Rice physical and chemical parameter extraction method fusing spectrum curve angle and spectrum index
By integrating the spectral curve angle and spectral index, a multi-term rice physical and chemical parameter extraction model was constructed, which solved the problem of high spatial and temporal dependence of rice physical and chemical parameter extraction, and achieved more efficient rice growth monitoring.
Patent Information
- Application Number
- CN202410978592.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-11-15
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing technology has a high spatial and temporal dependence when extracting the physical and chemical parameters of rice, and the impact of the observation altitude and time of the drone on the spectral data has not been fully studied, which has led to difficulties in obtaining physical and chemical information of large-scale rice.
Using the method of fusion spectral curve angle and spectral index, the multi-spectral image preprocessing of UAVs, calculating spectral index, simulating the canopy spectral reflectivity, constructing the spectral curve angle, and using the improved random forest regression model to construct the multi-year rice physics and chemistry Parameter extraction model.
The temporal and spatial dependence of the extraction of physical and chemical parameters of rice is reduced, the extraction accuracy and stability are improved, and it is adapted to rice growth monitoring at different scales.
Smart Images

Figure CN118968352A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of agricultural remote sensing, and in particular to a method for extracting rice physical and chemical parameters by integrating spectral curve angles and spectral indices. Background Art
[0002] Rice is one of the main food crops, and it is extremely important to effectively monitor the health of rice. Leaf or canopy physiological and biochemical information can be used as a good indicator to measure the growth of rice, reflecting its nutritional level and productivity information from different angles. For example, the leaf area index (LAI) is closely related to the ability to intercept photosynthetically active radiation and is a key variable to characterize photosynthesis, respiration and evaporation. Leaf chlorophyll content (LCC) reflects important information about the ability of crops to absorb intercepted photosynthetically active radiation and produce biomass (Aboveground biomass, AGB). LCC, which is elevated to the canopy level, namely canopy chlorophyll content (CCC), can evaluate the total canopy-scale productivity of crops. Timely and accurate acquisition of rice growth information can monitor its growth process in real time, thereby guiding field management such as fertilization and irrigation, and providing prior knowledge and estimation parameters for final yield estimation.
[0003] Many useful explorations have been made to obtain rice physical and chemical information by manual measurement and physical model inversion. However, due to the spatiotemporal variability of rice growth and the cost and regional restrictions, timely acquisition of large-scale rice physical and chemical information is challenging. In recent years, with the rapid development and integration of sensors, automatic control, computers, and aviation technologies, the use of drones and satellite remote sensing technology to extract physical and chemical information has become an important research direction for refined farmland management.
[0004] With the rapid development of remote sensing technology, breakthrough progress has been made in the rapid and non-destructive extraction of crop growth parameters in recent years. Since remote sensing images have the characteristics of high temporal and spatial resolution and unified image and spectrum, image features such as spectrum and spatial geometry are constantly being explored as important explanatory variables for the extraction of growth parameters. Spatial geometric information such as canopy height and texture commonly used in drone images is difficult to obtain in satellite images, which limits its application in the extraction of large-scale rice physical and chemical information. Spectral information can be easily obtained on both drone and satellite platforms. Therefore, spectral information shows great potential for the extraction of rice physical and chemical information at different scales.
[0005] When monitoring crop growth based on UAV remote sensing technology, we need to face the problem of spatial and temporal scales. When extracting crop physical and chemical parameters, the UAV detection method is usually planned based on the monitoring target, the size of the study area, the endurance of the UAV, etc. Most current studies have explained the flight altitude and time of the UAV, but they are somewhat blind, and few scholars have paid attention to the impact of different UAV observation altitudes and times on spectral data and monitoring targets. Therefore, there is no consensus on how to effectively reduce the spatiotemporal dependence of the rice physical and chemical parameter extraction model, and it is still worth further exploration. Summary of the invention
[0006] The invention provides a method for extracting physicochemical parameters of rice by fusing spectral curve angle and spectral index, so as to solve the defects existing in the prior art when extracting physicochemical parameters of rice.
[0007] The present invention provides a method for extracting rice physical and chemical parameters by integrating spectral curve angle and spectral index, comprising: Collecting UAV multispectral images of the study area, and preprocessing the UAV multispectral images to obtain UAV multispectral preprocessed images; Calculating the spectral index in the multi-spectral pre-processed image of the UAV based on the multi-band spectral reflectance; According to the rice leaf, soil and observation geometry parameters, the PROSAIL model was used to simulate the rice canopy spectrum, and the variation pattern of the rice canopy spectral reflectance with leaf area index and chlorophyll content was obtained. Constructing a spectral curve angle according to the variation law of the spectral reflectance of the rice canopy with the leaf area index and the chlorophyll content; Using RReliefF, the spectral index and the spectral curve angle are used to construct and train an improved random forest regression model to obtain a multi-period rice physical and chemical parameter extraction model; The multi-spectral image features extracted from the target area are input into the multi-stage rice physical and chemical parameter extraction model, wherein the multi-spectral image features include spectral index and spectral curve angle, and the multi-stage rice physical and chemical parameter extraction results are obtained.
[0008] According to a method for extracting rice physical and chemical parameters by fusing spectral curve angles and spectral indices provided by the present invention, a multispectral image of a research area is collected by a drone, and the multispectral image of the drone is preprocessed to obtain a multispectral preprocessed image of the drone, including: Determine the aerial photography method and aerial photography route of the drone according to the preset weather conditions of the study area, and obtain the multispectral image according to the aerial photography method and the aerial photography route; The multispectral image is subjected to geometric processing, radiation processing and spectral information extraction in sequence to obtain the multispectral preprocessed image of the UAV.
[0009] According to a method for extracting physical and chemical parameters of rice by fusing spectral curve angle and spectral index provided by the present invention, geometric processing, radiation processing and spectral information extraction are sequentially performed on the multispectral image to obtain the multispectral preprocessed image of the unmanned aerial vehicle, including: The geometric processing includes vignetting correction, band registration and distortion correction; The radiation processing includes radiation calibration, which is achieved by using a plurality of calibration blankets with different reflectivities and a piecewise linear calibration method; The spectral information extraction includes the delineation of regions of interest, the extraction of reflectance and the operation of bands.
[0010] According to a method for extracting rice physical and chemical parameters by fusing spectral curve angle and spectral index provided by the present invention, the spectral index in the multi-spectral preprocessed image of the unmanned aerial vehicle is calculated based on multi-band spectral reflectance, including: Acquire multiple reflectances of the multispectral preprocessed image in different bands; According to the multiple reflectivities, a difference spectral index, a ratio spectral index, a normalized spectral index and an improved spectral index are calculated; The difference spectral index includes the difference vegetation index, the ratio spectral index includes the ratio vegetation index, the normalized spectral index includes the normalized difference vegetation index and the normalized red edge index, and the improved spectral index includes the visible light atmospheric impedance index, the two-band enhanced vegetation index, the red chlorophyll index, the green chlorophyll index, the modified chlorophyll absorption reflectance index and the wide dynamic range vegetation index.
[0011] According to a rice physical and chemical parameter extraction method that integrates spectral curve angle and spectral index provided by the present invention, the PROSAIL model is used to simulate the rice canopy spectrum according to rice leaves, soil and observation geometry input parameters, and the variation law of rice canopy spectral reflectance with leaf area index and chlorophyll content is obtained, including: The PROSAIL model is obtained by coupling the leaf reflectance model PROSPECT-5B and the canopy reflectance model SAIL; The input parameters of the PROSPECT-5B model include leaf structural parameters, leaf chlorophyll content, carotenoid content, brown pigment content, leaf water content, and leaf dry matter content; The SAIL model input parameters include leaf area index, mean leaf inclination, hot spot parameters, solar zenith angle, observation zenith angle, soil brightness parameters, and observation relative azimuth; Obtaining the reflectivity of the rice leaf according to the leaf structure parameters, the leaf chlorophyll content, the carotenoid content, the brown pigment content, the leaf water content and the leaf dry matter content; The canopy spectral reflectance is obtained according to the rice leaf reflectance, the leaf area index, the average leaf inclination angle, the hot spot parameter, the solar zenith angle, the observation zenith angle, the soil brightness parameter and the observation relative azimuth.
[0012] According to a method for extracting rice physical and chemical parameters by fusing a spectral curve angle and a spectral index provided by the present invention, a spectral curve angle is constructed according to the spectral reflectance of the rice canopy, comprising: Based on the spectral reflectance of the rice canopy, obtaining a first reflectance, a second reflectance, a first wavelength and a second wavelength of any two bands; A first variable is obtained from the first reflectivity and the second reflectivity, and a second variable is obtained from the first wavelength and the second wavelength; The first variable and the second variable are calculated using an inverse tangent function to obtain the spectral curve angle.
[0013] According to a method for extracting rice physical and chemical parameters by fusing spectral curve angle and spectral index provided by the present invention, RReliefF is adopted to construct and train an improved random forest regression model using the spectral index and the spectral curve angle to obtain a multi-period rice physical and chemical parameter extraction model, including: Inputting the spectral index and the random linear and nonlinear combination of the spectral curve angle into the random forest model to obtain the improved random forest regression model; A training set is formed by the spectral index and the spectral curve angle, and RReliefF is used to randomly select any sample in the training set, and search and determine adjacent samples of the same category and adjacent samples of different categories of the any sample; Determine any two samples of the continuous eigenvalue, and obtain the difference between the two samples on the continuous eigenvalue by the ratio of the absolute value of the difference between the two samples and the difference between the maximum value and the minimum value of the continuous eigenvalue; Calculate the difference between the two samples on the continuous feature value, the approximate probability distribution of the adjacent samples of the same category and the adjacent samples of different categories to obtain the continuous feature weight; Convert the difference between the two arbitrary samples on the continuous feature value into a resampling probability of two similar samples, and use the resampling probability to obtain an updated continuous feature weight; The training set is screened based on the updated continuous feature weights, and the improved random forest regression model is trained using the screened feature variables to obtain an initial multi-period rice physicochemical parameter extraction model; The determination coefficient, root mean square error and relative root mean square error were used to estimate the model accuracy of the multi-period rice physicochemical parameter extraction model, and an optimized multi-period rice physicochemical parameter extraction model was obtained.
[0014] In a second aspect, the present invention further provides a rice physical and chemical parameter extraction system integrating spectral curve angle and spectral index, comprising: An acquisition preprocessing module is used to acquire UAV multispectral images of a research area, and preprocess the UAV multispectral images to obtain UAV multispectral preprocessed images; A calculation module, used for calculating the spectral index in the multi-spectral pre-processed image of the UAV based on the multi-band spectral reflectance; The simulation module uses the PROSAIL model to simulate the spectrum of the rice canopy based on the rice leaf, soil and observation geometry input parameters, and obtains the variation pattern of the spectral reflectance of the rice canopy with the leaf area index and chlorophyll content; A construction module, used to construct a spectral curve angle according to the variation law of the spectral reflectance of the rice canopy with the leaf area index and the chlorophyll content; A training module is used to construct and train an improved random forest regression model using the spectral index and the spectral curve angle by using RReliefF to obtain a multi-period rice physical and chemical parameter extraction model; The extraction module is used to input the multi-spectral image features extracted from the target area into the multi-stage rice physical and chemical parameter extraction model, wherein the multi-spectral image features include spectral index and spectral curve angle, to obtain the multi-stage rice physical and chemical parameter extraction results.
[0015] In a third aspect, the present invention also provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, it implements any of the above-described methods for extracting physical and chemical parameters of rice by fusing spectral curve angles and spectral indices.
[0016] In a fourth aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method for extracting physical and chemical parameters of rice by fusing spectral curve angles and spectral indices as described in any one of the above.
[0017] The method for extracting rice physical and chemical parameters by integrating spectral curve angle and spectral index provided by the present invention proposes and constructs SCA based on the analysis of the spectral variation law of rice canopy when simulating different leaf area indexes and chlorophyll contents based on the PROSAIL model, and fully integrates spectral index and SCA by improving the machine learning algorithm, realizing feature enhancement and variable screening, thereby improving the accuracy and stability of rice physical and chemical parameter extraction. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0019] Figure 1 It is a schematic flow chart of a method for extracting rice physical and chemical parameters by fusing spectral curve angle and spectral index provided by the present invention; Figure 2 Schematic diagram of the structure of the rice physical and chemical parameter extraction system integrating the spectral curve angle and the spectral index provided by the present invention; Figure 3 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION
[0020] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0021] Figure 1 FIG. 1 is a flow chart of a method for extracting rice physical and chemical parameters by fusing spectral curve angle and spectral index provided by an embodiment of the present invention. Figure 1 As shown, including: Step 100: Collecting drone multispectral images of the study area, and preprocessing the drone multispectral images to obtain drone multispectral preprocessed images; Step 200: Calculate the spectral index in the multi-spectral pre-processed image of the UAV based on the multi-band spectral reflectance; Step 300: According to the rice leaf, soil and observation geometry input parameters, the PROSAIL model is used to simulate the rice canopy spectrum to obtain the variation law of the rice canopy spectral reflectance with the leaf area index and the chlorophyll content; Step 400: constructing a spectral curve angle according to the variation law of the spectral reflectance of the rice canopy with the leaf area index and the chlorophyll content; Step 500: using RReliefF, using the spectral index and the spectral curve angle to construct and train an improved random forest regression model to obtain a multi-period rice physical and chemical parameter extraction model; Step 600: Input the multispectral image features extracted from the target area into the multi-stage rice physical and chemical parameter extraction model, wherein the multispectral image features include spectral index and spectral curve angle, to obtain the multi-stage rice physical and chemical parameter extraction results.
[0022] The embodiment of the present invention utilizes an improved machine learning algorithm to fuse the spectral curve angle and the spectral index to adapt to the extraction of rice growth parameters at different time and space scales. The specific technical solution is as follows: 1. UAV data acquisition and preprocessing The UAV remote sensing data of the present invention is obtained by S1000 UAV (S1000, SZ DJI Technology, Co., Ltd., Shenzhen, China) equipped with MCA series multispectral camera (Tetracam, Inc., Chatsworth, CA, USA). The UAV flight mission is carried out under clear, cloudless and windless conditions as much as possible. The multispectral camera used in the present invention is MCA-12 with 12 bands (the central bands are: 490, 520, 550, 570, 670, 680, 700, 720, 800, 850, 900, 950 nm). Before the UAV takes images, the UAV is first controlled to fly to the center of the study area, and the same fixed exposure time (set to 1 millisecond) is set. After adjusting the shooting direction, only vertical lifting and lowering operations are performed, and the height at which all rice fields in the study area are included in the image is the lowest UAV flight height, and the shooting control at multiple heights is completed within 3 minutes to ensure that the influence of solar radiation changes is minimized. When drones are used for shooting at different times, the center of the drone is selected as accurately as possible at the same position to ensure that the image positions obtained at different times are the same. Remote sensing images are all obtained by shooting single images to ensure that there is no light change caused by flight routes. Therefore, the drone images in the present invention do not involve image stitching, and the research results are expected to be applied on a satellite scale.
[0023] The UAV multispectral image processing process includes three parts: geometric processing, radiation processing and spectral information extraction. The geometric processing process of the UAV multispectral image acquired by the MCA camera includes vignetting correction, band registration and distortion correction, etc. This process is completed in the MCA image dedicated processing software PixelWrech2 (Tetracam, Inc., Chatsworth, CA, USA). Radiation processing is mainly a radiation calibration process. The present invention uses 8 calibration blankets with known reflectivity (reflectivities are 3%, 6%, 12%, 24%, 36%, 48%, 56% and 80% respectively) for radiation calibration, and uses a piecewise linear calibration method for radiation calibration. Spectral information extraction mainly includes the delineation of the region of interest (each field block is the region of interest except the edge part), reflectivity extraction (reflectivity mean within the region of interest) and band operation.
[0024] 2. Spectral Index Calculation The spectral index calculated based on spectral reflectance (two or more bands) has been proven by many scholars to have a very good effect in estimating plant growth parameters because it can enhance vegetation characteristics. The spectral index selected by the present invention includes ratio, difference, normalization and improved spectral index. The specific calculation formula is as follows: Difference Vegetation Index: DVI=R 850nm -R 670nm Ratio vegetation index: RVI=R 850nm / R 670nm Normalized Difference Vegetation Index: NDVI = (R 850nm -R 670nm ) / (R 850nm + R 670nm ) Normalized Red Edge Index: NDRE=(R 850nm -R 720nm ) / (R 850nm + R 720nm ) Visible atmospheric impedance index: VARI=(R 550nm -R 670nm ) / (R 550nm + R 670nm ) Two-band enhanced vegetation index: EVI2=2.5 (R 850nm -R 670nm ) / (R 850nm + 2.4 R 670nm + 1) Red chlorophyll index: CI red edge=R 850nm / R 720nm -1 Green chlorophyll index: CI green =R 850nm / R 550nm -1 Modified chlorophyll absorption reflectance index: MCARI=[(R 700nm -R 670nm )-0.2 (R 700nm -R 550nm )] / (R 700nm / R 670nm ) Wide Dynamic Range Vegetation Index: WDRVI=(0.2 R 850nm -R 670nm ) / (0.2 R 850nm + R 670nm ) In the formula, R 550 , R 670 , R 700 , R 720 , R 850 They represent the reflectance of the multispectral image in the corresponding bands.
[0025] 3. PROSAIL model simulates rice canopy spectrum The PROSAIL model has a good effect in simulating the directional spectrum of vegetation because it takes into account the influence of soil background, the leaf properties of vegetation and the geometric structure characteristics of the canopy. It is developed by coupling the leaf reflectance model PROSPECT and the canopy reflectance model SAIL. The present invention couples the PROSPECT-5B and 4SAIL models to obtain a physical model to simulate the multi-angle and multi-time spectral data of the rice canopy.
[0026] According to the input of different variables, the calculation process of rice canopy reflectance is as follows:
[0027]
[0028] Among them, R(leaf) and R(canopy) are the spectral reflectance of rice leaves and canopy, respectively, and the different variables in brackets are the corresponding model input parameters in Table 1.
[0029] Table 1 Input parameter settings of PROSAIL model
[0030] Using MATLAB 2023a as the platform, the measured soil spectrum is used to replace the soil spectrum data provided by the model. The PROSAIL model, which takes dynamic leaf area index, chlorophyll content, leaf inclination, solar azimuth, observation zenith angle and observation azimuth as input parameters, simulates the multi-angle, multi-time and multi-period rice canopy spectral data.
[0031] 4. Spectral curve angle (SCA) construction The SCA proposed in the present invention is defined as the angle between the line connecting the positions of the reflectances corresponding to the two bands in the wavelength-reflectance coordinate system and the horizontal direction. According to the response analysis of the spectral reflectance to the changes in leaf area index and chlorophyll content, different SCAs are constructed to extract the physical and chemical parameters of rice. Different SCAs need to be defined for different rice growth parameters. The SCA used to estimate the LAI and CCC of rice is α, and the SCA used to estimate SPAD is β. The SCA can be used to quantitatively analyze the spectral curve characteristics of the rice canopy. According to different band combinations, different SCAs are recorded as SCA (b1, b2) or SCA (b3, b4). For example, α can be expressed as SCA (670, 800) and β can be expressed as SCA (550, 670). The calculation formula is as follows: SCA = arctan(Δy / Δx) Δy = R 2 -R 1 Δx = (wav2-wav1) / 2500 Where R 1 and R 2 represents the reflectance of the selected band, and wav1 and wav2 represent the wavelength of the selected band.
[0032] 5. Extraction of physicochemical parameters of rice in multiple periods The present invention achieves high-precision extraction of different rice physicochemical parameters by improving the random forest regression algorithm (RFR), called feature-enhanced random forest regression (FE-RFR), and compares the model accuracy with RFR, GBDT, and XGBoost algorithms.
[0033] First, a higher dimensional feature space is created by adding random linear and nonlinear combinations of input variables in the random forest. Since the main factors affecting the error of the random forest algorithm are the strength of the predictor in the forest and the correlation between trees, more randomness can be introduced through the combination of the original variables (spectral index and SCA), and the similarity of decision trees can be reduced, thereby improving the model accuracy and enhancing the generalization ability of the model.
[0034] Secondly, the importance of different features is evaluated and screened by the RReliefF algorithm. For machine learning, not all variables are equally important. RReliefF is a typical nonlinear system feature selection method based on statistical correlation. It can be used to deal with regression problems where multiple features interact and are interrelated and the target attribute is a continuous value. By randomly selecting a sample R in the training set and then searching for its adjacent samples (H for the same category and M for different categories), the weight of each feature is updated according to the distance between R, H and M. For continuous feature values, two samples R of feature F are equal. 1 and R 2 Dif (F, R 1 ,R 2 ) is defined as:
[0035] Among them, Max(F) and Min(F) represent the maximum and minimum values of feature F respectively, and the Dif() function is used to calculate the distance between instances and find the nearest neighboring samples.
[0036] The weight W(F) of feature F can be given by the following approximate probability distribution:
[0037] in, and represents conditional probability.
[0038] In order to transform this classification problem into a regression problem, two resampling probabilities are introduced to determine whether different samples are in the same class. The determination of the probability makes it possible to model and predict the corresponding interval between the two resamplings. The introduced resampling probabilities are as follows:
[0039]
[0040]
[0041] Among them, P 1 and P 2 They are the modeling and predicted values of the distance probability of two similar samples, which are conditional probabilities.
[0042] According to the combination of the above calculation formulas, the weight of feature F can be obtained:
[0043] Finally, a rice physical and chemical parameter extraction model based on the feature enhanced random forest algorithm was established. Multiple different decision trees were trained by the final feature variable set screened by the RReliefF algorithm to construct extraction models for different rice growth parameters.
[0044] The present invention adopts the coefficient of determination (Coefficient of determination, R 2 ), root mean square error (RMSE) and relative root mean square error (RRMSE) are used to calculate the model accuracy. The calculation formula is as follows:
[0045]
[0046]
[0047] in y , and represent the measured values, estimated values and average values of different physicochemical parameters of rice, respectively, and n is the sample size.
[0048] VI. Stability Verification A variety of combined scenarios of UAV flight altitudes and times (including combinations of high altitude, low altitude, morning and afternoon in different periods) were randomly selected to evaluate the stability of FE-RFR algorithm in extracting rice growth parameters in multiple periods, and the accuracy was compared with spectral index, RFR, GBDT and XGBoost models to evaluate the spatiotemporal adaptability of the method proposed in the invention in extracting rice physical and chemical parameters.
[0049] Based on the analysis of the spectral variation law of rice canopy when simulating different leaf area indexes and chlorophyll contents based on the PROSAIL model, the present invention proposes and constructs SCA. By improving the machine learning algorithm, the spectral index and SCA are fully integrated to achieve feature enhancement and variable screening, thereby improving the accuracy and stability of rice physical and chemical parameter extraction.
[0050] The rice physical and chemical parameter extraction system for fusing spectral curve angles and spectral indices provided by the present invention is described below. The rice physical and chemical parameter extraction system for fusing spectral curve angles and spectral indices described below and the rice physical and chemical parameter extraction method for fusing spectral curve angles and spectral indices described above can be referenced to each other.
[0051] Figure 2 is a schematic diagram of the structure of a rice physical and chemical parameter extraction system integrating spectral curve angle and spectral index provided by an embodiment of the present invention. Figure 2As shown, it includes: an acquisition preprocessing module 21, a calculation module 22, a simulation module 23, a construction module 24, a training module 25 and an extraction module 26, wherein: The acquisition and preprocessing module 21 is used to acquire UAV multispectral images of the study area, preprocess the UAV multispectral images, and obtain the UAV multispectral preprocessed images; the calculation module 22 is used to calculate the spectral index in the UAV multispectral preprocessed images based on the multi-band spectral reflectance; the simulation module 23 is used to simulate the rice canopy spectrum using the PROSAIL model according to the rice leaves, soil and observation geometry input parameters, and obtain the variation law of the rice canopy spectral reflectance with the leaf area index and the chlorophyll content; the construction module 24 is used to construct the spectral curve angle according to the variation law of the rice canopy spectral reflectance with the leaf area index and the chlorophyll content; the training module 25 is used to use RReliefF to construct and train the improved random forest regression model using the spectral index and the spectral curve angle to obtain a multi-period rice physical and chemical parameter extraction model; the extraction module 26 is used to input the multispectral image features extracted from the target area into the multi-period rice physical and chemical parameter extraction model, the multispectral image features include spectral index and spectral curve angle, and obtain the multi-period rice physical and chemical parameter extraction results.
[0052] Figure 3 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 3 As shown, the electronic device may include: a processor (processor) 310, a communication interface (Communications Interface) 320, a memory (memory) 330 and a communication bus 340, wherein the processor 310, the communication interface 320, and the memory 330 communicate with each other through the communication bus 340. The processor 310 can call the logic instructions in the memory 330 to execute the rice physical and chemical parameter extraction method by fusing the spectral curve angle and the spectral index. The method includes: collecting unmanned aerial vehicle multispectral images of the study area, preprocessing the unmanned aerial vehicle multispectral images, and obtaining unmanned aerial vehicle multispectral preprocessed images; calculating the spectral index in the unmanned aerial vehicle multispectral preprocessed images based on multi-band spectral reflectance; simulating the rice canopy spectrum using the PROSAIL model according to the rice leaves, soil and observation geometry input parameters, and obtaining the variation law of the rice canopy spectral reflectance with the leaf area index and the chlorophyll content; constructing the spectral curve angle according to the variation law of the rice canopy spectral reflectance with the leaf area index and the chlorophyll content; using RReliefF, using the spectral index and the spectral curve angle to construct and train an improved random forest regression model to obtain a multi-period rice physical and chemical parameter extraction model; inputting the multi-spectral image features extracted from the target area into the multi-period rice physical and chemical parameter extraction model to obtain the multi-period rice physical and chemical parameter extraction results.
[0053] In addition, the logic instructions in the above-mentioned memory 330 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.
[0054] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the rice physical and chemical parameter extraction method provided by the above-mentioned methods by fusing spectral curve angles and spectral indices, the method comprising: collecting unmanned aerial vehicle multispectral images of the study area, preprocessing the unmanned aerial vehicle multispectral images, and obtaining unmanned aerial vehicle multispectral preprocessed images; calculating the spectral index in the unmanned aerial vehicle multispectral preprocessed images based on multi-band spectral reflectance; simulating the rice canopy spectrum using the PROSAIL model according to rice leaves, soil and observation geometry input parameters, and obtaining the variation law of the rice canopy spectral reflectance with leaf area index and chlorophyll content; constructing the spectral curve angle according to the variation law of the rice canopy spectral reflectance with leaf area index and chlorophyll content; using RReliefF, using the spectral index and the spectral curve angle to construct and train an improved random forest regression model to obtain a multi-period rice physical and chemical parameter extraction model; inputting the multispectral image features extracted from the target area into the multi-period rice physical and chemical parameter extraction model, the multispectral image features including the spectral index and the spectral curve angle, and obtaining the multi-period rice physical and chemical parameter extraction results.
[0055] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.
[0056] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.
[0057] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for extracting rice physical and chemical parameters by integrating spectral curve angle and spectral index, characterized in that: include: Collecting UAV multispectral images of the study area, and preprocessing the UAV multispectral images to obtain UAV multispectral preprocessed images; Calculating the spectral index in the multi-spectral pre-processed image of the UAV based on the multi-band spectral reflectance; The PROSAIL model was used to simulate the spectrum of rice canopy based on the input parameters of rice leaves, soil and observation geometry, and the variation pattern of rice canopy spectral reflectance with leaf area index and chlorophyll content was obtained. Constructing a spectral curve angle according to the variation law of the spectral reflectance of the rice canopy with the leaf area index and the chlorophyll content; The feature selection algorithm RReliefF is adopted to construct and train an improved random forest regression model using the spectral index and the spectral curve angle, and a multi-period rice physical and chemical parameter extraction model is obtained; The multi-spectral image features extracted from the target area are input into the multi-stage rice physical and chemical parameter extraction model, wherein the multi-spectral image features include spectral index and spectral curve angle, and the multi-stage rice physical and chemical parameter extraction results are obtained.
2. The method for extracting rice physical and chemical parameters by integrating spectral curve angle and spectral index according to claim 1, characterized in that: Collecting UAV multispectral images of the study area, preprocessing the UAV multispectral images to obtain UAV multispectral preprocessed images, including: According to the preset weather conditions of the study area, the drone's aerial photography method and aerial photography route are determined, and the multispectral image is acquired according to the aerial photography method and the aerial photography route; the multispectral image is subjected to geometric processing, radiation processing, and spectral information extraction in sequence to obtain the drone's multispectral preprocessed image.
3. The method for extracting rice physical and chemical parameters by fusing spectral curve angle and spectral index according to claim 2, characterized in that: The multispectral image is subjected to geometric processing, radiation processing and spectral information extraction in sequence to obtain the multispectral preprocessed image of the UAV, including: The geometric processing includes vignetting correction, band registration and distortion correction; The radiation processing includes radiation calibration, which is achieved by using a plurality of calibration blankets with different reflectivities and a piecewise linear calibration method; The spectral information extraction includes the delineation of regions of interest, the extraction of reflectance and the operation of bands.
4. The method for extracting rice physical and chemical parameters by fusing spectral curve angle and spectral index according to claim 1, characterized in that: Based on the multi-band spectral reflectance, the spectral index in the multi-spectral pre-processed image of the UAV is calculated, including: Acquire multiple reflectances of the multispectral preprocessed image in different bands; According to the multiple reflectivities, a difference spectral index, a ratio spectral index, a normalized spectral index and an improved spectral index are calculated; The difference spectral index includes the difference vegetation index, the ratio spectral index includes the ratio vegetation index, the normalized spectral index includes the normalized difference vegetation index and the normalized red edge index, and the improved spectral index includes the visible light atmospheric impedance index, the two-band enhanced vegetation index, the red chlorophyll index, the green chlorophyll index, the modified chlorophyll absorption reflectance index and the wide dynamic range vegetation index.
5. The method for extracting rice physical and chemical parameters by integrating spectral curve angle and spectral index according to claim 1, characterized in that: According to the rice leaf, soil and observation geometry input parameters, the PROSAIL model is used to simulate the rice canopy spectrum, and the variation law of rice canopy spectral reflectance with leaf area index and chlorophyll content is obtained, including: The PROSAIL model is obtained by coupling the leaf reflectance model PROSPECT-5B and the canopy reflectance model SAIL; The input parameters of the PROSPECT-5B model include leaf structural parameters, leaf chlorophyll content, carotenoid content, brown pigment content, leaf water content, and leaf dry matter content; The SAIL model input parameters include leaf area index, mean leaf inclination, hot spot parameters, solar zenith angle, observation zenith angle, soil brightness parameters, and observation relative azimuth; Obtaining the reflectivity of the rice leaf according to the leaf structure parameters, the leaf chlorophyll content, the carotenoid content, the brown pigment content, the leaf water content and the leaf dry matter content; The canopy spectral reflectance is obtained according to the rice leaf reflectance, the leaf area index, the average leaf inclination angle, the hot spot parameter, the solar zenith angle, the observation zenith angle, the soil brightness parameter and the observation relative azimuth.
6. The method for extracting rice physical and chemical parameters by fusing spectral curve angle and spectral index according to claim 1, characterized in that: The spectral curve angle is constructed according to the variation law of the spectral reflectance of the rice canopy with the leaf area index and the chlorophyll content, including: Based on the spectral reflectance of the rice canopy, obtaining a first reflectance, a second reflectance, a first wavelength and a second wavelength of any two bands; A first variable is obtained from the first reflectivity and the second reflectivity, and a second variable is obtained from the first wavelength and the second wavelength; The first variable and the second variable are calculated using an inverse tangent function to obtain the spectral curve angle.
7. The method for extracting rice physical and chemical parameters by fusing spectral curve angle and spectral index according to claim 1, characterized in that: By using RReliefF, the spectral index and the spectral curve angle are used to construct and train an improved random forest regression model, and a multi-period rice physical and chemical parameter extraction model is obtained, including: Inputting the spectral index and the random linear and nonlinear combination of the spectral curve angle into the random forest model to obtain the improved random forest regression model; A training set is formed by the spectral index and the spectral curve angle, and RReliefF is used to randomly select any sample in the training set, and search and determine adjacent samples of the same category and adjacent samples of different categories of the any sample; Determine any two samples of the continuous eigenvalue, and obtain the difference between the two samples on the continuous eigenvalue by the ratio of the absolute value of the difference between the two samples and the difference between the maximum value and the minimum value of the continuous eigenvalue; Calculate the difference between the two samples on the continuous feature value, the approximate probability distribution of the adjacent samples of the same category and the adjacent samples of different categories to obtain the continuous feature weight; Convert the difference between the two arbitrary samples on the continuous feature value into a resampling probability of two similar samples, and use the resampling probability to obtain an updated continuous feature weight; The training set is screened based on the updated continuous feature weights, and the improved random forest regression model is trained using the screened feature variables to obtain an initial multi-period rice physicochemical parameter extraction model; The determination coefficient, root mean square error and relative root mean square error were used to estimate the model accuracy of the multi-period rice physicochemical parameter extraction model, and an optimized multi-period rice physicochemical parameter extraction model was obtained.
8. A rice physical and chemical parameter extraction system integrating spectral curve angle and spectral index, characterized in that: include: An acquisition preprocessing module is used to acquire UAV multispectral images of a research area, and preprocess the UAV multispectral images to obtain UAV multispectral preprocessed images; A calculation module, used for calculating the spectral index in the multi-spectral pre-processed image of the UAV based on the multi-band spectral reflectance; The simulation module uses the PROSAIL model to simulate the spectrum of the rice canopy based on the rice leaf, soil and observation geometry input parameters, and obtains the variation pattern of the spectral reflectance of the rice canopy with the leaf area index and chlorophyll content; A construction module, used to construct a spectral curve angle according to the variation law of the spectral reflectance of the rice canopy; A training module is used to construct and train an improved random forest regression model using the spectral index and the spectral curve angle by using RReliefF to obtain a multi-period rice physical and chemical parameter extraction model; The extraction module is used to input the multi-spectral image features extracted from the target area into the multi-stage rice physical and chemical parameter extraction model, wherein the multi-spectral image features include spectral index and spectral curve angle, to obtain the multi-stage rice physical and chemical parameter extraction results.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method for extracting rice physical and chemical parameters by fusing spectral curve angles and spectral indices as described in any one of claims 1 to 7 is implemented.
10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for extracting rice physical and chemical parameters by fusing spectral curve angles and spectral indices as claimed in any one of claims 1 to 7 is implemented.
Citation Information
Cited By
Saline-alkali soil corn growth vigor and salt stress monitoring method based on unmanned aerial vehicle multispectrum
CN121962983A
Estimation method for crop yield
TWI909968B
