Method for dynamically monitoring field growth and development of crops
By segmenting the planting area into calibrated and non-calibrated areas, and using regression model and neural network model to correct the plant height feature index, the problem of estimation deviation of corn plant height under the influence of terrain factors in remote sensing technology is solved, and the accuracy of corn growth monitoring is improved sex.
Patent Information
- Application Number
- CN202410957412.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-17
- Publication Date
- 2024-11-12
AI Technical Summary
The prior art fails to effectively consider topographic factors when estimating corn plant height using remote sensing technology, resulting in plant height estimation deviations, affecting the accuracy of corn growth monitoring.
By dividing the planting area to be monitored into calibration areas and non-calibrated areas, the average plant height of the calibration area is measured, a regression model of plant height and vegetation index is constructed, and a neural network model is used to correct the plant height characteristic index of the non-calibrated areas to reduce the ground The influence of height discrepancy on the characteristic parameters of remote sensing images.
It improves the accuracy of corn plant height estimation, more accurately reflects the growth status of corn crops in the planted area, and reduces the estimation deviation caused by uneven ground heights.
Smart Images

Figure CN118941883A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of crop growth monitoring, and in particular to a method for dynamically monitoring crop field growth and development. Background Art
[0002] Corn cultivation accounts for a large proportion of my country's agriculture, and its output is second only to rice and wheat. The planting area is stable at more than 600 million mu all year round, making it one of China's main food crops. The growth trend of crops is an important research direction in modern agriculture. The use of remote sensing to monitor the growth of crops is suitable for a large-scale analysis of crop growth. When crops grow, due to the influence of external factors, the growth of the same variety of the same type of crops in the same region may also be different.
[0003] The plant height of corn is an important parameter to measure its growth. At present, the method of estimating corn plant height using remote sensing technology such as inversion is generally to obtain remote sensing images of corn crops in the planting area and extract characteristic parameters that characterize the overall plant height of corn in the image, such as normalized vegetation index, green normalized vegetation index, green leaf vegetation index, etc. Then, the relevant model is constructed using the index parameters to estimate the plant height of corn, so as to realize dynamic monitoring of corn growth in different periods. However, this method does not take into account the influence of terrain factors on plant height estimation. Generally, remote sensing images are taken by drones or spacecraft at a certain distance above the planting area. For corn crops in the field, the ground height in the planting area is often uneven, which makes the characteristic parameters of the extracted remote sensing images different from the actual situation, resulting in an increase in the estimation deviation of plant height. To this end, we propose a method for dynamic monitoring of crop field growth and development. Summary of the invention
[0004] The main purpose of the present invention is to provide a method for dynamically monitoring the growth and development of crops in the field, which can effectively solve the problems in the background technology.
[0005] In order to achieve the above object, the technical solution adopted by the present invention is:
[0006] A method for dynamically monitoring the growth and development of crops in a field, comprising:
[0007] The planting area to be monitored is divided into grids with a step size of λ, and the calibration area r is selected. c , measure the calibration area r c Average plant height of crops in the range
[0008] Obtain the remote sensing image of the crops in the monitored planting area, and extract the calibration area r in the remote sensing image c Plant height characteristic index X of crops within the range j, construct the average plant height of crops in the monitored planting area and plant height characteristic index regression model, where the plant height characteristic index X j The acquisition process includes the following steps:
[0009] The shooting height of the remote sensing image to be measured is used as the calibration shooting height H c ;
[0010] Extract the commonly used vegetation index x of the calibration area in the remote sensing image to be tested i , represented by the i-th commonly used vegetation index, i=1,2,...,n; wherein the commonly used vegetation index at least includes normalized vegetation index, green normalized vegetation index, renormalized vegetation index, ratio vegetation index, terrestrial chlorophyll index, vegetation extraction color index, overgreen vegetation index, green leaf vegetation index, plant pigment ratio, red-green-blue vegetation index, terrestrial chlorophyll index, moisture index, improved red edge ratio vegetation index, and vegetation attenuation index;
[0011] Calculate the commonly used vegetation index and average plant height The correlation coefficient between them is calculated as:
[0012]
[0013] In the formula, ρ i is the correlation coefficient between the i-th commonly used vegetation index and the average plant height; is the mean value of the i-th commonly used vegetation index; is the mean of the average plant height;
[0014] Screen the commonly used vegetation indices whose correlation coefficients are within the correlation threshold range [0.5, 1] as the plant height characteristic index X j , expressed as the jth commonly used vegetation index, j = 1, 2, ..., m; where m ≤ n;
[0015] Get the relative shooting height H r Calibration area r c The contrast remote sensing image of crops within the range is extracted, and the calibration area r is c Plant height characteristic index X of crops within the range j ', where the relative shooting height H r With the calibration shooting height H c Not equal;
[0016] The shooting height difference ΔH is used as the input parameter, and the plant height characteristic index difference ΔX j The neural network model was constructed by taking the height difference ΔH and the plant height characteristic index difference ΔX as input parameters. j The calculation formulas are: ΔH=Hr -H c ; ΔX j =X j '-X j , the network model outputs the shooting height difference ΔH and the plant height characteristic index difference ΔX j The mapping relationship is as follows, wherein the neural network model is a BP neural network model having an input layer with one neuron, an intermediate layer with s neurons, and an output layer with j neurons, wherein the number of neurons s in the intermediate layer of the neural network model is determined according to the following formula:
[0017]
[0018] Wherein, a is an integer from 1 to 9;
[0019] Get the non-calibrated area and the calibrated area r corresponding to the kth grid in the planting area to be monitored c The ground height difference Δh k , the calculation formula is: Δh k =h k -h c ; In the formula, h k 、h c They are respectively represented as the surface height values of the non-calibrated area and the calibrated area corresponding to the kth grid. According to the obtained surface height difference Δh k And mapping relationship, to correct various plant height characteristic indexes of the non-calibrated area of the monitored planting area in the remote sensing image. The specific process includes the following steps:
[0020] Get the non-calibrated area and the calibrated area r corresponding to the kth grid in the planting area to be monitored c The ground height difference Δh k ;
[0021] The height difference from the ground surface is Δh k The shooting height difference ΔH with the same value is used as the input parameter, and the plant height characteristic index difference ΔX of the non-calibrated area corresponding to the kth grid is obtained according to the mapping relationship. kj ;
[0022] Get the plant height characteristic index of the crop in the non-calibrated area corresponding to the kth grid in the remote sensing image to be tested
[0023] According to the obtained and plant height characteristic index difference ΔX j , calculate the corrected plant height characteristic index X of the crop in the non-calibrated area corresponding to the kth grid kj ,in,
[0024] The corrected plant height characteristic index was obtained, and the average plant height of crops in each non-calibrated area was determined according to the regression model.
[0025] The present invention has the following beneficial effects:
[0026] Compared with the prior art, the technical solution of the present invention divides the planting area to be monitored into calibration areas r c and non-calibrated area, measure the calibrated area r c Average plant height of crops in the range Obtain the remote sensing image of the crops in the monitored planting area, and extract the calibration area r in the remote sensing image c Plant height characteristic index X of crops within the range j , construct the average plant height of crops in the monitored planting area The relative shoot height H was obtained by using the regression model of plant height characteristic index r Calibration area r c The contrast remote sensing image of crops within the range is extracted, and the calibration area r is c Plant height characteristic index X of crops within the range j ', taking the shooting height difference ΔH as the input parameter and the plant height characteristic index difference ΔX j The neural network model is constructed as input parameters, and the shooting height difference ΔH and plant height characteristic index difference ΔX are output through the network model. j The mapping relationship between the non-calibrated area and the calibrated area r corresponding to the kth grid in the planting area to be monitored is obtained. c The ground height difference Δh k , according to the obtained surface height difference Δh k And mapping relationship, the plant height characteristic indexes of the non-calibrated area of the monitored planting area in the remote sensing image are corrected to obtain the corrected plant height characteristic index, and the average plant height of the crops in each non-calibrated area is determined according to the regression model. This can reduce the difference between the characteristic parameters of the extracted remote sensing graphics and the actual situation caused by the uneven ground height in the planting area, thereby improving the accuracy of corn plant height estimation, which is conducive to better reflecting the growth status of corn crops in the planting area. BRIEF DESCRIPTION OF THE DRAWINGS
[0027] Figure 1 The present invention is a schematic flow chart of a method for dynamically monitoring the growth and development of crops in the field. DETAILED DESCRIPTION
[0028] The present invention will be further described below in conjunction with specific implementation methods, wherein the accompanying drawings are only used for exemplary descriptions and represent only schematic diagrams rather than actual drawings, and should not be understood as limiting the present invention. In order to better illustrate the specific implementation methods of the present invention, some parts of the accompanying drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product.
[0029] Example 1
[0030] like Figure 1 As shown, the implementation process of the technical solution of the present invention includes the following steps:
[0031] Step 1: Divide the planting area to be monitored into grids with a step size of λ, and select a relatively flat and regular area in the planting area to be monitored as the calibration area r c , the sampling survey method is used to measure the calibration area r c Average plant height of crops in the range The calculation formula is: Where u is the calibration area r c Number of corn crops sampled within range.
[0032] Step 2: Obtain the remote sensing image of the crops in the monitored planting area. In this embodiment, the RGB camera and the RedEdge-M multispectral camera can be used to acquire the remote sensing image. By setting the flight altitude, heading overlap, lateral overlap, and flight speed data of the drone, adjusting the RGB camera to select the automatic shooting mode, and setting the shooting time interval of the multispectral camera, before the drone enters the route and after the flight, it is necessary to take pictures of the radiation calibration plate for the calibration of the multispectral image reflectivity. The drone data collection selects a clear and cloudless time period with good lighting to obtain high-quality remote sensing images of the crops in the monitored planting area. Based on the visible light orthophoto and the multispectral orthophoto reflectivity map after radiation correction, the remote sensing image is interpreted to extract the calibration area r in the remote sensing image to be tested. c Common vegetation index x for crops within the range j Among them, the commonly used vegetation indices include the spectral parameters of the surface vegetation coverage summarized according to the spectral characteristics of vegetation chlorophyll, etc., which can effectively reflect the coverage and growth status of rice, including the normalized vegetation index NDVI, the green normalized vegetation index GNDVI, the renormalized vegetation index RDVI, and the ratio vegetation index RVI 1 / RVI2, which can effectively reflect the terrestrial chlorophyll index MTCI, the vegetation extraction color index C IVE, the overgreen vegetation index ExG, the green leaf vegetation index GLA, the plant pigment ratio PPR, the red, green and blue vegetation index RGBVI, the moisture index WI, the improved red edge ratio vegetation index MSR, and the vegetation attenuation index PSRI. The calculation formulas of each index are shown in the following table:
[0033] Table 1 Commonly used vegetation index calculation formulas
[0034]
[0035]
[0036] Among them, r, g, and b in the formula are the RGB sensor band data, r is the red band, g is the green band, and b is the blue band; R R , R G , R B , R NIR , R Red-edge is the multispectral sensor band data, R R is the red band, R Gv is the green band, R B is the blue band, R NIR is the near infrared band, R Red-edge It is the red edge band.
[0037] Step 3: Obtain the calibration area r c Average plant height of crops in the range Compared with the commonly used vegetation index x j Conduct correlation analysis and calculate the common vegetation indexes and average plant height The correlation coefficient between them is calculated as:
[0038]
[0039] In the formula, ρ i is the correlation coefficient between the i-th commonly used vegetation index and the average plant height; is the mean value of the i-th commonly used vegetation index; is the mean of the average plant height;
[0040] Step 4: Select the commonly used vegetation index with correlation coefficient within the correlation threshold range [0.5, 1] as the plant height characteristic index X j , expressed as the jth commonly used vegetation index, j = 1, 2, ..., m; where m ≤ n;
[0041] It should be noted that the common vegetation indices with correlation coefficients within the correlation threshold range [0.5, 1] show a linear strong positive correlation with the average plant height, and can be used as a characteristic quantity reflecting the average plant height;
[0042] Step 5: Construct the average plant height of crops in the planting area to be monitored Regression model with plant height characteristic index;
[0043] Step 6: Use the shooting height of the remote sensing image to be measured as the calibration shooting height H c , get the relative shooting height Hr Calibration area r c The contrast remote sensing image of crops within the range is extracted, and the calibration area r is c Plant height characteristic index X of crops within the range j ', calculate the shooting height difference ΔH and plant height characteristic index difference ΔX based on the acquired parameters j , where the shooting height difference ΔH and the plant height characteristic index difference ΔX j The calculation formulas are: ΔH=H r -H c ; ΔX j =X j '-X j , and the relative shooting height H r With the calibration shooting height H c Not equal;
[0044] Step 7: Take the normalized value of the shooting height difference ΔH as the input parameter and the plant height characteristic index difference ΔX j The normalized value of is used as the input parameter to build a neural network model, and the network model outputs the shooting height difference ΔH and the plant height characteristic index difference ΔX j The mapping relationship is as follows: x is the normalized parameter value, x 0 is the parameter value before normalization, x max 、x min They are the maximum and minimum values of the parameter values before normalization;
[0045] It should be noted that in this solution, the constructed neural network model is a BP neural network model with an input layer of one neuron, an intermediate layer of s neurons, and an output layer structure of j neurons, wherein the number of neurons s in the intermediate layer of the neural network model is determined according to the following formula: Wherein, a is an integer from 1 to 9;
[0046] Step 8: Obtain the non-calibrated area and the calibrated area r corresponding to the kth grid in the planting area to be monitored c The ground height difference Δh k , the calculation formula is: Δh k =h k -h c ; In the formula, h k 、h c They are respectively represented as the surface height values of the non-calibrated area and the calibrated area corresponding to the kth grid. According to the obtained surface height difference Δh k And mapping relationship, to correct various plant height characteristic indexes of the non-calibrated area of the monitored planting area in the remote sensing image. The specific process includes the following steps:
[0047] Get the non-calibrated area and the calibrated area r corresponding to the kth grid in the planting area to be monitored c The ground height difference Δh k ;
[0048] The height difference from the ground surface is Δh k The shooting height difference ΔH with the same value is used as the input parameter, and the plant height characteristic index difference ΔX of the non-calibrated area corresponding to the kth grid is obtained according to the mapping relationship. kj ;
[0049] Get the plant height characteristic index of the crop in the non-calibrated area corresponding to the kth grid in the remote sensing image to be tested
[0050] According to the obtained and plant height characteristic index difference ΔX j , calculate the corrected plant height characteristic index X of the crop in the non-calibrated area corresponding to the kth grid kj ,in,
[0051] The corrected plant height characteristic index was obtained, and the average plant height of crops in each non-calibrated area was determined according to the regression model.
[0052] In this embodiment, a method for obtaining a ground surface height value is provided, specifically:
[0053] Set the drone's flight altitude to a fixed value;
[0054] By carrying distance sensors to cruise the monitored planting area, the cruise route passes through the center of each grid area;
[0055] When the drone passes above the center of the grid area, the reading of the distance sensor is read as the ground height value of the area.
[0056] Example 2
[0057] The technical solution of the present invention can also be used for dynamic monitoring of plant height of other field crops, such as rice and wheat. The specific implementation process includes the following steps:
[0058] The planting area to be monitored is divided into grids with a step size of λ, and the calibration area r is selected. c , measure the calibration area r c Average plant height of crops in the range
[0059] Obtain the remote sensing image of the crops in the monitored planting area, and extract the calibration area r in the remote sensing image c Plant height characteristic index X of crops within the range j, construct the average plant height of crops in the monitored planting area and plant height characteristic index regression model, where the plant height characteristic index X j The acquisition process includes the following steps:
[0060] The shooting height of the remote sensing image to be measured is used as the calibration shooting height H c ;
[0061] Extract the commonly used vegetation index x of the calibration area in the remote sensing image to be tested i , represented by the i-th commonly used vegetation index, i=1,2,...,n; wherein the commonly used vegetation index includes at least one of the normalized vegetation index, green normalized vegetation index, renormalized vegetation index, ratio vegetation index, terrestrial chlorophyll index, vegetation extraction color index, overgreen vegetation index, green leaf vegetation index, plant pigment ratio, red-green-blue vegetation index, terrestrial chlorophyll index, moisture index, improved red edge ratio vegetation index, and vegetation attenuation index;
[0062] Calculate various commonly used vegetation indices and average plant height The correlation coefficient between them is calculated as:
[0063]
[0064] In the formula, ρ i is the correlation coefficient between the i-th commonly used vegetation index and the average plant height; is the mean value of the i-th commonly used vegetation index; is the mean of the average plant height;
[0065] Screen the commonly used vegetation indices whose correlation coefficients are within the correlation threshold range [0.5, 1] as the plant height characteristic index X j , expressed as the jth commonly used vegetation index, j = 1, 2, ..., m; where m ≤ n;
[0066] Get the relative shooting height H r Calibration area r c The contrast remote sensing image of crops within the range is extracted, and the calibration area r is c Plant height characteristic index X of crops within the range j ', where the relative shooting height H r With the calibration shooting height H c Not equal;
[0067] The shooting height difference ΔH is used as the input parameter, and the plant height characteristic index difference ΔX j The neural network model was constructed by taking the height difference ΔH and the plant height characteristic index difference ΔX as input parameters. j The calculation formulas are: ΔH=Hr -H c ; ΔX j =X j '-X j , the network model outputs the shooting height difference ΔH and the plant height characteristic index difference ΔX j The mapping relationship is as follows, wherein the neural network model is a BP neural network model having an input layer with one neuron, an intermediate layer with s neurons, and an output layer with j neurons. The number of neurons s in the intermediate layer of the neural network model is determined according to the following formula:
[0068]
[0069] Wherein, a is an integer from 1 to 9;
[0070] Get the non-calibrated area and the calibrated area r corresponding to the kth grid in the planting area to be monitored c The ground height difference Δh k , the calculation formula is: Δh k =h k -h c ; In the formula, h k 、h c They are respectively represented as the surface height values of the non-calibrated area and the calibrated area corresponding to the kth grid. According to the obtained surface height difference Δh k And mapping relationship, to correct various plant height characteristic indexes of the non-calibrated area of the monitored planting area in the remote sensing image. The specific process includes the following steps:
[0071] Get the non-calibrated area and the calibrated area r corresponding to the kth grid in the planting area to be monitored c The ground height difference Δh k ;
[0072] The height difference from the ground surface is Δh k The shooting height difference ΔH with the same value is used as the input parameter, and the plant height characteristic index difference ΔX of the non-calibrated area corresponding to the kth grid is obtained according to the mapping relationship. kj ;
[0073] Get the plant height characteristic index of the crop in the non-calibrated area corresponding to the kth grid in the remote sensing image to be tested
[0074] According to the obtained and plant height characteristic index difference ΔX j , calculate the corrected plant height characteristic index X of the crop in the non-calibrated area corresponding to the kth grid kj ,in,
[0075] The corrected plant height characteristic index was obtained, and the average plant height of crops in each non-calibrated area was determined according to the regression model.
[0076] It should be noted that for different field crops, the commonly used vegetation index types obtained are different, and the corresponding plant height characteristic index is also different. Therefore, in the actual operation process, it is necessary to select according to the characteristics of the monitored object, that is, the type of field crops, to determine the plant height characteristic index of the remote sensing image of the field crops to be monitored.
[0077] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.
Claims
1. A method for dynamically monitoring crop field growth and development, It is characterized in that include: The planting area to be monitored is divided into grids with a step size of λ, and the calibration area r is selected. c , measure the calibration area r c Average plant height of crops in the range Obtain the remote sensing image of the crops in the monitored planting area, and extract the calibration area r in the remote sensing image c Plant height characteristic index X of crops within the range j , construct the average plant height of crops in the monitored planting area Plant height characteristic index X j The regression model of Get the relative shooting height H r Calibration area r c The contrast remote sensing image of crops within the range is extracted, and the calibration area r is c Plant height characteristic index X of crops within the range j '; The shooting height difference ΔH is used as the input parameter, and the plant height characteristic index difference ΔX j The neural network model is constructed as input parameters, and the shooting height difference ΔH and plant height characteristic index difference ΔX are output through the network model. j The mapping relationship; Get the non-calibrated area and the calibrated area r corresponding to the kth grid in the planting area to be monitored c The ground height difference Δh k , according to the obtained surface height difference Δh k and mapping relationship, and correct various plant height characteristic indexes of the non-calibrated area of the monitored planting area in the remote sensing image; The corrected plant height characteristic index was obtained, and the average plant height of crops in each non-calibrated area was determined according to the regression model.
2. A method for dynamically monitoring crop field growth and development according to claim 1, It is characterized in that Plant height characteristic index X j The acquisition process includes the following steps: The shooting height of the remote sensing image to be measured is used as the calibration shooting height H c ; Extract the commonly used vegetation index x of the calibration area in the remote sensing image to be tested i , represented by the i-th commonly used vegetation index, i = 1, 2, ..., n; Calculate the commonly used vegetation index and average plant height The correlation coefficient between them is calculated as: In the formula, ρ i is the correlation coefficient between the i-th commonly used vegetation index and the average plant height; is the mean value of the i-th commonly used vegetation index; is the mean of the average plant height; Screen the commonly used vegetation indices whose correlation coefficients are within the correlation threshold range as the plant height characteristic index X j , represented by the jth commonly used vegetation index, j = 1, 2, ..., m; where m ≤ n.
3. A method for dynamically monitoring crop field growth and development according to claim 2, It is characterized in that The correlation threshold value ranges from [0.5, 1].
4. A method for dynamically monitoring crop field growth and development according to claim 1, It is characterized in that Relative shooting height H r Not equal to the calibrated shooting height H c .
5. A method for dynamically monitoring crop field growth and development according to claim 1, It is characterized in that Shooting height difference ΔH and plant height characteristic index difference ΔX j The calculation formulas are: ΔH=H r -H c ; ΔX j =X j '-X j 。 6. A method for dynamically monitoring crop field growth and development according to claim 1, It is characterized in that Surface height difference Δh k The calculation formula is: Δh k =h k -h c In the formula, h k 、h c They are respectively represented as the surface height values of the non-calibrated area and the calibrated area corresponding to the kth grid.
7. A method for dynamically monitoring crop field growth and development according to claim 1, It is characterized in that The commonly used vegetation indices include at least normalized vegetation index, green normalized vegetation index, renormalized vegetation index, ratio vegetation index, terrestrial chlorophyll index, vegetation extraction color index, overgreen vegetation index, green leaf vegetation index, plant pigment ratio, red, green and blue vegetation index, terrestrial chlorophyll index, moisture index, improved red edge ratio vegetation index, and vegetation attenuation index.
8. A method for dynamically monitoring crop field growth and development according to claim 1, It is characterized in that The plant height characteristic index correction process for non-calibrated areas includes the following steps: Get the non-calibrated area and the calibrated area r corresponding to the kth grid in the planting area to be monitored c The ground height difference Δh k ; The height difference from the ground surface is Δh k The shooting height difference ΔH with the same value is used as the input parameter, and the plant height characteristic index difference ΔX of the non-calibrated area corresponding to the kth grid is obtained according to the mapping relationship. kj ; Get the plant height characteristic index of the crop in the non-calibrated area corresponding to the kth grid in the remote sensing image to be tested According to the obtained and plant height characteristic index difference ΔX j , calculate the corrected plant height characteristic index X of the crop in the non-calibrated area corresponding to the kth grid kj ,in, 9. A method for dynamically monitoring crop field growth and development according to claim 1, It is characterized in that The neural network model is a BP neural network model having an input layer with one neuron, a middle layer with s neurons, and an output layer with j neurons, wherein the number of neurons s in the middle layer of the neural network model is determined according to the following formula: In the formula, a is an integer of 1-9.
