Cotton yield cross-scale estimation method and system based on multi-modal remote sensing
By combining multimodal remote sensing technology with optical imagery and lidar data, precise cross-scale analysis of cotton yield was achieved, solving the problems of insufficient accuracy and efficiency in existing cotton yield prediction methods and realizing high-precision and stable yield estimation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-29
- Publication Date
- 2026-03-27
AI Technical Summary
Existing methods for predicting cotton yield are insufficient in terms of accuracy and efficiency. Traditional methods rely on indirect data or manual surveys, resulting in large errors, high costs, low efficiency, and difficulty in achieving refined management.
By employing multimodal remote sensing technology, combined with UAV optical imagery and lidar point cloud data, and through cotton plant identification models, plant height information extraction, and plant height-phenological period-cotton boll number relationship models, we can achieve precise cross-scale analysis from individual plants to field plots, and integrate crop growth physiological mechanisms.
It improves the accuracy and stability of cotton yield estimation, achieves precise analysis from individual plants to field plots, breaks through the population homogenization assumption of traditional remote sensing yield estimation, and enhances the accuracy and mechanism of yield estimation.
Smart Images
Figure CN121746931A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of agricultural information technology and relates to a method and system for cross-scale estimation of cotton yield based on multimodal remote sensing. Background Technology
[0002] Cotton is an important economic crop in my country, occupying a key position in the oilseed and fiber raw material sectors, with its consumption and output ranking among the world's top. Timely and accurate understanding of cotton growth and yield forecasting are of significant practical importance for strengthening cotton production management, formulating scientific production plans, optimizing foreign trade strategies, and promoting high-quality regional economic development.
[0003] However, current cotton yield forecasting methods vary widely, exhibiting significant shortcomings in both accuracy and efficiency. Some methods rely on indirect data such as cumulative shipments for rough estimations, resulting in low accuracy. Other methods employ field surveys, estimating planting area before deducing yield, leading to substantial errors. Still others utilize a bottom-up statistical model, relying on manual field records, hierarchical reporting of paper data, and manual data aggregation and analysis. This results in heavy workloads, lengthy processes, and high costs, and the acquired data is difficult to process and effectively mine. Due to inconsistent survey methods and varying levels of personnel expertise, existing methods have significant limitations in the accuracy and applicability of information collection, severely hindering the refined management of my country's cotton industry and the sustainable development of its agricultural economy. Summary of the Invention
[0004] To address the problems in existing technologies, this invention provides a method and system for cross-scale estimation of cotton yield based on multimodal remote sensing. It achieves accurate cross-scale analysis from individual plants to field plots and incorporates crop growth physiological mechanisms, effectively improving the accuracy and stability of yield estimation.
[0005] To achieve the above objectives, the present invention employs the following technical solution: In a first aspect, the present invention provides a method for estimating cotton yield across scales based on multimodal remote sensing, comprising the following steps: S1. Acquire multi-temporal UAV multimodal remote sensing data of the target cotton field during the cotton growth cycle. The multimodal remote sensing data includes synchronously acquired optical images and lidar point cloud data. S2. Based on the optical image, target detection and recognition are performed using the trained cotton plant identification model to obtain the location information of each cotton plant in the target cotton field and to count the total number of cotton plants. S3. Based on the lidar point cloud data, extract the plant height information of each cotton plant in the target cotton field; S4. For the target cotton field, combining the phenological information of the current time phase and the plant height information of each cotton plant, the number of cotton bolls per cotton plant is estimated using a pre-constructed plant height-phenological period-number of cotton bolls relationship model. S5. Calculate the total cotton yield of the target cotton field based on the number of bolls per cotton plant, the preset average weight of a single cotton boll, and the total number of cotton plants.
[0006] Preferably, the method for constructing the trained cotton plant identification model includes: Obtain training and validation sets of UAV optical images containing cotton plant labeling information; An improved YOLO recognition model is constructed. The improved YOLO recognition model uses the YOLOv8 network as the baseline framework, replaces its backbone feature extraction network with the MobileNetv3 network, and embeds the ECANet attention mechanism in the channel attention module of its neck feature fusion network. The improved YOLO recognition model was trained and its parameters optimized using the aforementioned UAV optical image training set. The trained model is evaluated using a validation set, and the model parameters are adjusted based on the evaluation results until the preset accuracy requirements are met, thus obtaining the trained cotton plant identification model.
[0007] Preferably, the step of extracting the plant height information of each cotton plant in the target cotton field based on the lidar point cloud data includes: The lidar point cloud data is preprocessed, including noise reduction and classification, to separate the cotton canopy point cloud; Based on the location information of the individual cotton plant, individual point cloud clusters corresponding to each cotton plant are segmented from the cotton canopy point cloud; For each individual cotton plant cluster, the plant height is calculated based on the statistical difference between the highest and lowest points.
[0008] Preferably, the step of segmenting the single-plant point cloud clusters corresponding to each cotton plant from the cotton canopy point cloud includes: The identified locations of individual cotton plants are converted into vector surface data; the vector surface data of individual cotton plants are spatially superimposed and analyzed with the preprocessed lidar point cloud data to extract the point cloud falling within the range of each vector surface, forming the individual cotton plant point cloud cluster.
[0009] Preferably, the method for constructing the plant height-phenological stage-cotton boll number relationship model includes: Within the target cotton field, multi-temporal UAV lidar point cloud data, corresponding phenological information, and on-site measurements of the actual plant height and actual number of cotton bolls were collected simultaneously. The remote sensing inversion plant height of each cotton plant was obtained based on the lidar point cloud data. Using the phenological period information, the remotely sensed plant height or the actual plant height as input features, and the actual number of cotton bolls as the target variable, a machine learning regression algorithm is used for training to obtain the plant height-phenological period-number of cotton bolls relationship model.
[0010] Preferably, the formula for calculating the total cotton yield of the target cotton field is:
[0011] In the formula, Total cotton production; For the first The number of bolls per cotton plant; The preset average weight of a single cotton boll; This represents the total number of cotton plants.
[0012] Preferably, after calculating the total cotton yield of the target cotton field, the method further includes: During the entire growth cycle of cotton, S1~S5 are executed at preset time intervals to obtain the estimated total cotton yield at different time phases. The estimated yield at each time phase is compared and analyzed with the final measured yield. The optimal observation time window is determined based on the earliest time phase with the highest estimation accuracy or the earliest time phase that reaches stable estimation accuracy.
[0013] Secondly, the present invention provides a multi-scale cotton yield estimation system based on multimodal remote sensing, comprising: Data acquisition module: used to acquire multi-temporal UAV multimodal remote sensing data of the target cotton field during the cotton growth cycle. The multimodal remote sensing data includes synchronously acquired optical images and lidar point cloud data. Plant identification module: Based on the optical image, it uses a trained cotton plant identification model to perform target detection and identification, obtain the location information of each cotton plant in the target cotton field, and count the total number of cotton plants; Plant height extraction module: used to extract the plant height information of each cotton plant in the target cotton field based on the lidar point cloud data; Cotton Boll Estimation Module: This module is used to estimate the number of cotton bolls per cotton plant for the target cotton field by combining the phenological information of the current time phase and the plant height information of each cotton plant, and by using a pre-constructed plant height-phenological period-cotton boll number relationship model. Yield calculation module: used to calculate the total cotton yield of the target cotton field based on the number of cotton bolls per cotton plant, the preset average weight of a single cotton boll, and the total number of cotton plants.
[0014] Thirdly, the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of a multi-scale estimation method for cotton yield based on multimodal remote sensing.
[0015] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of a multi-scale estimation method for cotton yield based on multimodal remote sensing.
[0016] Compared with the prior art, the present invention has the following beneficial effects: This invention utilizes optical imagery for precise identification and positioning of individual cotton plants, reliably obtaining the number of plants per unit area. It extracts precise three-dimensional plant height information for each cotton plant from lidar point clouds, overcoming the limitations of traditional remote sensing in characterizing crop structural parameters. By combining key phenological information, plant height, a growth indicator, is transformed into the number of bolls per plant through a mechanistic model, establishing a crucial bridge connecting remote sensing observation with the physiological processes of yield formation. Based on the number of bolls per plant, preset boll weight, and total number of plants, it achieves a plant-by-plant cumulative calculation of the total yield of the field. This method breaks the coarse assumption of population homogeneity in traditional remote sensing yield estimation, achieving precise cross-scale analysis from individual plants to the entire field, and integrating crop growth physiological mechanisms, effectively improving the accuracy and stability of yield estimation. Attached Figure Description
[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0018] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation
[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.
[0020] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.
[0021] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.
[0022] In the description of the embodiments of the present invention, it should be noted that if terms such as "upper," "lower," "horizontal," or "inner" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of the invention is in use, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, terms such as "first" and "second" are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.
[0023] Furthermore, the use of the term "horizontal" does not imply that the component must be absolutely horizontal, but rather that it can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted.
[0024] In the description of the embodiments of the present invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in the present invention according to the specific circumstances.
[0025] The present invention will now be described in further detail with reference to the accompanying drawings: The first objective of this invention is to provide a method for cross-scale estimation of cotton yield based on multimodal remote sensing, such as... Figure 1 As shown, it includes the following steps: S1. Acquire multi-temporal UAV multimodal remote sensing data of the target cotton field during the cotton growth cycle. The multimodal remote sensing data includes synchronously acquired optical images and lidar point cloud data. S2. Based on the optical image, target detection and recognition are performed using the trained cotton plant identification model to obtain the location information of each cotton plant in the target cotton field and to count the total number of cotton plants. S3. Based on the lidar point cloud data, extract the plant height information of each cotton plant in the target cotton field; S4. For the target cotton field, combining the phenological information of the current time phase and the plant height information of each cotton plant, the number of cotton bolls per cotton plant is estimated using a pre-constructed plant height-phenological period-number of cotton bolls relationship model. S5. Calculate the total cotton yield of the target cotton field based on the number of bolls per cotton plant, the preset average weight of a single cotton boll, and the total number of cotton plants.
[0026] The formula for calculating the total cotton yield of the target cotton field is as follows:
[0027] In the formula, Total cotton production; For the first The number of bolls per cotton plant; The preset average weight of a single cotton boll; This represents the total number of cotton plants.
[0028] This invention presents a multi-modal remote sensing-based method for estimating cotton yield across scales. By deeply fusing UAV optical imagery and lidar point cloud data, it achieves accurate yield estimation from the individual plant to the field scale. Specifically: First, the optical imagery is processed to accurately identify the location of each cotton plant and count the total number of plants, providing a reliable spatial distribution benchmark for yield estimation. Second, the plant height information of each cotton plant is accurately extracted from lidar point cloud data, obtaining key phenotypic parameters that are difficult to obtain on a large scale using traditional methods. Next, combined with current phenological stage information, a pre-constructed plant height-phenological stage-boll number relationship model is used to convert the remotely sensed plant height into the number of bolls per plant, establishing a bridge connecting remote sensing observation and the physiological process of yield formation. Finally, based on the number of bolls per plant, the preset weight of a single boll, and the total number of plants, the total yield is calculated by summing. This method breaks through the coarse assumption of population homogeneity in traditional remote sensing yield estimation. Through multimodal data collaboration, deep learning and crop physiological models are deeply integrated, forming a new cross-scale analysis paradigm from pixel to plant to field. This not only significantly improves the accuracy and mechanism of yield estimation, but also provides scalable technical support for smart agricultural management.
[0029] For example, the method for constructing the trained cotton plant identification model includes: Obtain training and validation sets of UAV optical images containing cotton plant labeling information; An improved YOLO recognition model is constructed. The improved YOLO recognition model uses the YOLOv8 network as the baseline framework, replaces its backbone feature extraction network with the MobileNetv3 network, and embeds the ECANet attention mechanism in the channel attention module of its neck feature fusion network. The improved YOLO recognition model was trained and its parameters optimized using the aforementioned UAV optical image training set. The trained model is evaluated using a validation set, and the model parameters are adjusted based on the evaluation results until the preset accuracy requirements are met, thus obtaining the trained cotton plant identification model.
[0030] This invention significantly improves the accuracy and efficiency of cotton plant identification in complex field scenarios by specifically optimizing the deep learning model architecture. Specifically, the method first obtains a professionally labeled UAV optical image dataset as the training foundation. Then, two key improvements are made to the baseline YOLOv8 network: the backbone network is replaced with the lightweight MobileNetv3, which greatly reduces the number of model parameters and computational complexity, making the model more suitable for deployment on edge computing devices or mobile devices, thus improving its practicality; simultaneously, an ECANet attention mechanism module is embedded in the neck feature fusion network. This module learns the importance between channels through efficient one-dimensional convolution, enabling the model to focus more on the key features of the cotton plant and effectively suppress interference from background weeds, soil, and other information, thereby improving the robustness of identification in complex situations such as adhesion and occlusion. Then, the improved model is fully trained and its parameters optimized using the training set; finally, accuracy is evaluated and iteratively tuned using an independent validation set to ensure the model achieves the preset performance indicators such as accuracy and recall.
[0031] For example, extracting the plant height information of each cotton plant in the target cotton field based on the lidar point cloud data includes: The lidar point cloud data is preprocessed, including noise reduction and classification, to separate the cotton canopy point cloud; Based on the location information of the individual cotton plant, individual point cloud clusters corresponding to each cotton plant are segmented from the cotton canopy point cloud; For each individual cotton plant cluster, the plant height is calculated based on the statistical difference between the highest and lowest points.
[0032] This invention effectively filters out noise interference from ground points, weeds, and other sources by denoising and classifying the original point cloud, accurately separating the effective point cloud representing the cotton canopy, providing a high-quality three-dimensional data foundation for subsequent single-plant-scale analysis. Secondly, based on the single-plant location information identified in optical images, the invention precisely segments the single-plant point cloud clusters corresponding to each cotton plant from the canopy point cloud, achieving precise spatial matching and fusion of multimodal data at the plant level. Finally, the invention uses the difference between the highest and lowest points to calculate plant height. This method is intuitive, computationally stable, and can efficiently and non-destructively obtain precise single-plant height parameters that are difficult to achieve on a large scale using traditional manual measurements.
[0033] The step of segmenting individual cotton plant point cloud clusters from the cotton canopy point cloud includes: The identified locations of individual cotton plants are converted into vector surface data; the vector surface data of individual cotton plants are spatially superimposed and analyzed with the preprocessed lidar point cloud data to extract the point cloud falling within the range of each vector surface, forming the individual cotton plant point cloud cluster.
[0034] By converting the location of individual cotton plants identified through optical imagery (typically the center point of the detection frame or the smallest bounding polygon) into vector surface data with clearly defined geographic coordinates, and then accurately overlaying it with high-precision lidar point clouds on a GIS spatial analysis platform, this method can automatically extract 3D point cloud clusters falling within the vector surface range of each individual plant based on accurate georegistration. This enables precise correlation between 2D planar positioning and 3D structural information at the individual plant scale.
[0035] For example, the method for constructing the plant height-phenological stage-cotton boll number relationship model includes: Within the target cotton field, multi-temporal UAV lidar point cloud data, corresponding phenological information, and on-site measurements of the actual plant height and actual number of cotton bolls were collected simultaneously. The remote sensing inversion plant height of each cotton plant was obtained based on the lidar point cloud data. Using the phenological period information, the remotely sensed plant height or the actual plant height as input features, and the actual number of cotton bolls as the target variable, a machine learning regression algorithm is used for training to obtain the plant height-phenological period-number of cotton bolls relationship model.
[0036] This method effectively solves the challenge of accurately estimating the key yield component of cotton (number of bolls). Specifically, the simultaneous acquisition of lidar point clouds, field measurement data, and phenological information ensures strict temporal and spatial consistency between input and target variables, providing a high-quality, highly relevant dataset for modeling. The method also uses remotely sensed plant height or field-measured plant height along with phenological information as input features. This not only fully utilizes the rapid and non-destructive monitoring advantages of UAV remote sensing but also captures the dynamic changes in the relationship between plant height and boll formation during cotton growth by introducing the key temporal variable of phenological period, giving the model a solid foundation in crop physiology. Finally, leveraging the powerful nonlinear fitting capabilities of machine learning regression algorithms, a quantitative model capable of characterizing the complex relationship between plant height, phenological period, and number of bolls was successfully constructed. This deep integration of macroscopic physical signals from remote sensing observations with the microscopic physiological processes of crop growth enables high-precision and high-efficiency inversion of the number of bolls per plant using only UAV remote sensing data and phenological information, without the need for large-scale manual surveys. This provides core algorithmic support for accurate large-area yield prediction from a mechanistic perspective.
[0037] For example, after calculating the total cotton yield of the target cotton field, the method further includes: During the entire growth cycle of cotton, S1~S5 are executed at preset time intervals to obtain the estimated total cotton yield at different time phases. The estimated yield at each time phase is compared and analyzed with the final measured yield. The optimal observation time window is determined based on the earliest time phase with the highest estimation accuracy or the earliest time phase that reaches stable estimation accuracy.
[0038] This subsequent step, by tracking the time-series evolution of yield estimation accuracy at different growth stages (such as seedling stage, bud stage, flowering and boll-forming stage, and boll-opening stage), can not only assess the stability of the model's performance throughout the entire growth cycle, but more importantly, can scientifically identify the critical time point when stable estimation accuracy or the peak accuracy is achieved.
[0039] On the one hand, this method can clearly indicate the most critical and efficient time nodes for UAV remote sensing monitoring in production decision-making. While ensuring prediction accuracy, it significantly reduces unnecessary repeated flights and data collection, greatly improving the economy and feasibility of the technology. On the other hand, through time-series analysis of the correlation between the estimation accuracy of different growth stages and the final yield, it scientifically reveals the key stages of remote sensing response in the yield formation process, deepens the understanding of crop physiological mechanisms, and provides an important basis for model feature optimization and mechanism explanation, promoting the development of technology towards mechanism-driven intelligentization.
[0040] A second objective of this invention is to provide a multi-scale cotton yield estimation system based on multimodal remote sensing, comprising: Data acquisition module: used to acquire multi-temporal UAV multimodal remote sensing data of the target cotton field during the cotton growth cycle. The multimodal remote sensing data includes synchronously acquired optical images and lidar point cloud data. Plant identification module: Based on the optical image, it uses a trained cotton plant identification model to perform target detection and identification, obtain the location information of each cotton plant in the target cotton field, and count the total number of cotton plants; Plant height extraction module: used to extract the plant height information of each cotton plant in the target cotton field based on the lidar point cloud data; Cotton Boll Estimation Module: This module is used to estimate the number of cotton bolls per cotton plant for the target cotton field by combining the phenological information of the current time phase and the plant height information of each cotton plant, and by using a pre-constructed plant height-phenological period-cotton boll number relationship model. Yield calculation module: used to calculate the total cotton yield of the target cotton field based on the number of cotton bolls per cotton plant, the preset average weight of a single cotton boll, and the total number of cotton plants.
[0041] In this system, the data acquisition module ensures the synchronous acquisition and spatiotemporal consistency of multimodal data; the plant identification module enables precise positioning of individual plants in complex field scenarios; the plant height extraction module efficiently acquires key three-dimensional phenotypic parameters; the cotton boll estimation module completes the accurate inversion of yield components through an intelligent model that integrates physiological mechanisms; and the yield calculation module finally completes the cross-scale yield aggregation from individual plants to the entire field.
[0042] In one embodiment of the present invention, a computer device is provided, comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions from the computer storage medium to achieve a corresponding method flow or function. The processor described in this embodiment can be used for the operation of a multi-scale cotton yield estimation method based on multimodal remote sensing.
[0043] This invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the cotton yield cross-scale estimation method based on multimodal remote sensing in the above embodiments.
[0044] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0045] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0046] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0047] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0048] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.
Claims
1. A method for estimating cotton yield across scales based on multimodal remote sensing, characterized in that, Includes the following steps: S1. Acquire multi-temporal UAV multimodal remote sensing data of the target cotton field during the cotton growth cycle. The multimodal remote sensing data includes synchronously acquired optical images and lidar point cloud data. S2. Based on the optical image, target detection and recognition are performed using the trained cotton plant identification model to obtain the location information of each cotton plant in the target cotton field and to count the total number of cotton plants. S3. Based on the lidar point cloud data, extract the plant height information of each cotton plant in the target cotton field; S4. For the target cotton field, combining the phenological information of the current time phase and the plant height information of each cotton plant, the number of cotton bolls per cotton plant is estimated using a pre-constructed plant height-phenological period-number of cotton bolls relationship model. S5. Calculate the total cotton yield of the target cotton field based on the number of bolls per cotton plant, the preset average weight of a single cotton boll, and the total number of cotton plants.
2. The method for cross-scale estimation of cotton yield based on multimodal remote sensing according to claim 1, characterized in that, The method for constructing the trained cotton plant identification model includes: Obtain training and validation sets of UAV optical images containing cotton plant labeling information; An improved YOLO recognition model is constructed. The improved YOLO recognition model uses the YOLOv8 network as the baseline framework, replaces its backbone feature extraction network with the MobileNetv3 network, and embeds the ECANet attention mechanism in the channel attention module of its neck feature fusion network. The improved YOLO recognition model was trained and its parameters optimized using the aforementioned UAV optical image training set. The trained model is evaluated using a validation set, and the model parameters are adjusted based on the evaluation results until the preset accuracy requirements are met, thus obtaining the trained cotton plant identification model.
3. The method for cross-scale estimation of cotton yield based on multimodal remote sensing according to claim 1, characterized in that, The step of extracting the plant height information of each cotton plant in the target cotton field based on the lidar point cloud data includes: The lidar point cloud data is preprocessed, including noise reduction and classification, to separate the cotton canopy point cloud; Based on the location information of the individual cotton plant, individual point cloud clusters corresponding to each cotton plant are segmented from the cotton canopy point cloud; For each individual cotton plant cluster, the plant height is calculated based on the statistical difference between the highest and lowest points.
4. The method for cross-scale estimation of cotton yield based on multimodal remote sensing according to claim 3, characterized in that, The step of segmenting individual cotton plant point cloud clusters from the cotton canopy point cloud includes: The identified locations of individual cotton plants are converted into vector surface data; the vector surface data of individual cotton plants are spatially superimposed and analyzed with the preprocessed lidar point cloud data to extract the point cloud falling within the range of each vector surface, forming the individual cotton plant point cloud cluster.
5. The method for cross-scale estimation of cotton yield based on multimodal remote sensing according to claim 1, characterized in that, The method for constructing the plant height-phenological period-cotton boll number relationship model includes: Within the target cotton field, multi-temporal UAV lidar point cloud data, corresponding phenological information, and on-site measurements of the actual plant height and actual number of cotton bolls were collected simultaneously. The remote sensing inversion plant height of each cotton plant was obtained based on the lidar point cloud data. Using the phenological period information, the remotely sensed plant height or the actual plant height as input features, and the actual number of cotton bolls as the target variable, a machine learning regression algorithm is used for training to obtain the plant height-phenological period-number of cotton bolls relationship model.
6. The method for cross-scale estimation of cotton yield based on multimodal remote sensing according to claim 1, characterized in that, The formula for calculating the total cotton yield of the target cotton field is as follows: In the formula, Total cotton production; For the first The number of bolls per cotton plant; The preset average weight of a single cotton boll; This represents the total number of cotton plants.
7. The method for cross-scale estimation of cotton yield based on multimodal remote sensing according to claim 1, characterized in that, After calculating the total cotton yield of the target cotton field, the method further includes: During the entire growth cycle of cotton, S1~S5 are executed at preset time intervals to obtain the estimated total cotton yield at different time phases. The estimated yield at each time phase is compared and analyzed with the final measured yield. The optimal observation time window is determined based on the earliest time phase with the highest estimation accuracy or the earliest time phase that reaches stable estimation accuracy.
8. A multi-scale cotton yield estimation system based on multimodal remote sensing, characterized in that, include: Data acquisition module: used to acquire multi-temporal UAV multimodal remote sensing data of the target cotton field during the cotton growth cycle. The multimodal remote sensing data includes synchronously acquired optical images and lidar point cloud data. Plant identification module: Based on the optical image, it uses a trained cotton plant identification model to perform target detection and identification, obtain the location information of each cotton plant in the target cotton field, and count the total number of cotton plants; Plant height extraction module: used to extract the plant height information of each cotton plant in the target cotton field based on the lidar point cloud data; Cotton Boll Estimation Module: This module is used to estimate the number of cotton bolls per cotton plant for the target cotton field by combining the phenological information of the current time phase and the plant height information of each cotton plant, and by using a pre-constructed plant height-phenological period-cotton boll number relationship model. Yield calculation module: used to calculate the total cotton yield of the target cotton field based on the number of cotton bolls per cotton plant, the preset average weight of a single cotton boll, and the total number of cotton plants.
9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1-7.
10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-7.