replacing a 7*7 convolution layer and a 3*3 maximum pooling layer of the original residual error network Resnet 50 model, and 3*3 maximum pooling operation is needed after the two convolutions, so that better fine-grained visual characteristics can be obtained, and the identification of early diseases and insect pests of crops is improved; furthermore, the last layer of the network consists of a dense layer of the softmax activation function with an output size of 1056 elements during pre-training; in order to reduce errors caused by uneven pest type distribution in the data set, focal loss is selected as a loss function; finally, the pooling layer before full concatenation is modified, using adaptive pooling, so that data of different resolutions can be used.

Where focal loss is a loss function, defined as follows:

FL(p _t )＝-α _t (1-p _t ) ^γ log(p _t )；

in the above formula, (1-p) _t ) ^γ For adjusting the factor, gamma is more than or equal to 0 and is an adjustable focusing parameter, and the value of gamma is 2.

It should be noted that before training the residual network Resnet 50 model with the pre-processed data after image enhancement, the weight parameters of the model need to be initialized.

When the method is applied, as shown in fig. 8, the crop pest image is input into the pest detection model, and the corresponding type of the crop pest and the detection time are obtained.

S5, crop sowing data and growth data are obtained, a crop traceability model is built based on the alliance block chain, and the crop traceability model is obtained;

s5.1, obtaining crop sowing data and growth data and constructing a crop tracing database;

and S5.2, as shown in the figure 10, constructing a seeding chain by using the seeding data, constructing a growth chain by using the growth data, and integrating the seeding chain and the growth chain based on the alliance block chain to obtain a crop traceability model.

S5.2.1, constructing a sowing chain according to sowing data;

specifically, the sowing chain firstly records crop seed sources and seed detection analysis reports, and the crop seed sources and the seed detection analysis reports are used as the initial position of the sowing chain; then, recording the application amount of the fertilizer and the use record of the fertilizer and the pesticide, and carrying out real-time digital description on the sowing stage; and after the seeding is finished, chaining the whole seeding condition and broadcasting the whole network.

S5.2.2, constructing a growing chain according to the growing data;

specifically, the growing chain mainly takes environment accurate monitoring, weeding and pest killing records and growing and harvesting records as main upper chain nodes; the environment is accurately monitored, the future weather condition is predicted through an antagonistic neural network, and the optimal growth harvesting time is selected; the record of weeding and deinsectization is used for recording the medicament, the dosage and the time used for weeding and deinsectization; the growing harvest record is used to record the number of crops harvested and their corresponding harvest time.

S5.2.3, and integrating a seeding chain and a growing chain based on the alliance block chain to obtain a crop traceability model.

As shown in fig. 9, initializing a system and a sequencing node, starting a peer node based on a configuration file, generating a channel, and adding the peer node with a certificate into the channel.

When the method is applied, firstly, a farmland three-dimensional model is built for a farmland to be detected, an index tag is marked, the index tag is stored in a block chain, and an index address is linked to a cloud database; then, a crop pest and disease damage affected area is identified by carrying out change detection on the farmland three-dimensional model, and accurate pesticide spraying, weeding and pest killing are carried out; in the final harvesting step, the optimal harvesting time interval is selected by collecting regional weather change conditions and predicting the future weather conditions by combining with the antagonistic neural network, and the data is linked up, and the traceability information of the data is shown in fig. 11, including variety, producing area, harvest time, harvest quantity and the like.

And S6, integrating a land parcel change detection model, a crop yield estimation model, a pest detection model and a crop traceability model based on a block chain technology to obtain a crop growth environment monitoring model.

When the method is applied, firstly, a crop growth environment monitoring model is utilized to ensure that a grower can accurately know information such as plot change, pest and disease damage conditions, crop yield, agricultural product traceability and the like in the crop production process, and the problems of more prominent agricultural disasters, incapability of traceability of agricultural products and the like in agricultural modernization are solved; and secondly, monitoring results obtained by the crop growth environment monitoring model are stored in a corresponding database in real time, and authenticity, effectiveness and integrity of agricultural product traceability are guaranteed by the characteristics of the block chain technology in the aspects of data storage decentralization, data non-falsification and the like.

As shown in fig. 2, a crop growth environment monitoring system based on a block chain comprises:

The contents in the method embodiments are all applicable to the system embodiments, the functions specifically implemented by the system embodiments are the same as those in the method embodiments, and the beneficial effects achieved by the system embodiments are also the same as those achieved by the method embodiments.

While the preferred embodiments of the present invention have been illustrated and described, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.

Claims

1. A crop growth environment monitoring method based on a block chain is characterized by comprising the following steps:

acquiring land image data and constructing a land change detection model;

acquiring crop disease and insect pest image data and constructing a disease and insect pest detection model;

obtaining crop sowing data and growth data and constructing a crop traceability model based on the alliance block chain;

2. The method for monitoring the crop growth environment based on the block chain as claimed in claim 1, wherein the step of obtaining the image data of the land and constructing the land change detection model specifically comprises:

acquiring land parcel image data and constructing a land parcel database;

based on the wavelet coefficient, obtaining a fused image through wavelet reconstruction of the two-dimensional image;

3. The method for monitoring crop growth environment based on blockchain according to claim 1, wherein the step of obtaining crop yield remote sensing data and constructing a crop yield estimation model specifically comprises:

4. The method for monitoring the crop growth environment based on the blockchain according to claim 1, wherein the step of obtaining the crop pest image data and constructing a pest detection model specifically comprises the steps of:

selecting training data from a disease and pest database, and carrying out preprocessing and image enhancement to obtain preprocessed data after image enhancement;

5. The method for monitoring the crop growing environment based on the block chain as claimed in claim 1, wherein the step of obtaining crop seeding data and growing data and constructing the crop traceability model based on the alliance block chain specifically comprises:

6. The method for monitoring the crop growth environment based on the block chain as claimed in claim 5, wherein the sowing chain is used for recording the sowing condition, and comprises:

in the sowing process, the sowing stage is digitally described in real time by recording the application amount and usage record of the fertilizer and the pesticide;

and (4) after sowing, chaining the sowing conditions before and during sowing.

7. The method for monitoring the crop growth environment based on the block chain as claimed in claim 5, wherein the growth chain comprises three nodes of environment accurate monitoring, weeding and pest killing record and growth and harvest record, and specifically comprises:

the weeding and pest killing record is used for recording the medicament, the dosage and the time used for weeding and pest killing;

8. The crop growth environment monitoring method based on the blockchain as claimed in claim 1, further comprising a data acquisition step, specifically comprising:

collecting second-class data through a satellite remote sensing technology;

acquiring third-class data through a fixed sensor and a camera;

and dividing the integrated data to obtain weather data, plot image data, crop yield remote sensing data, crop pest image data and crop seeding data and growth data.

9. A crop growth environment monitoring system based on a block chain, comprising: