Cotton effective leaf area monitoring method based on unmanned aerial vehicle optical image

Abstract

The invention discloses an unmanned aerial vehicle optical image-based cotton effective leaf area monitoring method. The method comprises the following steps of: acquiring a group of unmanned aerial vehicle images of a cotton canopy in a target area by utilizing an unmanned aerial vehicle carried optical sensor; segmenting the unmanned aerial vehicle image into a plurality of image grid units according to the unmanned aerial vehicle flight parameters and the sensor key parameters; calculating the porosity of the cotton canopy in each image grid unit by adopting a deep learning method; in the target growth period of cotton, the unmanned aerial vehicle is used for observing a cotton canopy from multiple observation angles, target leaf dip angle distribution parameters under each observation angle and an observation value G (theta) observation value of a G (theta) function are determined, and an effective leaf area index LAIe is calculated; wherein the G (theta) function is a function that an independent variable leaf dip angle distribution parameter x and an observation angle theta change, and a result of the G (theta) function represents a ratio of a projection area of a leaf in a canopy under the zenith angle theta to a projection area under the zenith angle 0; and carrying out geometric averaging on the LAIe at all observation angles to obtain a final effective leaf area index.

Description

Technical Field

[0001] The present invention relates to the field of agricultural technologies, and relates to, but is not limited to, a method for monitoring the effective leaf area of cotton based on unmanned aerial vehicle (UAV) optical images. Background Art

[0002] With the development of UAV technology, UAVs are increasingly widely used in the agricultural field. Precision agriculture requires efficient and accurate monitoring of crop growth conditions to optimize management decisions and improve crop yield and quality. Cotton is an important cash crop, and its leaf area index (LAI) is one of the important parameters for evaluating the growth status and yield of cotton. The effective leaf area (LAIe) refers to the leaf area with normal photosynthetic ability and is the core part of photosynthesis. Traditional methods for monitoring the effective leaf area of cotton, such as ground measurement and satellite remote sensing, have problems such as low measurement efficiency, high time cost, and low resolution. Therefore, it is of great significance to develop an efficient and accurate method for monitoring the effective leaf area of cotton based on UAV optical images.

[0003] Currently, the methods for monitoring the cotton leaf area mainly include the ground measurement method and remote sensing technology. The ground measurement method calculates the leaf area index by manually collecting sample leaf data. This method has high accuracy, but it has a large workload, takes a long time, and is not applicable to large-scale cotton fields. Satellite remote sensing technology uses satellite images to monitor the cotton leaf area. Although the coverage range is wide, it is limited by the resolution and frequency of satellite images and cannot obtain information on cotton fields in real time and with high resolution. In addition, satellite images are easily affected by factors such as clouds and climate conditions, which affect the accuracy of data. Summary of the Invention

[0004] In view of this, an embodiment of the present invention provides a method for monitoring the effective leaf area of cotton based on UAV optical images, which solves at least the problems of low measurement efficiency, high time cost, and low resolution in the prior art.

[0005] The technical solution of the embodiment of the present invention is implemented as follows:

[0006] An embodiment of the present invention provides a method for monitoring the effective leaf area of cotton based on UAV optical images, the method comprising:

[0007] Using a UAV equipped with an optical sensor to collect UAV images of cotton leaves in a target area; dividing the UAV images into a plurality of image grid units according to UAV flight parameters and key sensor parameters; wherein, the plurality of image grid units correspond to a plurality of observation angles; calculating the porosity of the cotton canopy in each of the image grid units by using a deep learning method; during the target growth period of cotton, using the UAV to observe the cotton canopy from the plurality of observation angles to determine the target leaf inclination distribution parameter and the observed value G(θ) of the G(θ) function at each of the observation angles观测 ; wherein, the G(θ) function is a function that varies with the leaf inclination distribution parameter x of the independent variable and the observation angle θ, and the result represents the ratio of the projected area of the leaves in the canopy at the zenith angle θ to the projected area at the zenith angle of 0; based on the G(θ) 观测 and the porosity, calculate the effective leaf area index LAI at each observation angle e , and then perform geometric mean to obtain the final effective leaf area index.

[0008] In some possible embodiments, the key parameters of the sensor include the diagonal field of view angle, the horizontal field of view angle, and the vertical field of view angle; the flight parameters of the UAV include the UAV flight altitude, the forward overlap rate, and the side overlap rate.

[0009] In some possible embodiments, both the forward overlap rate and the side overlap rate are set to 85%, and the UAV flight altitude is set to 10 to 30 meters.

[0010] In some possible embodiments, the step of dividing the UAV image into a plurality of image grid units according to the UAV flight parameters and the key parameters of the sensor includes: determining the actual ground length corresponding to the pixels of the UAV image according to the flight altitude of the UAV and the horizontal field of view angle of the sensor; determining the actual ground width corresponding to the pixels of the UAV image according to the flight altitude of the UAV and the vertical field of view angle of the sensor; determining the number N of image grid units in the horizontal / vertical direction according to the forward overlap rate and the side overlap rate of the UAV; dividing the actual ground length corresponding to the pixels and the actual ground width corresponding to the pixels into N equal parts respectively, and dividing the UAV image into N*N image grid units.

[0011] In some possible embodiments, the method further includes: determining the observation angle of the corresponding image grid unit according to the pixel row and column numbers of each pixel in the image grid unit from the geometric center of the image grid unit, the actual ground width corresponding to the pixel, the actual ground length corresponding to the pixel, and the UAV flight altitude; taking the geometric mean of the observation angles of all pixels in each image grid unit as the observation angle of the corresponding image grid unit.

[0012] In some possible embodiments, the step of using the UAV to observe the cotton canopy from the multiple observation angles and determining the leaf inclination distribution parameter and the observed value G(θ) of the G(θ) function at each observation angle 观测 , includes:

[0013] Calculate a set of actual observed values G’(θ) at different observation angles based on the observation data 观测 =-cos(θ)lnP(θ) 观测 ; wherein, P(θ) 观测is the porosity of the cotton canopy calculated by deep learning method;

[0014] Through the set of actual observation values G’(θ) 观测 Compare with G(θ) of multiple leaf inclination distribution parameters x in sequence 理论 Perform correlation comparison and select the most relevant G(θ) 理论 And deduce the actual leaf inclination distribution parameter x 实际 ;

[0015] Determine the empirical parameters of the 5 growth stages of cotton according to the expert experience method, including the minimum value x of the leaf inclination distribution parameter min And the maximum value x max , as well as the minimum porosity P(θ) min And the maximum porosity P(θ) max ;

[0016] Based on the empirical parameters, adjust the actual leaf inclination distribution parameter x 实际 And the porosity P(θ) 观测 To determine the final target leaf inclination distribution parameter x 综合 , porosity P(θ) 综合 ;

[0017] Substitute the target leaf inclination distribution parameter x 综合 And the observation angle θ into the following formula to calculate G(θ) 理论 As the practical G(θ) 观测 :

[0018]

[0019] Among them, the variable x represents the leaf inclination distribution parameter. When x>1 ε1 is a function of x, defined as ε1=(1 - x -2 ) 1 / 2 .

[0020] In some possible embodiments, the adjusting the actual leaf inclination distribution parameter x 实际 And the porosity P(θ) 观测 Based on the empirical parameters to determine the final target leaf inclination distribution parameter x 综合 , porosity P(θ) 综合 , including: if the leaf inclination distribution parameter x 实际 Is greater than the minimum value x min And less than the maximum value x max , take the leaf inclination distribution parameter x 实际 As the target leaf inclination distribution parameter x 综合 ; if the leaf inclination distribution parameter x实际 Less than or equal to the minimum value x min , with the minimum value x min as the target leaf inclination angle distribution parameter x 综合 ; if the leaf inclination angle distribution parameter x 实际 is greater than or equal to the maximum value x max , with the maximum value x max as the target leaf inclination angle distribution parameter x 综合 ; if the porosity P(θ) 观测 is greater than the minimum porosity P(θ) min and less than the maximum porosity P(θ) max , with the porosity P(θ) 观测 as the porosity P(θ) 综合 ; if the porosity P(θ) 观测 is less than the minimum porosity P(θ) min , with the minimum porosity P(θ) min as the porosity P(θ) 综合 ; if the porosity P(θ) 观测 is greater than or equal to the maximum porosity P(θ) max , with the maximum porosity P(θ) max as the porosity P(θ) 综合 .

[0021] In some possible embodiments, calculating the porosity of the cotton canopy in each of the image grid units by using a deep learning method includes: using a cotton canopy image sample data set with porosity labels, and performing porosity detection on each of the image grid units by combining a porosity detection model trained with a deep learning U-Net network to output a detection result image; calculating the porosity of each of the image grid units according to the porosity area and the unit image area in the detection result image.

[0022] In some possible embodiments, calculating the effective leaf area index LAI at each of the observation angles based on the G(θ) 观测 and the porosity respectively includes: substituting the G(θ) e and the porosity P(θ) 观测 into the following formula to calculate LAI 综合 : e :

[0023]

[0024] where θ is the observation angle.

[0025] The beneficial effects brought by the technical solutions provided by the embodiments of the present invention at least include:

[0026] In the embodiments of the present invention, first, an unmanned aerial vehicle (UAV) is used to carry an optical sensor to obtain high-resolution image data in real time. Compared with traditional ground measurements, the monitoring efficiency is greatly improved. At the same time, compared with satellite remote sensing, the influence of climate is reduced. Secondly, according to the UAV flight parameters and the key parameters of the sensor, the UAV images are segmented into multiple image grid units, so as to accurately calculate the inclination angle distribution of cotton leaves by multi-angle observation, and improve the accuracy of the effective leaf area index calculation. The UAV has a low flight altitude and is less affected by climate conditions, and can stably obtain data under various meteorological conditions. This method provides an efficient, accurate and economical method for monitoring the effective leaf area of cotton, which can obtain information on cotton fields in real time and with high resolution, is applicable to large-scale cotton fields and is not affected by climate, and has broad application prospects and popularization value. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to these drawings, where:

[0028] Figure 1 is a schematic flow chart of the method for monitoring the effective leaf area of cotton based on UAV optical images provided by the embodiments of the present invention;

[0029] Figure 2 is a schematic diagram of the sensor principle provided by the embodiments of the present invention;

[0030] Figure 3 is a schematic diagram of the imaging principle of the camera provided by the embodiments of the present invention;

[0031] Figure 4 is a schematic diagram of the flight scene and activity trajectory of the UAV provided by the embodiments of the present invention;

[0032] Figure 5 is a schematic diagram of the forward overlap rate during UAV flight provided by the embodiments of the present invention;

[0033] Figure 6 is a schematic diagram of the side overlap rate during UAV flight provided by the embodiments of the present invention;

[0034] Figure 7 is a schematic diagram of the segmentation effect of multiple observation angle units of UAV images provided by the embodiments of the present invention;

[0035] Figure 8 is a schematic diagram of the G(θ) function curve of different leaf inclination distribution parameters x provided by the embodiments of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. The following embodiments are used to illustrate the present invention, but are not used to limit the scope of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0037] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments and can be combined with each other without conflict.

[0038] It should be noted that the terms "first / second / third" involved in the embodiments of the present invention are only used to distinguish similar objects and do not represent a specific order for the objects. It can be understood that "first / second / third" can be interchanged with a specific order or sequence when permitted, so that the embodiments of the present invention described here can be implemented in an order other than that illustrated or described here.

[0039] Those skilled in the art of the present technology can understand that, unless otherwise defined, all terms (including technical terms and scientific terms) used here have the same meaning as the general understanding of those of ordinary skill in the technical field to which the embodiments of the present invention belong. It should also be understood that terms such as those defined in a general dictionary should be understood to have a meaning consistent with the meaning in the context of the prior art and will not be interpreted with an idealized or overly formal meaning unless specifically defined as here.

[0040] The embodiments of the present invention provide a method for monitoring the effective leaf area of cotton based on unmanned aerial vehicle (UAV) optical images, as Figure 1 shown, the method at least includes the following steps:

[0041] Step S110, using a UAV equipped with an optical sensor to collect UAV images of cotton leaves in the target area.

[0042] Here, the UAV can quickly cover a large area of cotton fields and obtain high-resolution UAV image data in real time. Compared with traditional ground measurements, the monitoring efficiency is greatly improved. At the same time, the UAV has a low flight altitude and is less affected by climate conditions. Compared with satellite remote sensing, it can stably obtain data under various meteorological conditions. The present invention collects and analyzes a group of UAV images with multiple side and horizontal overlaps.

[0043] Step S120: Segment the UAV images into multiple image grid units according to the UAV flight parameters and the key sensor parameters.

[0044] Here, the key UAV flight parameters include flight altitude, forward overlap rate, and side overlap rate, which affect the coverage range and stitching accuracy of the images.

[0045] The key sensor parameter, namely the field of view (FOV), is the viewing angle range that the sensor can capture, including the diagonal field of view (DFOV), horizontal field of view (HFOV), and vertical field of view (VFOV). The field of view is determined by the focal length (F), width (W), and height (H) of the sensor.

[0046] Based on the flight altitude and the field of view, the ground resolution corresponding to each pixel can be calculated. According to the forward and side overlap rates, the UAV images can be segmented into multiple image grid units. The multiple image grid units correspond to multiple observation angles, and each image grid unit corresponds to one observation angle. By using multi-angle observations, the inclination angle distribution of cotton leaves can be accurately calculated, improving the accuracy of leaf area index calculation.

[0047] Step S130: Use a deep learning method to calculate the porosity of the cotton canopy in each of the image grid units.

[0048] Here, the porosity of the cotton canopy refers to the proportion of the void part in the total area of the canopy. By segmenting the cotton canopy pores in the UAV images according to multiple image grid units, the porosity corresponding to the image grid units observed at each observation angle can be calculated. The present invention uses multi-angle observations and combines a deep learning model to accurately calculate the precise data of the cotton canopy pores in the image grid units, improving the accuracy of cotton effective leaf area index calculation.

[0049] There has been certain research and application of deep learning methods in the determination of porosity, including using neural network models such as U-net. These methods improve the accuracy and efficiency of porosity prediction by extracting pore structure features, optimizing model parameters, using specific loss functions, and optimization algorithms.

[0050] Step S140: During the target growth period of cotton, use the UAV to observe the cotton canopy from the multiple observation angles, and determine the target leaf inclination angle distribution parameter and the observed value G(θ) of the G(θ) function at each observation angle. 观测 。

[0051] Here, the target growth period of cotton is, for example, the budding stage, early flowering stage, full flowering stage, full boll stage, boll opening stage, etc.

[0052] Leaf inclination angle generally refers to the angle between the leaf and the horizontal plane, which is an important parameter describing the plant canopy structure. The spatial structure information of the leaves can be obtained through field measurement, image processing technology or using techniques such as three-dimensional lidar scanning, and then the leaf inclination angle can be calculated.

[0053] The G(θ) function is a function that varies with the independent variable of the leaf inclination angle distribution parameter x and the observation angle θ. Its result represents the ratio of the projected area of the leaves in the canopy at the zenith angle θ to the projected area at the zenith angle of 0. In the study of leaf inclination angle distribution, the G(θ) function is used to convert the leaf inclination angle distribution in three-dimensional space into a projection on a two-dimensional plane, which helps to simplify the problem and conduct visual analysis. For cotton, the leaf inclination angle is different at different growth stages. Currently, there are studies that can describe the G(θ) function to form the curve shape of G(θ)-θ under different leaf inclination angle distribution parameters x, such as Figure 8 shown, which shows the G(θ) function curves when x is 1.1, 1.5, 2, 3, 3.5, and 5 respectively. This step aims to obtain the observed value G(θ) of the corresponding G(θ) function under the target leaf inclination angle distribution parameter x that conforms to the actual situation. 观测 .

[0054] Step S140, calculate the effective leaf area index LAI at each of the observation angles based on the G(θ) 观测 and the porosity, and then perform geometric mean to obtain the final effective leaf area index. e

[0055] Here, the effective leaf area index (Effective Leaf Area Index, LAI e ) is the ratio of the area of the plant leaves on the vertical projection plane to the ground area of its layer. It is an important indicator for measuring vegetation coverage and is of great significance for understanding plant photosynthesis, water use efficiency, and the carbon cycle of the ecosystem, etc. In the embodiments of the present invention, the effective leaf area index at the target growth stage, that is, any growth stage, is calculated by combining the leaf inclination angle distribution parameter, G(θ) 观测 and the porosity.

[0056] In the embodiments of the present invention, first, an unmanned aerial vehicle (UAV) is used to carry an optical sensor to obtain high-resolution image data in real time. Compared with traditional ground measurement, the monitoring efficiency is greatly improved; at the same time, compared with satellite remote sensing, the influence of climate is reduced. Secondly, according to the UAV flight parameters and the key parameters of the sensor, the UAV images are segmented into multiple image grid units, so as to accurately calculate the inclination distribution of cotton leaves from multiple angles and improve the accuracy of leaf area index calculation. The UAV flies at a low altitude and is less affected by climate conditions, and can stably obtain data under various meteorological conditions. This method provides an efficient, accurate and economical method for monitoring the effective leaf area of cotton, which can obtain information on cotton fields in real time and with high resolution, is applicable to large-scale cotton fields and is not affected by climate, and has broad application prospects and popularization value.

[0057] In some possible embodiments, the key parameters of the sensor include the diagonal field of view angle, the horizontal field of view angle, and the vertical field of view angle; the UAV flight parameters include the UAV flight altitude, the heading overlap rate, and the side overlap rate.

[0058] Here, the field of view angle refers to the field of view range that the lens matches the CMOS (sensor) to be able to photograph. Specifically, it is the angle formed by two edges of the maximum range that the image of the measured target can be seen through the lens with the vertex of the lens of the optical instrument, as Figure 2 shown, which is divided into the horizontal field of view angle (HFOV), the vertical field of view angle (VFOV), and the diagonal field of view angle (DFOV). Among them, the diagonal field of view angle is the largest, the horizontal field of view angle is the second largest, and the vertical field of view angle is the smallest. Generally, the field of view angle usually refers to the diagonal field of view angle.

[0059] As Figure 3 shown is the imaging principle of the camera. The imaging width of each camera is fixed. For different focal lengths, the value of the field of view angle α is different, and the corresponding relationship is as follows: the longer the focal length, the smaller the field of view angle; the shorter the focal length, the larger the field of view angle. For the field of view range, the longer the focal length, the farther the field of view range; the shorter the focal length, the shorter the field of view range. The determining factors of the horizontal field of view angle are the focal length f and the width W of the imaging area; the determining factors of the vertical field of view angle are the focal length f and the height H of the imaging area; the determining factors of the diagonal field of view angle are the focal length f and the width W and height H of the imaging area. The specific calculation formulas are as follows:

[0060]

[0061] It is known that the diagonal size of the CMOS is 1 / 2.3” inches, and the diagonal length D is 16 mm. After conversion, 16 / 2.3” = 6.95 mm; it is known that the aspect ratio of the CMOS is 4:3. Using the converted diagonal length D of 6.95 mm, the values of the imaging height H and the imaging width W can be obtained.

[0062] Given the diagonal length of the CMOS and the diagonal field of view angle of the CMOS, Let σ represent the diagonal length, and the focal length f at the moment of taking a photo can be obtained. Then, according to formula (1), it can be calculated that: the horizontal field of view angle HFOV = 70.48 degrees; the vertical field of view angle WFOV = 55.8 degrees; the diagonal field of view angle DFOV = 82.8 degrees.

[0063] As Figure 4 shown is a schematic diagram of the flight scene and activity trajectory of the drone provided by an embodiment of the present invention. The forward overlap rate and the side overlap rate are usually used to describe the overlap degree between adjacent images and the overlap degree between images on the same flight line when the drone or aircraft is performing photogrammetry. The overlap rate is to ensure that there is sufficient information for stereo matching and image stitching.

[0064] In some possible embodiments, both the forward overlap rate and the side overlap rate are set to 85%, and the flight height of the drone is set to 10 to 30 meters.

[0065] Here, as Figure 5 shown, the forward overlap rate is expressed as where P x is the overlapping part of adjacent left and right images 1 and 2, and L x is the length of one image. As Figure 6 shown, the side overlap rate is expressed as where P y is the overlapping part of adjacent upper and lower images I-1 and II-1, and L y is the width of one image.

[0066] In the above embodiments, both the forward overlap rate and the side overlap rate set for the drone flight are 85%, which can reduce the blank area between images and improve the accuracy of subsequent image stitching; setting the flight height to 10 to 30 meters (m) can balance the image resolution and the coverage area.

[0067] In some possible embodiments, the above step 120 "segment the drone images into multiple image grid units according to the drone flight parameters and the key sensor parameters" is further implemented through the following process:

[0068] S121, determine the actual length corresponding to the pixels of the drone image according to the flight height of the drone and the horizontal field of view angle of the sensor.

[0069] Here, given that the flight height d is 30m, the horizontal field of view angle HVOF in the length direction is 70.48 degrees, and the number of pixels in the length direction is 4056 pixels, the actual length l corresponding to the pixels is calculated by the following formula to obtain the result 42.42:

[0070]

[0071] Further, dividing the actual length corresponding to the pixel by the number of pixels in the length direction gives a resolution of 0.0105 m, which usually refers to the actual ground length represented by each pixel in the image.

[0072] S122. Determine the actual ground width corresponding to the pixels of the UAV image according to the flight altitude of the UAV and the vertical field of view angle of the sensor.

[0073] Here, given that the flight altitude d is 30 m, the vertical field of view angle WVOF in the width direction is 55.8 degrees, and the number of pixels in the width direction is 3040 pixels, the actual ground width w corresponding to the pixels is calculated by the following formula to obtain the result 31.77:

[0074]

[0075] Further, dividing the actual ground width corresponding to the pixel by the number of pixels in the width direction gives a resolution of 0.0105 m, which usually refers to the actual ground width represented by each pixel in the image.

[0076] S123. Determine the number N of image grid units in the horizontal / vertical direction according to the forward overlap rate and the side overlap rate of the UAV.

[0077] Here, dividing the number of pixel units N according to the forward overlap rate and the side overlap rate, we can get The length of the UAV image is 4056 pixels, divided into 7 equal parts, with each pixel size being 580 pixels; the width is 3040 pixels, divided into 7 equal parts, with each pixel size being 434 pixels.

[0078] S124. Divide the actual ground length corresponding to the pixel and the actual ground width corresponding to the pixel into N equal parts respectively, and divide the UAV image into N*N image grid units.

[0079] Here, taking the UAV flight altitude of 30 m as an example, the actual ground length corresponding to the UAV pixel is 42.42 m, and the width is 31.77 m. The present invention is not limited thereto.

[0080]

[0081] In the above formula, dividing the actual ground length l, width w and N corresponding to the pixel by 2 makes the grid division smaller, making the image of each divided image grid unit clearer. Finally, the entire image area is divided into a grid of 7 rows * 7 columns of 6 m × 4.54 m, that is, 49 units of observation angles, as Figure 7 shown.

[0082] In some possible embodiments, the method further includes: determining the observation angle of a corresponding image grid unit according to the row and column numbers of each pixel from the geometric center of the image grid unit within each image grid unit, the actual field width corresponding to the pixel, the actual field length corresponding to the pixel, and the flight altitude of the drone; and taking the geometric mean of the observation angles of all pixels within each image grid unit as the observation angle of the corresponding image grid unit.

[0083] Here, according to the following formula, the observation angle of each pixel can be calculated as follows:

[0084]

[0085] In the formula, d is the flight altitude of the drone, θ is the observation angle. x and y are the row and column numbers of the pixel from the geometric center in the image, that is, its position in the image matrix, and w and l are the pixel width and pixel length of the image. For a rectangular image, the geometric center is usually the midpoint of the width and height. Using the flight altitude and field of view angle of the drone, these row and column numbers are converted into observation angles.

[0086] The image is segmented into multiple image grid units, and each image grid unit contains a certain number of pixels. For the pixels within each image grid unit, the average value of their observation angles is calculated as the observation angle of the image grid unit. This method can effectively associate each pixel in the image with the actual observation angle, providing a basis for subsequent analysis and processing.

[0087] In some possible embodiments, the above step S140 "using the drone to observe the cotton canopy from the multiple observation angles to determine the leaf inclination distribution parameter and the observed value G(θ) of the G(θ) function at each of the observation angles 观测 " is further implemented through the following steps:

[0088] S131, calculating a set of actual observed values G’(θ)obs = -cos(θ)lnP(θ) at different observation angles based on the observation data 观测 ; where P(θ) 观测 is the porosity of the cotton canopy calculated using a deep learning method;

[0089] Here, at different growth stages of cotton, the drone is used to take images of the cotton canopy from multiple observation angles to obtain image data. From the image data obtained from the drone, cos(θ) and P(θ) 观测 are extracted and the actual observed value G’(θ) 观测 is calculated, as shown in Table 1:

[0090] Table 1 Example of G'(θ) calculation at different observation angles

[0091] Observation angle θ <![CDATA[p(θ) 观测 > <![CDATA[ln p(θ) 观测 > cos(θ) G'(θ) 0 0.142 -1.9661 1 1.9661 8.5 0.143 -1.9661 0.9890 1.9445 10.55 0.151 -1.8971 0.9831 1.8650 16.7 0.152 -1.8971 0.9578 1.8171 21.8 0.155 -1.8643 0.9285 1.7311 22.55 0.164 -1.8325 0.9236 1.6926 31.5 0.165 -1.8325 0.8527 1.5627

[0092] It should be noted that in other methods, the empirical parameters of the five growth periods of cotton can also be combined to adjust P(θ) 观测 to ensure that the P(θ) substituted into the calculation 观测 is reasonable.

[0093] S132. By comparing the correlation of a set of actual observed values G’(θ) 观测 with G(θ) of multiple leaf inclination distribution parameters x in sequence 理论 to select the most relevant G(θ) 理论 and deduce the leaf inclination distribution parameter x that conforms to the actual situation 实际 .

[0094] Here, for a set of actual observed values G’(θ) 观测 , the correlation analysis is carried out with G(θ) of the preset multiple leaf inclination distribution parameters x 理论 . This usually involves statistical methods, such as the Pearson correlation coefficient, to measure the linear correlation between the observed values and the theoretical model. Select the leaf inclination distribution parameter with the highest correlation with the actual observed values as x 实际 . This can be achieved by maximizing the correlation coefficient.

[0095] It should be noted that LAI e is a constant, and there is a strong correlation between G’(θ) 观测 and G(θ) 观测 . At the same time, there is also a strong correlation between G(θ) 观测 and G(θ) 理论 . Therefore, the correlation between G’(θ) 观测 and G(θ) under multiple leaf inclination distribution parameters x can be compared. Select the leaf inclination distribution parameter corresponding to the maximum correlation as the leaf inclination distribution parameter x that conforms to the actual situation 理论 实际 . As shown in Table 2, the leaf inclination distribution parameter x 理论 corresponding to the most relevant G(θ) 实际 is 1.1. And based on x 实际 the corresponding G(θ) 观测。 min

[0096] Table 2 Comparison of the theoretical correlation between the observation of G’(θ) and G(θ) with different leaf inclination distribution parameters x

[0097]

[0098] S133. Determine the empirical parameters of the five growth periods of cotton according to the expert experience method, including the minimum value x of the leaf inclination distribution parameter minand the maximum value x max , as well as the minimum porosity P(θ) min and the maximum porosity P(θ) max .

[0099] Here, the empirical parameters for the five growth stages of cotton are shown in Table 3 below:

[0100] Table 3 Empirical parameters for the five growth stages of cotton

[0101]

[0102] S134. Based on the empirical parameters, adjust the actual leaf inclination distribution parameter x 实际 and the porosity P(θ) 观测 to determine the final target leaf inclination distribution parameter x 综合 , porosity P(θ) 综合 .

[0103] Here, the leaf inclination distribution parameter x obtained from step S133 实际 is 1.1. Combining with the empirical parameters of the five growth stages of cotton, the leaf inclination distribution parameter x during the full-bloom stage is between 1.4 and 2.2. After integrating the two, the most relevant target leaf inclination distribution parameter x 综合 = 1.4.

[0104] S135. Substitute the target leaf inclination distribution parameter x 综合 and the observation angle θ into G(θ) calculated by the following formula 理论 as the actual G(θ) 观测 :

[0105]

[0106] where the variable x represents the leaf inclination distribution parameter. When x > 1 ε1 is a function of x and is defined as ε1 = (1 - x -2 ) 1 / 2 .

[0107] It should be noted that the leaf inclination distribution parameter x can represent the degree of inclination of the leaf relative to the horizontal plane. A larger x value may mean that the leaf is more inclined to be perpendicular to the ground, while a smaller x value may mean that the leaf is more inclined to be horizontally distributed. ε1 is a function of x. When x approaches 1, ε1 approaches 0, indicating that the leaf is almost horizontally distributed; when x increases, ε1 also increases, indicating that the leaf is more inclined to be vertically distributed.

[0108] In the above embodiments, by combining expert experience and actual observation data, the calculation of the leaf inclination distribution parameter and the porosity is optimized, improving the adaptability and reliability of the method.

[0109] In some possible embodiments, the above step of "adjusting the actual leaf inclination angle distribution parameter x 实际 and the porosity P(θ) 观测 to determine the final target leaf inclination angle distribution parameter x 综合 and the porosity P(θ) 综合 " is further implemented through the following process: If the leaf inclination angle distribution parameter x 实际 is greater than the minimum value x min and less than the maximum value x max , then use the leaf inclination angle distribution parameter x 实际 as the target leaf inclination angle distribution parameter x 综合 ; if the leaf inclination angle distribution parameter x 实际 is less than or equal to the minimum value x min , then use the minimum value x min as the target leaf inclination angle distribution parameter x 综合 ; if the leaf inclination angle distribution parameter x 实际 is greater than or equal to the maximum value x max , then use the maximum value x max as the target leaf inclination angle distribution parameter x 综合 ; if the porosity P(θ) 观测 is greater than the minimum porosity P(θ) min and less than the maximum porosity P(θ) max , then use the porosity P(θ) 观测 as the porosity P(θ) 综合 ; if the porosity P(θ) 观测 is less than the minimum porosity P(θ) min , then use the minimum porosity P(θ) min as the porosity P(θ) 综合 ; if the porosity P(θ) 观测 is greater than or equal to the maximum porosity P(θ) max , then use the maximum porosity P(θ) max as the porosity P(θ) 综合 .

[0110] Exemplarily, during the cotton bud stage, the leaf inclination angle distribution parameter obtained through actual observation by an unmanned aerial vehicle is x 实际 = 2.0, the minimum value x min determined by expert experience is 1.7, and the maximum value x max is 2.4. The porosity calculated by the deep learning algorithm is P(θ) 观测 = 0.20, the minimum porosity P(θ) min determined by expert experience is 0.12, and the maximum porosity P(θ) max is 0.38.

[0111] For the leaf inclination distribution parameter, since x 实际 > x min and x 实际 < x max , so x 综合 = x 实际 = 2.0. For the porosity, since P(θ) 观测 > P(θ) min and P(θ) 观测 < P(θ) max , so P(θ) 综合 = P(θ) 观测 = 0.20.

[0112] This method combines the expert experience method and actual observation data. By setting reasonable thresholds, the parameters can be effectively adjusted and optimized to make them more in line with the actual observation situation.

[0113] In some possible embodiments, the above step S130 "calculating the porosity of the cotton canopy in each of the image grid units by using a deep learning method" is further implemented through the following process: by using a cotton canopy image sample data set with porosity labels, combining the porosity detection model trained by the deep learning U-Net network to perform porosity detection on each of the image grid units, and outputting a detection result image; calculating the porosity of each image grid unit according to the porosity area and the unit image area in the detection result image.

[0114] Here, first, make the porosity sample labels of the cotton canopy: use the UAV images to make the porosity sample labels of the cotton canopy, manually label the pores in the cotton canopy by using an image annotation tool (such as Labelme or VGG Image Annotator) to generate a binary label image, where the pore part is white (1) and the non-pore part is black (0). Finally, a set of cotton canopy image sample data sets with porosity labels are obtained, and these samples will be used for subsequent deep learning model training.

[0115] Then, divide the sample data set into a training set, a validation set, and a test set according to 8:1:1. Select the deep learning U-Net network, use the training set to train the model, verify the performance of the model on the validation set, and optimize the model by adjusting the hyperparameters (such as the learning rate and batch size, etc.). When evaluating the model on the test set, use indicators such as accuracy, recall rate, and F1 score for evaluation. Finally, a trained porosity detection model is obtained.

[0116] Finally, detect the cotton canopy pores in the UAV images and verify: After preprocessing and segmenting the grid of the UAV images, use the trained pore detection model to detect the pores in the preprocessed images and output the detection result images. Then calculate the porosity of each image grid unit according to the detection result images. The image grid unit is a fixed-size area divided in the preprocessed image, and each image grid unit contains multiple pixels. The porosity calculation formula is pore area / unit area. In other embodiments, the detection results can also be compared with the manually labeled images to verify the accuracy and reliability of the model.

[0117] In some possible embodiments, the above step S140 "Based on the G(θ) 观测 and the porosity, calculate the effective leaf area index LAI at each of the observation angles e ", includes:

[0118] Substitute the G(θ) 观测 and the porosity P(θ) 综合 into the following formula to calculate LAI e :

[0119]

[0120] where θ is the observation angle.

[0121] Here, by using multi-angle observation and deep learning techniques to accurately calculate the inclination distribution of cotton leaves and the canopy porosity, the accuracy of leaf area index calculation can be improved. At the same time, by combining expert experience and actual observation data to optimize the calculation of leaf inclination distribution parameters and porosity P(θ), the adaptability and reliability of the method can be improved.

[0122] For the target growth stage, obtain the observation data at multiple observation angles from the UAV images. For each observation angle, use formula (8) to calculate the effective leaf area index LAI at a single observation angle e , and substitute it into the geometric mean method formula to calculate the final effective leaf area index LAI e_avg . Through this method, the influences at different observation angles can be comprehensively considered to obtain a more comprehensive cotton effective leaf area index, providing more accurate decision-making support for agricultural production.

[0123] The cotton effective leaf area monitoring method based on UAV optical images provided by the embodiments of the present invention has the following advantages and values in using UAV technology combined with deep learning methods to monitor the cotton leaf area index:

[0124] (1) Improvement in monitoring efficiency: Drones can quickly cover large areas of cotton fields and obtain high-resolution image data. This not only saves time and human resources but also reduces errors caused by human factors.

[0125] (2) Improvement in monitoring accuracy: Multi-angle observations can capture the structural characteristics of the cotton canopy more comprehensively. Combining with deep learning techniques, the inclination distribution of cotton leaves and canopy porosity can be calculated more accurately. This helps to estimate the leaf area index more precisely.

[0126] (3) Reducing the impact of climate: Due to the relatively low flight altitude of drones, compared with satellite remote sensing, they are less affected by climate conditions. This means that drones can work stably under various meteorological conditions and obtain reliable data.

[0127] (4) Combination of expert experience and actual observation data: Combining expert experience and actual observation data can optimize the calculation of leaf inclination distribution parameters and porosity. This method improves the adaptability and reliability of the calculation results, making them more in line with the actual agricultural environment and crop growth conditions.

[0128] (5) Economy: Compared with traditional ground measurement and satellite remote sensing technologies, drone technology has advantages in terms of cost and operation difficulty. This makes this method more suitable for large-scale promotion and application.

[0129] (6) Application prospects and promotion value: This method can not only be used for monitoring the leaf area of cotton but also be extended to the monitoring of the leaf area of other crops, with broad application prospects. Its efficient and accurate monitoring capabilities are of great significance for aspects such as agricultural production management, pest control, and crop growth monitoring.

[0130] It should be noted that in the embodiments of the present invention, the above-mentioned method for monitoring the effective leaf area of cotton based on drone optical images can be implemented in the form of software function modules. When sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiments of the present invention, in essence, or the part that contributes to the related technology, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing an electronic device to execute all or part of the methods described in the various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROMs), magnetic disks, or optical discs that can store program codes. In this way, the embodiments of the present invention are not limited to any specific combination of hardware and software.

[0131] It should be understood that the "one embodiment" or "an embodiment" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present invention. Therefore, the "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present invention, the magnitudes of the serial numbers of the above processes do not mean the sequence of execution. The execution sequence of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention. The serial numbers of the embodiments of the present invention above are only for description and do not represent the advantages or disadvantages of the embodiments.

[0132] It should be noted that in this article, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including one..." does not exclude the presence of additional identical elements in the process, method, article or device including the element.

[0133] In several embodiments provided by the present invention, it should be understood that the disclosed method can be implemented in other ways. The methods disclosed in several method embodiments provided by the present invention can be combined arbitrarily without conflict to obtain new method embodiments. The features disclosed in several method embodiments provided by the present invention can be combined arbitrarily without conflict to obtain new method embodiments.

[0134] The above is only the implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A method for monitoring the effective leaf area of cotton based on UAV optical images, characterized in that, The method includes: Using an unmanned aerial vehicle (UAV) equipped with an optical sensor to collect UAV images of cotton leaves in the target area; Dividing the UAV images into multiple image grid units according to the UAV flight parameters and the key sensor parameters; wherein, the multiple image grid units correspond to multiple observation angles; Calculating the porosity of the cotton canopy in each of the image grid units by using a deep learning method; During the target growth period of cotton, a drone is used to observe the cotton canopy from the multiple observation angles, and the target leaf inclination distribution parameter and the observed value G(θ) of the G(θ) function are determined for each of the observation angles. 观测 Among them, the G(θ) function is a function that varies with the independent variable leaf inclination distribution parameter x and the observation angle θ, and its result represents the ratio of the projected area of the leaves in the canopy at the zenith angle of θ to the projected area at the zenith angle of 0. Based on the G(θ) 观测 and the porosity, calculate the effective leaf area index LAI at each of the observation angles e , and then perform geometric averaging to obtain the final effective leaf area index.

2. The method according to claim 1, wherein The key sensor parameters include the diagonal field of view, the horizontal field of view, and the vertical field of view; the UAV flight parameters include the UAV flight altitude, the heading overlap rate, and the side overlap rate.

3. The method according to claim 1, wherein Both the heading overlap rate and the side overlap rate are set to 85%, and the UAV flight altitude is set to 10 to 30 meters.

4. The method according to claim 2, wherein The dividing of the UAV images into multiple image grid units according to the UAV flight parameters and the key sensor parameters includes: Determining the actual ground length corresponding to the pixels of the UAV images according to the flight altitude of the UAV and the horizontal field of view of the sensor; Determining the actual ground width corresponding to the pixels of the UAV images according to the flight altitude of the UAV and the vertical field of view of the sensor; Determining the number N of image grid units in the horizontal / vertical direction according to the heading overlap rate and the side overlap rate of the UAV; Performing N equal divisions on the actual ground length corresponding to the pixels and the actual ground width corresponding to the pixels respectively, and dividing the UAV images into N*N image grid units.

5. The method according to any one of claims 1 to 4, characterized in that, The method further includes: Determining the observation angle of the corresponding image grid unit according to the pixel row and column numbers of each pixel in the image grid unit from the geometric center of the image grid unit, the actual ground width corresponding to the pixel, the actual ground length corresponding to the pixel, and the UAV flight altitude; Taking the geometric mean of the observation angles of all pixels in each image grid unit as the observation angle of the corresponding image grid unit.

6. The method according to any one of claims 1 to 4, characterized in that Using the drone to observe the cotton canopy from the multiple observation angles to determine the target leaf inclination distribution parameter and the observed value G(θ) of the G(θ) function at each of the observation angles 观测 , including: A set of actual observed values G’(θ) at different observation angles are calculated based on the observed data 观测 = -cos(θ)lnP(θ) 观测 ; where P(θ) 观测 is the porosity of the cotton canopy calculated by using the deep learning method Through the set of actual observation values G’(θ) 观测 successively compare with G(θ) of multiple leaf inclination distribution parameters x 理论 for correlation comparison, and select the most relevant G(θ) 理论 and deduce the leaf inclination distribution parameter x that conforms to the actual situation 实际 ; Determine the empirical parameters for the five growth stages of cotton according to the expert experience method, including the minimum value x min and the maximum value x max , as well as the minimum porosity P(θ) min and the maximum porosity P(θ) max ; Based on the empirical parameters, adjust the actual leaf inclination angle distribution parameter x 实际 and the porosity P(θ) 观测 to determine the final target leaf inclination angle distribution parameter x 综合 , the porosity P(θ) 综合 ; Substitute the target leaf inclination distribution parameter x 综合 and the observation angle θ into the following formula to calculate G(θ) 理论 as the practical G(θ) 观测 : Among them, the variable x represents the leaf inclination angle distribution parameter, under the condition of x > 1 ε1 is a function of x, defined as ε1 = (1 - x -2 ) 1 / 2 .

7. The method according to claim 6, wherein Based on the empirical parameters, the actual leaf inclination distribution parameter x 实际 and the porosity P(θ) 观测 are adjusted to determine the final target leaf inclination distribution parameter x 综合 , the porosity P(θ) 综合 , including: If the leaf inclination angle distribution parameter x 实际 is greater than the minimum value of x min and less than the maximum value of x max , take the leaf inclination angle distribution parameter x 实际 as the target leaf inclination angle distribution parameter x 综合 ; If the leaf inclination distribution parameter x 实际 is less than or equal to the minimum value x min , then use the minimum value x min as the target leaf inclination distribution parameter x 综合 ; If the leaf inclination angle distribution parameter x 实际 is greater than or equal to the maximum value x max , then use the maximum value x max as the target leaf inclination angle distribution parameter x 综合 ; If the porosity P(θ) 观测 is greater than the minimum porosity P(θ) min and less than the maximum porosity P(θ) max , take the porosity P(θ) 观测 as the said porosity P(θ) 综合 ; If the porosity P(θ) 观测 is less than the minimum porosity P(θ) min , then use the minimum porosity P(θ) min as the porosity P(θ) 综合 ; If the porosity P(θ) 观测 is greater than or equal to the maximum porosity P(θ) max , then the maximum porosity P(θ) max is taken as the porosity P(θ) 综合 .

8. The method according to any one of claims 1 to 4, characterized in that, The calculating of the porosity of the cotton canopy in each of the image grid units by using a deep learning method includes: Performing pore detection on each of the image grid units by using a pore detection model trained by combining a deep learning U-Net network with a cotton canopy image sample dataset with pore labels, and outputting a detection result image; Calculating the porosity of each image grid unit according to the pore area and the unit image area in the detection result image.

9. The method according to claim 6, characterized in that, Based on the G(θ) 观测 and the porosity, calculate the effective leaf area index LAI at each of the observation angles e , including: Substitute the G(θ) 观测 and the porosity P(θ) 综合 into the following formula to calculate the LAI e : Wherein, θ is the observation angle.

Images