A tea bud yield monitoring method combining ground-air spectral technology

By combining near-Earth hyperspectral imaging technology and drone multispectral remote sensing technology, a tea bud yield model is constructed, which solves the accuracy and efficiency of tea yield monitoring, and realizes accurate monitoring and real-time update of tea bud yield.

CN118552870BActive Publication Date: 2024-11-01SOUTH CHINA AGRICULTURAL UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410997527.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-24
Publication Date
2024-11-01
Estimated Expiration
2044-07-24

AI Technical Summary

Technical Problem

The existing technology is difficult to quickly and accurately monitor the yield of large areas of tea. The near-Earth spectrum cameras measure slowly and have a small coverage area. The reflectivity information obtained by the drone equipped with the spectral camera is not accurate enough, resulting in low accuracy in measuring the yield of tea buds and leaves.

Method used

Using a combined method of ground-space spectroscopy technology, combining near-Earth hyperspectral imaging technology and drone multispectral remote sensing technology, the first and second tea bud yield models are constructed, and the optimal characteristic spectral data is obtained through hyperspectral images and drone image processing. , calculate the density and quantity of tea buds and leaves, combine the area index and characteristic spectrum data of tea buds and leaves to achieve accurate monitoring of tea buds and leaves yield.

Benefits of technology

It realizes accurate monitoring of tea bud leaf yield, improves tea bud yield monitoring efficiency, can update the final tea bud yield in real time, and is suitable for planning of tea picking and spring tea stir-frying and other links.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118552870B_ABST
    Figure CN118552870B_ABST
Patent Text Reader

Abstract

The present invention provides a tea bud yield monitoring method combined with ground-air spectral technology, including collecting hyperspectral images and drone images, picking tea buds on the tea ridge plane, recording the weight of the tea buds, and calculating the density of the tea buds according to the first type of tea buds and the second type of tea buds. The drone image is processed based on the hyperspectral image to obtain the optimal characteristic spectral data. A first tea bud yield model is constructed according to the tea bud density and the optimal characteristic spectral data, and a second tea bud yield model is constructed according to the number of the first type of tea buds, the number of the second type of tea buds and the optimal characteristic spectral data. Combined with the first tea bud yield model and the second tea bud yield model, the final tea bud yield is calculated. The above method combines near-ground hyperspectral imaging technology with drone multispectral remote sensing technology to accurately monitor the yield of tea buds. Since only one model needs to be established to monitor the changes in the final tea bud yield in real time, the above method has a high tea bud yield monitoring efficiency.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of tea bud yield monitoring and image processing, and in particular to a tea bud yield monitoring method combined with ground-air spectral technology. Background Art

[0002] In recent years, the consumption of tea has gradually increased. Currently, the estimation of tea production mostly relies on expert experience. There is a lack of methods for rapid estimation of remote sensing production on a large scale, which limits the planning of tea picking and spring tea frying.

[0003] The operation of measuring tea yield with near-ground spectral cameras is slow and covers a small area, while the spectral cameras carried by drones are not accurate enough in obtaining reflectance information and have low accuracy in measuring tea bud and leaf yield. Therefore, a method based on near-ground hyperspectral imaging technology and drone multispectral remote sensing technology is needed to accurately monitor tea bud and leaf yield and has a high tea bud yield monitoring efficiency. Summary of the invention

[0004] In order to overcome the problems existing in the related technology, the purpose of the present invention is to provide a tea bud yield monitoring method combined with ground-to-air spectral technology. The method combines near-ground hyperspectral imaging technology with unmanned aerial vehicle multispectral remote sensing technology to construct a first tea bud yield model and a second tea bud yield model, and performs weighted summation on the first tea bud yield model and the second tea bud yield model. The yield of tea buds and leaves can be accurately monitored, and the tea bud yield has a high monitoring efficiency.

[0005] A tea bud yield monitoring method using ground-air spectral technology, comprising:

[0006] Acquiring a hyperspectral image, wherein the hyperspectral image corresponds to a tea ridge plane including a marked position;

[0007] Picking tea buds and leaves on the plane of the tea ridge, and recording the weight of the tea buds and leaves;

[0008] Calculate the tea bud density based on the first-category tea bud and the second-category tea bud;

[0009] Collecting drone images, and processing the drone images based on the hyperspectral images to obtain optimal characteristic spectrum data; the drone images are used to reflect multispectral information;

[0010] Constructing a first tea bud yield model according to the tea bud leaf density and the optimal characteristic spectrum data;

[0011] Constructing a second tea bud yield model according to the number of the first type of tea buds, the number of the second type of tea buds and the optimal characteristic spectrum data;

[0012] The final tea bud yield is calculated by combining the first tea bud yield model and the second tea bud yield model.

[0013] In a preferred technical solution of the present invention, the first type of tea buds and leaves include one bud and one leaf, and the second type of tea buds and leaves include one bud and two leaves.

[0014] In a preferred technical solution of the present invention, the tea bud density is calculated based on the first type of tea buds and the second type of tea buds, comprising:

[0015] The tea bud density was calculated according to the following formula:

[0016] ;

[0017] in, is the tea bud leaf area index, is the tea bud leaf area, which is the sum of the areas of the tea bud, the first leaf under the tea bud and the second leaf under the tea bud in one bud and one leaf or one bud and two leaves. The area occupied by tea trees.

[0018] In a preferred technical solution of the present invention, the step of collecting drone images and processing the drone images based on the hyperspectral images to obtain optimal characteristic spectral data includes:

[0019] Preprocessing the hyperspectral image to obtain a preprocessed image; wherein the preprocessing includes black and white correction;

[0020] Extracting near-ground spectral reflectance information of the tea ridge plane according to the preprocessed image;

[0021] Acquire UAV images, perform radiation correction and atmospheric correction on the UAV images, and obtain multispectral information;

[0022] Based on the near-ground spectral reflectance information, the multispectral information is corrected using B-spline interpolation to obtain corrected multispectral information;

[0023] Constructing a partial least squares regression model based on the corrected multispectral information;

[0024] The characteristic spectral data with the largest modeling determination coefficient is screened out from the least squares regression model to obtain the optimal characteristic spectral data; wherein the characteristic spectral data includes any one of SPA characteristics and vegetation index.

[0025] In a preferred technical solution of the present invention, based on the near-ground spectral reflectance information, the multispectral information is corrected using B-spline interpolation to obtain the corrected multispectral information, including:

[0026] Using a third-order B-spline interpolation method to fit the near-ground spectral reflectance information to obtain a B-spline curve;

[0027] Determining a plurality of bands to be corrected in the multispectral information;

[0028] Searching for a target position corresponding to each of the to-be-corrected bands on the B-spline curve;

[0029] The multispectral information is corrected using the numerical value of the B-spline curve corresponding to the target position to obtain corrected multispectral information.

[0030] In a preferred technical solution of the present invention, the first tea bud yield model is constructed according to the tea bud leaf density and the optimal characteristic spectrum data, including:

[0031] The first tea bud yield model was constructed according to the following formula:

[0032] ;

[0033] in, The first tea bud yield, is the first parameter of the first tea bud yield model, is the second parameter of the first tea bud yield model, is the tea bud leaf area index, is the optimal characteristic spectrum data.

[0034] In a preferred technical solution of the present invention, the second tea bud yield model is constructed according to the number of the first type of tea buds, the number of the second type of tea buds and the optimal characteristic spectrum data, including:

[0035] The second tea bud yield model was constructed according to the following formula:

[0036] ;

[0037] in, For the second tea bud yield, is the first parameter of the second tea bud yield model, is the second parameter of the second tea bud yield model, is the third parameter of the second tea bud yield model, is the number of one bud and one leaf, is the number of one bud and two leaves.

[0038] In a preferred technical solution of the present invention, after collecting the hyperspectral image, the method further includes:

[0039] Acquire a visible light image taken by the hyperspectral camera;

[0040] The YOLOV5 model is used to identify the number of the first type of tea buds and the number of the second type of tea buds in the visible light image.

[0041] In a preferred technical solution of the present invention, the method of combining the first tea bud yield model and the second tea bud yield model to calculate the final tea bud yield includes:

[0042] Calculating the first tea bud yield according to the first tea bud yield model;

[0043] calculating the second tea bud yield according to the second tea bud yield model;

[0044] The first tea bud yield and the second tea bud yield are weightedly summed to obtain a final tea bud yield.

[0045] In a preferred technical solution of the present invention, after recording the weight of the tea buds, the method further comprises:

[0046] Recording the number of the first type of tea buds and the number of the second type of tea buds in the tea buds;

[0047] Spreading the tea buds and leaves on a test paper, and scanning them with a printer to obtain a scanned image of the tea buds and leaves;

[0048] Segmenting the tea bud leaf region from the tea bud leaf scanned image using the Otsu method;

[0049] Calculating the ratio of the number of pixels in the tea bud leaf area to the number of pixels in the scanned image of the tea bud leaf to obtain an area ratio;

[0050] The area ratio was multiplied by the area of ​​the test paper to obtain the tea bud leaf area.

[0051] The beneficial effects of the present invention are:

[0052] The present invention provides a tea bud yield monitoring method combined with ground-air spectral technology, which includes collecting a hyperspectral image, and the hyperspectral image corresponds to a tea ridge plane containing a marked position. The tea ridge plane includes a plurality of tea buds, and the tea buds are divided into a first-class tea bud and a second-class tea bud, and the first-class tea bud and the second-class tea bud have different morphological characteristics. Pick the tea buds on the tea ridge plane, record the weight of the tea buds, and calculate the tea bud density according to the first-class tea bud and the second-class tea bud. The weight and density of the tea buds can not only accurately grasp the growth of tea trees, but also provide a basis for remote sensing yield estimation of tea. Collect drone images, process the drone images, and obtain optimal characteristic spectral data. Construct a first tea bud yield model according to the tea bud density and the optimal characteristic spectral data, and construct a second tea bud yield model according to the number of the first-class tea buds, the number of the second-class tea buds, and the optimal characteristic spectral data. Combine the first tea bud yield model and the second tea bud yield model to calculate the final tea bud yield. The first type of tea bud leaf model is based on the tea bud leaf area index, and the second type of tea bud leaf model is based on the number of the first type of tea bud leaves and the number of the second type of tea bud leaves. The tea bud leaf area index, the number of the first type of tea bud leaves and the number of the second type of tea bud leaves can be combined to accurately monitor the tea bud leaf yield. The above method combines near-ground hyperspectral imaging technology with drone multispectral remote sensing technology. The drone can quickly collect images of the tea ridge plane to obtain drone images. After constructing the first tea bud yield model and the second tea bud yield model, the variables of the first tea bud yield model and the variables of the second tea bud yield model are continuously updated according to the hyperspectral image and the drone image, such as the number of one bud and one leaf, the number of one bud and two leaves, and the optimal characteristic spectral data, so as to update the final tea bud yield. Since the model only needs to be established once to monitor the changes in the final tea bud yield in real time, the above method has a high tea bud yield monitoring efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a schematic flow chart of the tea bud yield monitoring method combined with ground-air spectroscopy technology provided by the present invention;

[0054] Figure 2 It is a schematic diagram of the process of processing UAV images based on hyperspectral images provided by the present invention;

[0055] Figure 3 is a schematic diagram of a scanned image of a tea bud leaf provided by the present invention;

[0056] Figure 4 It is a schematic diagram of the tea bud leaf region provided by the present invention. DETAILED DESCRIPTION

[0057] The preferred embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the preferred embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to make the present invention more thorough and complete, and to fully convey the scope of the present invention to those skilled in the art.

[0058] Example 1

[0059] like Figure 1 As shown, this embodiment provides a tea bud yield monitoring method combining ground-air spectral technology, including:

[0060] S1: Acquire a hyperspectral image, wherein the hyperspectral image corresponds to a tea ridge plane including a marked position.

[0061] A near-ground hyperspectral camera was used to photograph the tea ridge plane including the marked position in clear weather. The wavelength range of the hyperspectral camera was 400-1000nm and the number of channels was 256. At the same time, a multispectral camera carried by a drone was used to photograph the tea ridge plane including the marked position at a height of 10-15 meters from the tea ridge plane. The number of channels of the multispectral camera was 6, namely blue, green, red, red edge, near infrared and visible light.

[0062] S2: picking tea buds and leaves on the plane of the tea ridge, and recording the weight of the tea buds and leaves.

[0063] All the first and second tea buds on the tea surface at the marked position are plucked, and the plucked tea buds are weighed to record the weight of the tea buds. This embodiment takes destructive sampling of tea buds and weighing the tea buds after destructive sampling to obtain the weight of the tea buds as an example.

[0064] S3: Calculate the tea bud density based on the first-category tea bud and the second-category tea bud.

[0065] The first type of tea buds and leaves include one bud and one leaf, and the second type of tea buds and leaves include one bud and two leaves.

[0066] The tea bud density was calculated according to the following formula:

[0067] ;

[0068] in, is the tea bud leaf area index, is the tea bud leaf area, which is the sum of the areas of the tea bud, the first leaf under the tea bud and the second leaf under the tea bud in one bud and one leaf or one bud and two leaves. The area occupied by tea trees.

[0069] For one bud and one leaf and one bud and two leaves, first calculate the sum of the areas of the tea bud, the first leaf under the tea bud, and the second leaf under the tea bud to get the total area, then calculate the ratio of the total area to the area occupied by the tea tree to get the tea bud leaf area index. Use a plumb line to measure 6 points along the edge of the tea tree, connect the adjacent points, and use graph paper to calculate the area occupied by the tea tree.

[0070] The tea bud leaf area index is related to the density, structure, biological characteristics of trees and environmental conditions of vegetation. It is a comprehensive indicator of the light energy utilization status and canopy structure of vegetation. The tea bud leaf area index is used to indicate the density of tea bud leaves.

[0071] S4: Collecting drone images, and processing the drone images based on the hyperspectral images to obtain optimal characteristic spectral data; the drone images are used to reflect multispectral information.

[0072] The near-ground spectral reflectance information is used to correct the multi-spectral information of the unmanned aircraft to obtain the corrected multi-spectral information. The corrected multi-spectral information is more accurate and can be used as a supplement to the near-ground spectral reflectance information.

[0073] S5: constructing a first tea bud yield model according to the tea bud leaf density and the optimal characteristic spectrum data.

[0074] The first tea bud yield model was constructed according to the following formula:

[0075] ;

[0076] in, The first tea bud yield, is the first parameter of the first tea bud yield model, is the second parameter of the first tea bud yield model, is the tea bud leaf area index, is the optimal characteristic spectrum data.

[0077] S6: constructing a second tea bud yield model according to the number of the first type of tea buds and leaves, the number of the second type of tea buds and leaves, and the optimal characteristic spectrum data.

[0078] The second tea bud yield model was constructed according to the following formula:

[0079] ;

[0080] in, For the second tea bud yield, is the first parameter of the second tea bud yield model, is the second parameter of the second tea bud yield model, is the third parameter of the second tea bud yield model, is the number of one bud and one leaf, is the number of one bud and two leaves.

[0081] S7: Calculate the final tea bud yield by combining the first tea bud yield model and the second tea bud yield model.

[0082] Step S7 includes the following steps S71-S73:

[0083] S71: Calculating the first tea bud yield according to the first tea bud yield model.

[0084] S72: Calculating the second tea bud yield according to the second tea bud yield model.

[0085] S73: performing weighted summation on the first tea bud yield and the second tea bud yield to obtain a final tea bud yield.

[0086] The first tea bud yield is multiplied by the first weight to obtain the adjusted first tea bud yield. The second tea bud yield is multiplied by the second weight to obtain the adjusted second tea bud yield. The adjusted first tea bud yield and the adjusted second tea bud yield are added to obtain the final tea bud yield. The sum of the first weight and the second weight is 1. In this embodiment, the first weight and the second weight are both set to 0.5 as an example.

[0087] After establishing the first tea bud yield model and the second tea bud yield model, it is only necessary to adjust the variables in the model according to the hyperspectral images and drone images, such as the number of one bud and one leaf, the number of one bud and two leaves, and the optimal characteristic spectral data, to update the final tea bud yield, thereby realizing real-time monitoring of the changes in the final tea bud yield. Therefore, the above method has a high tea bud yield monitoring efficiency.

[0088] The present embodiment provides a tea bud yield monitoring method combined with ground-air spectral technology, including collecting hyperspectral images, and the hyperspectral images correspond to tea ridge planes containing marked positions. The tea ridge plane includes a plurality of tea buds, and the tea buds are divided into first-class tea buds and second-class tea buds, and the first-class tea buds and second-class tea buds have different morphological characteristics. Pick the tea buds on the tea ridge plane, record the weight of the tea buds, and calculate the tea bud density based on the first-class tea buds and the second-class tea buds. The weight and density of the tea buds can not only accurately grasp the growth of tea trees, but also provide a basis for remote sensing yield estimation of tea. Collect drone images, process the drone images, and obtain optimal characteristic spectral data. Construct a first tea bud yield model based on the tea bud density and the optimal characteristic spectral data, and construct a second tea bud yield model based on the number of first-class tea buds, the number of second-class tea buds, and the optimal characteristic spectral data. Combine the first tea bud yield model and the second tea bud yield model to calculate the final tea bud yield. The first type of tea bud leaf model is based on the tea bud leaf area index, and the second type of tea bud leaf model is based on the number of the first type of tea bud leaves and the number of the second type of tea bud leaves. The tea bud leaf area index, the number of the first type of tea bud leaves and the number of the second type of tea bud leaves can be combined to accurately monitor the tea bud leaf yield. The above method combines near-ground hyperspectral imaging technology with drone multispectral remote sensing technology. The drone can quickly collect images of the tea ridge plane to obtain drone images. After constructing the first tea bud yield model and the second tea bud yield model, the variables of the first tea bud yield model and the variables of the second tea bud yield model are continuously updated according to the hyperspectral image and the drone image, thereby updating the final tea bud yield. Since the model only needs to be established once to monitor the changes in the final tea bud yield in real time, the above method has a high tea bud yield monitoring efficiency.

[0089] Example 2

[0090] This embodiment provides a tea bud yield monitoring method combined with ground-air spectroscopy technology. This embodiment only describes the differences from Example 1, such as Figure 2 As shown, the collecting of drone images and processing of the drone images based on the hyperspectral images to obtain optimal characteristic spectral data include:

[0091] S41: Preprocessing the hyperspectral image to obtain a preprocessed image; wherein the preprocessing includes black and white correction.

[0092] S42: Extracting the near-ground spectral reflectance information of the tea ridge plane according to the preprocessed image.

[0093] S43: Acquire a drone image, perform radiation correction and atmospheric correction on the drone image, and obtain multispectral information.

[0094] S44: Based on the near-ground spectral reflectance information, the multispectral information is corrected using B-spline interpolation to obtain corrected multispectral information.

[0095] S45: constructing a partial least squares regression model based on the corrected multispectral information.

[0096] S46: Filter out characteristic spectral data with the largest modeling determination coefficient from the least squares regression model to obtain optimal characteristic spectral data; wherein the characteristic spectral data includes any one of SPA characteristics and vegetation index.

[0097] Due to the presence of ambient lighting, atmosphere, environment, and sensor noise, it is necessary to perform black-and-white correction and noise removal on the hyperspectral image. The hyperspectral image is corrected for black-and-white according to the following formula:

[0098] ;

[0099] in, This is the image after black and white correction. is a hyperspectral image, Calibrate the image for the whiteboard, Correct the image for blackboard.

[0100] Binarize the black-and-white corrected image to obtain a binary image. Use a morphological algorithm to remove image noise in the binary image and fill the holes to obtain a mask image. Perform an AND operation on the black-and-white corrected image and the mask image to remove the background area in the black-and-white corrected image to obtain a preprocessed image. Extract the average reflectance information of the marked tea ridge from the preprocessed image to obtain the near-ground spectral reflectance information.

[0101] The drone is equipped with a hyperspectral camera, which can simultaneously capture visible light images and hyperspectral images of the marked tea ridges. The images obtained by the drone are corrected for radiation and atmosphere using DJI's supporting software DJI Maps, and the central wavelength position is set to (450, 560, 650, 730, 840) to obtain the corresponding reflectance information.

[0102] Step S44 includes the following steps:

[0103] S441: Use a third-order B-spline interpolation method to fit the near-ground spectral reflectance information to obtain a B-spline curve.

[0104] S442: Determine a plurality of bands to be corrected in the multispectral information.

[0105] S443: Searching for a target position corresponding to each of the to-be-corrected bands on the B-spline curve.

[0106] S444: Correct the multispectral information using the value of the B-spline curve corresponding to the target position to obtain corrected multispectral information.

[0107] Determine the central band of the multispectral information of the drone corresponding to the near-ground spectral reflectance information, select the information reflectance points of M channels on the left and right of the central band, and determine M bands that need to be corrected based on the information reflectance points of the M channels. In this embodiment, M=20 is taken as an example.

[0108] The interpolation method uses the values ​​of surrounding points to calculate the value of the interpolation point. The smoothness of the curves fitted by different interpolation methods is: third-order B-spline interpolation > bicubic interpolation > bilinear interpolation > nearest neighbor interpolation. Compared with piecewise linear interpolation, B-spline interpolation is differentiable at the nodes and has smoothness.

[0109] Each B-spline curve corresponds to a set of coefficients, and the coefficients of the B-spline curve are fitted using the least squares method. The least squares method finds the best function that matches the data by minimizing the sum of squares of errors. When the sample size is large, the least squares method can be used to fit the coefficients of the B-spline curve with higher accuracy, so that the B-spline curve can better fit the near-ground spectral reflectance information.

[0110] The multispectral information is corrected using the numerical value of the B-spline curve corresponding to the target position to obtain the corrected multispectral information, and the corrected multispectral information has a higher accuracy.

[0111] The present embodiment collects drone images, processes the drone images, and obtains optimal characteristic spectral data, including preprocessing the hyperspectral images to obtain preprocessed images, wherein the preprocessing includes black and white correction. According to the preprocessed image, the near-ground spectral reflectance information of the tea ridge plane is extracted, the drone image is obtained, and the drone image is subjected to radiation correction and atmospheric correction to obtain multispectral information. Based on the near-ground spectral reflectance information, the multispectral information is corrected using B-spline interpolation to obtain the corrected multispectral information. A partial least squares regression model is constructed based on the corrected multispectral information, and the characteristic spectral data with the largest modeling determination coefficient is screened out from the least squares regression model to obtain the optimal characteristic spectral data, wherein the characteristic spectral data includes any one of the SPA features and the vegetation index. Each B-spline curve corresponds to a set of coefficients, and the coefficients of the B-spline curve are fitted using the least squares method. The least squares method finds the best function that matches the data by minimizing the sum of squares of the errors. When the sample size is large, the coefficients of the B-spline curve with higher accuracy can be fitted using the least squares method, so that the B-spline curve better fits the near-ground spectral reflectance information.

[0112] Example 3

[0113] This embodiment provides a tea bud yield monitoring method combined with ground-air spectral technology. This embodiment only describes the differences from Embodiment 1. After collecting the hyperspectral image, it also includes:

[0114] S21': Acquire a visible light image taken by the hyperspectral camera.

[0115] S22': using the YOLOV5 model to identify the number of the first-category tea buds and the number of the second-category tea buds in the visible light image.

[0116] The YOLOV5 model includes a backbone network, a neck network and a head network. The backbone network is used to extract features from visible light images. The neck network is located between the backbone network and the head network. The neck network fuses features of different levels extracted by the backbone network. The neck network inputs the fused features into the head network, and the head network predicts based on the fused features and outputs the number of the first type of tea buds and the number of the second type of tea buds. The first type of tea buds in this embodiment includes one bud and one leaf, and the second type of tea buds includes one bud and two leaves.

[0117] The neck network adopts the PANet structure. PANet introduces a bottom-up path based on FPN. After the top-down feature fusion, the bottom-up feature fusion is performed. Through the bottom-up path enhancement, the accurate low-level positioning signal is used to enhance the entire feature hierarchy, thereby shortening the information path between the low-level and top-level features and enhancing the positioning capability at multiple scales.

[0118] After recording the weight of the tea buds, the method further comprises:

[0119] S31': recording the number of the first type of tea buds and the number of the second type of tea buds in the tea buds.

[0120] S32': Spread the tea buds and leaves on a test paper, and scan them with a printer to obtain a scanned image of the tea buds and leaves.

[0121] S33': Segmenting the tea bud and leaf area from the tea bud and leaf scanned image using the Otsu method.

[0122] S34': Calculate the ratio of the number of pixels in the tea bud leaf area to the number of pixels in the tea bud leaf scanned image to obtain an area ratio.

[0123] S35': multiply the area ratio by the area of ​​the test paper to obtain the tea bud leaf area.

[0124] like Figure 3-Figure 4As shown, this embodiment takes the test paper as A4 paper as an example, and uses the printer scanning method to obtain the tea bud leaf scanned image placed on the A4 paper. The Otsu method is also called the maximum inter-class difference method. The Otsu method uses a threshold to divide the entire tea bud leaf scanned image into two types of areas. The grayscale of the pixels in the first type of area is less than the threshold, and the grayscale of the pixels in the second type of area is greater than or equal to the threshold. If the variance of the grayscale of the pixels in these two types of areas is the largest, it means that the threshold used is the optimal threshold. Variance is a measure of the uniformity of grayscale distribution. The larger the inter-class variance between the background and the foreground, the greater the difference between the two parts of the image. When part of the foreground is wrongly divided into the background or part of the background is wrongly divided into the foreground, the difference between the two parts will become smaller. Therefore, the segmentation method with the largest inter-class variance is used to segment the tea bud leaf area from the tea bud leaf scanned image, and the foreground and the background can be distinguished to the greatest extent.

[0125] As an example, the number of pixels in the tea bud area is 100, and the number of pixels in the scanned image of the tea bud is 300, so the area ratio is 1 / 3. Multiply the area of ​​the A4 paper by 1 / 3 to get the area of ​​the tea bud.

[0126] After recording the weight of the tea buds, the present embodiment also includes recording the number of the first type of tea buds and the number of the second type of tea buds in the tea buds, spreading the tea buds on the test paper, scanning with a printer, and obtaining a scanned image of the tea buds. The tea bud area is segmented from the scanned image of the tea buds by the Otsu method, and the ratio of the number of pixels in the tea bud area to the number of pixels in the scanned image of the tea buds is calculated to obtain the area ratio. The area ratio is multiplied by the area of ​​the test paper to obtain the tea bud area. The distribution of the tea buds on the test paper is uneven, and the shape of the tea buds is also irregular. The ratio of the area of ​​the tea bud area to the area of ​​the scanned image of the tea buds is determined by image segmentation. After the tea bud area is segmented, the tea bud area can be calculated by a division operation and a multiplication operation, and the area of ​​the tea buds does not need to be measured, and the operation efficiency is high.

[0127] Example 4

[0128] This embodiment provides an electronic device, which includes a memory and a processor.

[0129] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0130] The memory may include various types of storage units such as system memory, read-only memory (ROM), and permanent storage.

[0131] The memory stores executable codes, and when the executable codes are processed by the processor, the processor can execute part or all of the tea bud yield monitoring method combined with ground-to-air spectral technology mentioned above.

[0132] Unless otherwise specifically stated, the relative arrangement, numerical expressions and numerical values ​​of the parts and steps set forth in these embodiments do not limit the scope of the application. In all examples shown and discussed here, any specific value should be interpreted as being merely exemplary, rather than as a restriction. Therefore, other examples of exemplary embodiments can have different values. It should be noted that similar reference numerals and letters represent similar items in the following drawings, and therefore, once a certain item is defined in an accompanying drawing, it does not need to be further discussed in the subsequent accompanying drawings.

[0133] It should be understood that spatially relative terms are intended to encompass different orientations of the device in use or operation in addition to the orientation depicted in the figures. For example, if the device in the figures is inverted, a device described as "above" or "above" other devices or structures would then be positioned "below" or "below" the other devices or structures. Thus, the exemplary term "above" can include both "above" and "below". The device may also be oriented in other different ways (rotated 90 degrees or at other orientations), and the spatially relative descriptions used herein should be interpreted accordingly.

[0134] In addition, it should be noted that the use of terms such as "first" and "second" to limit components is only for the convenience of distinguishing the corresponding components. Unless otherwise stated, the above terms have no special meaning and therefore cannot be understood as limiting the scope of protection of this application.

[0135] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A tea bud yield monitoring method using ground-air spectral technology, characterized in that: include: Acquiring a hyperspectral image, wherein the hyperspectral image corresponds to a tea ridge plane including a marked position; Picking tea buds and leaves on the plane of the tea ridge, and recording the weight of the tea buds and leaves; Calculate the tea bud density based on the first-category tea bud and the second-category tea bud; Collecting drone images, and processing the drone images based on the hyperspectral images to obtain optimal characteristic spectrum data; the drone images are used to reflect multispectral information; Constructing a first tea bud yield model according to the tea bud leaf density and the optimal characteristic spectrum data; Constructing a second tea bud yield model according to the number of the first type of tea buds, the number of the second type of tea buds and the optimal characteristic spectrum data; The final tea bud yield is calculated by combining the first tea bud yield model and the second tea bud yield model.

2. The tea bud yield monitoring method using ground-air spectroscopy technology according to claim 1 is characterized in that: The first type of tea buds and leaves include one bud and one leaf, and the second type of tea buds and leaves include one bud and two leaves.

3. The tea bud yield monitoring method using ground-air spectroscopy technology according to claim 2 is characterized in that: The method of calculating the tea bud density according to the first type of tea buds and the second type of tea buds comprises: The tea bud density was calculated according to the following formula: ; in, is the tea bud leaf area index, is the tea bud leaf area, which is the sum of the areas of the tea bud, the first leaf under the tea bud and the second leaf under the tea bud in one bud and one leaf or one bud and two leaves. The area occupied by tea trees.

4. The tea bud yield monitoring method using ground-air spectroscopy technology according to claim 1 is characterized in that: The collecting of drone images and processing of the drone images based on the hyperspectral images to obtain optimal characteristic spectral data include: Preprocessing the hyperspectral image to obtain a preprocessed image; wherein the preprocessing includes black and white correction; Extracting near-ground spectral reflectance information of the tea ridge plane according to the preprocessed image; Acquire UAV images, perform radiation correction and atmospheric correction on the UAV images, and obtain multispectral information; Based on the near-ground spectral reflectance information, the multispectral information is corrected using B-spline interpolation to obtain corrected multispectral information; Constructing a partial least squares regression model based on the corrected multispectral information; The characteristic spectral data with the largest modeling determination coefficient is screened out from the least squares regression model to obtain the optimal characteristic spectral data; wherein the characteristic spectral data includes any one of SPA characteristics and vegetation index.

5. The tea bud yield monitoring method using ground-air spectroscopy technology according to claim 4 is characterized in that: The method of correcting the multispectral information based on the near-ground spectral reflectance information by using B-spline interpolation to obtain corrected multispectral information includes: Using a third-order B-spline interpolation method to fit the near-ground spectral reflectance information to obtain a B-spline curve; Determining a plurality of bands to be corrected in the multispectral information; Searching for a target position corresponding to each of the to-be-corrected bands on the B-spline curve; The multispectral information is corrected using the value of the B-spline curve corresponding to the target position to obtain corrected multispectral information.

6. The tea bud yield monitoring method using ground-air spectroscopy technology according to claim 2 is characterized in that: The step of constructing a first tea bud yield model according to the tea bud leaf density and the optimal characteristic spectrum data comprises: The first tea bud yield model was constructed according to the following formula: ; in, The first tea bud yield, is the first parameter of the first tea bud yield model, is the second parameter of the first tea bud yield model, is the tea bud leaf area index, is the optimal characteristic spectrum data.

7. The tea bud yield monitoring method using ground-air spectroscopy technology according to claim 6 is characterized in that: The method of constructing a second tea bud yield model according to the number of the first type of tea buds, the number of the second type of tea buds and the optimal characteristic spectrum data comprises: The second tea bud yield model was constructed according to the following formula: ; in, For the second tea bud yield, is the first parameter of the second tea bud yield model, is the second parameter of the second tea bud yield model, is the third parameter of the second tea bud yield model, is the number of one bud and one leaf, is the number of one bud and two leaves.

8. The tea bud yield monitoring method using ground-air spectroscopy technology as claimed in claim 2 is characterized in that: After the hyperspectral image is collected, the method further includes: Obtain visible light images taken by a hyperspectral camera; The YOLOV5 model is used to identify the number of the first type of tea buds and the number of the second type of tea buds in the visible light image.

9. The tea bud yield monitoring method using ground-air spectroscopy technology according to claim 7 is characterized in that: The step of combining the first tea bud yield model and the second tea bud yield model to calculate the final tea bud yield comprises: Calculating the first tea bud yield according to the first tea bud yield model; calculating the second tea bud yield according to the second tea bud yield model; The first tea bud yield and the second tea bud yield are weightedly summed to obtain a final tea bud yield.

10. The tea bud yield monitoring method using ground-air spectroscopy technology according to claim 1, characterized in that: After recording the weight of the tea buds, the method further comprises: Recording the number of the first type of tea buds and the number of the second type of tea buds in the tea buds; Spreading the tea buds and leaves on a test paper, and scanning them with a printer to obtain a scanned image of the tea buds and leaves; Segmenting the tea bud leaf region from the tea bud leaf scanned image using the Otsu method; Calculating the ratio of the number of pixels in the tea bud leaf area to the number of pixels in the tea bud leaf scanned image to obtain an area ratio; The area ratio was multiplied by the area of ​​the test paper to obtain the tea bud leaf area.