Comprehensive evaluation system and evaluation method for pepper harvesting operation quality

Abstract

The invention discloses a pepper harvesting operation quality comprehensive evaluation system and evaluation method, and belongs to the technical field of intelligent agricultural machinery, the evaluation system is composed of an unmanned aerial vehicle airborne detection system and a harvester airborne detection system, and the unmanned aerial vehicle airborne detection system is composed of an image acquisition module and a first satellite RTK-GPS positioning module; a harvester onboard detection system is composed of a vehicle-mounted sensor unit, a harvester positioning unit and a data transmission and processing unit which are connected in parallel. Real-time collection, accurate analysis and comprehensive evaluation of multi-dimensional data in the pepper harvesting process become a critical path for improving the pepper harvesting operation quality and agricultural production benefits, and through the technical means of high-precision positioning, multi-dimensional data collection, real-time data processing, intelligent evaluation and the like, the agricultural production efficiency is improved. According to the method, the quality evaluation efficiency and accuracy of pepper harvesting operation are remarkably improved, powerful technical support is provided for farmers and operators, crop loss can be reduced, the operation efficiency is improved, and resource allocation is optimized.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent agricultural machinery, and relates to a fruit and vegetable harvesting operation quality evaluation system and evaluation method based on airborne sensing monitoring integration, specifically to a damaged and impurity-containing operation quality evaluation system and evaluation method during the pepper harvesting process. Background Art

[0002] In the field of agricultural production, the quality assessment of pepper harvesting operations is of great importance. The traditional quality assessment of pepper harvesting operations mainly relies on manual sampling inspection, which has many drawbacks. On the one hand, the efficiency of manual inspection is extremely low. In large-scale planting areas, it consumes a large amount of manpower and time, and it is difficult to quickly obtain comprehensive operation quality information. For example, in a large-scale pepper planting farm, manually checking the pepper breakage rate, impurity rate, and dropping situation may require many workers to conduct one-by-one screening in the fields for a long time, and as the harvesting process progresses, the data update lags severely. On the other hand, the accuracy of manual detection is greatly affected by subjective factors. It is difficult to unify the judgment criteria for breakage and impurities among different detection personnel, resulting in large deviations in the evaluation results, which cannot provide a reliable basis for precision agricultural management, and thus affect the optimization of subsequent operation plans and equipment improvement decisions. Summary of the Invention

[0003] The purpose of the present invention is to address the significant deficiencies such as the accuracy of traditional pepper harvesting operation quality assessment being easily interfered by human factors, incomplete data acquisition, low quality assessment efficiency, and difficulty in quickly and accurately obtaining key information during the pepper harvesting process, such as the breakage rate, impurity rate, and crop dropping loss rate, in large-scale pepper planting areas, and being unable to meet the requirements of modern agricultural production for precision management and efficient decision-making. A comprehensive evaluation system and evaluation method for pepper harvesting operation quality are proposed. Through the collaborative operation of an unmanned aerial vehicle (UAV) and an on-board detection system of a harvester, and through sensing, positioning, and data processing technologies, automatic and high-precision monitoring and comprehensive evaluation of the pepper harvesting operation quality can be achieved.

[0004] The comprehensive evaluation system for pepper harvesting operation quality provided by this application adopts the following technical solutions:

[0005] A comprehensive evaluation system for pepper harvesting operation quality, characterized in that the evaluation system includes:

[0006] A UAV on-board detection system, which consists of an imaging acquisition module and a first satellite RTK-GPS positioning module;

[0007] The imaging acquisition module is used to obtain image information of the operation scene after the harvesting operation;

[0008] With the first satellite RTK-GPS positioning module, the drone can obtain high-precision geographic coordinate data, provide accurate geographic location and flight altitude information, ensure that the collected images have accurate spatial positioning information, and facilitate subsequent stitching and analysis.

[0009] The on-board detection system of the harvester consists of a vehicle-mounted sensor unit, a harvester positioning unit, and a data transmission and processing unit connected in parallel.

[0010] The vehicle-mounted sensor unit consists of a breakage rate sensor, an impurity rate sensor, and a vehicle-mounted data communication bus.

[0011] The breakage rate sensor is used to detect whether the chili peppers are damaged during the harvesting process and provide breakage rate information.

[0012] The impurity rate sensor is used to detect non-target substances such as weeds and soil mixed in the harvested product, so as to obtain the impurity rate information of the operation.

[0013] The data communication bus is used to transmit the breakage rate and impurity rate information to the data transmission and processing module in real time.

[0014] The harvester positioning unit consists of a second satellite RTK-GPS positioning module. Through the positioning module, the harvester can obtain accurate geographic coordinates in real time, so as to track and evaluate the quality of the harvesting operation.

[0015] The data transmission and processing unit consists of a data communication module, an operation quality information processing module, and a data storage and analysis module connected in parallel.

[0016] The data communication module is responsible for data transmission and communication between various devices in the entire system, and transmits the data flow between the imaging acquisition module, the vehicle-mounted sensor unit, and the operation quality information processing module in real time to ensure the synchronous operation of the system information.

[0017] The operation quality information processing module is responsible for receiving and processing the drone image data and vehicle-mounted sensor data, processing and analyzing the data, generating an operation quality report, and providing decision support for subsequent optimization.

[0018] The data storage and analysis module is responsible for data storage of the entire system, analysis of historical data, and monitoring of long-term trends, ensuring the integrity and security of the data, and providing support for subsequent analysis to help optimize the operation plan and equipment configuration.

[0019] By adopting the above technical solution, the problems of low efficiency and poor accuracy in manual detection during the traditional quality assessment of pepper harvesting operations are solved. Through the collaboration of the onboard detection systems of the unmanned aerial vehicle (UAV) and the harvester, automatic and accurate monitoring of key indicators such as the breakage rate and impurity content rate during the pepper harvesting process is achieved, as well as effective acquisition and analysis of the operation scene images. With the help of the data transmission and processing unit, efficient data flow and comprehensive processing are completed, a operation quality report is generated and subsequent operation optimization is supported, improving the intelligent level and overall efficiency of the quality assessment of pepper harvesting operations.

[0020] The comprehensive evaluation method for the quality of pepper harvesting operations provided by this application adopts the following technical solution:

[0021] A comprehensive evaluation method for the quality of pepper harvesting operations, characterized by including the following steps:

[0022] (1) During the pepper harvesting operation, the breakage rate and impurity content rate of peppers are obtained through the vehicle-mounted sensor unit, and at the same time, the trajectory data of the harvester, the trajectory and path data of the UAV are respectively collected through the satellite RTK-GPS positioning module;

[0023] (2) After the harvesting operation is completed, the UAV flies into the operation area and conducts historical path tracing flight along the operation path of the harvester, and uses the imaging acquisition module to collect operation scene images at different positioning points;

[0024] (3) Through the positioning information and operation trajectory of the RTK-GPS positioning module, the operation scene images are matched with the operation trajectory, and multiple operation scene images are stitched into a complete operation area image through image stitching technology;

[0025] (4) The stitched image is processed using machine vision algorithms to identify the fallen peppers in the image and calculate the number of fallen peppers per unit area;

[0026] (5) An evaluation index for the crop dropping loss rate is introduced. This evaluation index calculates the loss rate based on the number of fallen peppers per unit area and divides the loss rate into three levels: high, medium, and low;

[0027] (6) According to the evaluation results, operation quality evaluation information is generated through the vehicle-mounted sensor data and the UAV image data, and the operation quality is uploaded to the data storage and analysis module.

[0028] By adopting the above technical solution, the problem that the traditional evaluation method cannot achieve dynamic and continuous monitoring of the key indicators (breakage rate, impurity content rate, crop dropping loss rate) of the entire process of pepper harvesting operations is solved. Through orderly operations at different stages during and after the harvesting process, a complete operation quality data chain is formed, making up for the defect that only static and single-point detection could be carried out in the past.

[0029] Further, the specific method for obtaining the chili damage rate in step (1) is as follows:

[0030] Through the imaging acquisition module and image vision processing algorithm, the surface and internal damage of chili peppers are monitored and analyzed in real time. Combining with the machine learning model to identify the damaged areas, thereby accurately judging the damage rate of chili peppers during the harvesting process; through multi-angle acquisition by the imaging acquisition module to capture the surface details of chili peppers, especially under different lighting conditions, to ensure the quality and clarity of the images. To comprehensively obtain different-angle information of chili peppers and avoid missing damaged areas due to shooting angle problems, the system can be configured with multiple cameras to shoot chili peppers from different perspectives. By rotating or moving the platform, a full-range inspection of the surface and internal of chili peppers can be achieved. The acquired images need to be subjected to a series of preprocessing, such as denoising, contrast adjustment, edge enhancement, etc., to improve the recognition rate of damaged areas. The damage recognition algorithm uses the convolutional neural network model in deep learning to train and optimize the algorithm specifically for chili pepper damage detection. Through a large amount of labeled data, the model can identify damaged features such as cracks, depressions, and color changes on the surface of chili peppers. According to the image data collected by the camera and combined with the algorithm recognition results, the damage rate of chili peppers is calculated.

[0031] By adopting the above technical solution, the problems of inaccurate and incomplete detection of the damage rate during the chili harvesting process are solved. By combining imaging acquisition with advanced algorithms, the limitations of lighting and shooting angle are overcome. Using multiple cameras and a movable platform to achieve a full-range inspection, thereby accurately calculating the damage rate, improving the accuracy and reliability of damage detection, and providing more accurate data support for the quality assessment of chili harvesting operations.

[0032] Further, the specific method for obtaining the impurity content rate of chili peppers in step (1) is as follows:

[0033] Adopt a near-infrared spectroscopy sensor and an automated weight distribution system. The near-infrared spectroscopy sensor can quickly identify and classify substances in the harvested products, capture the chemical composition characteristics of different substances through the reflection spectrum. The molecular structures of substances such as chili peppers, weeds, and soil have different reflection characteristics in the near-infrared band, and the reflection spectrum is used to distinguish these substances. Combining with the automated weight distribution system, the weight change of the material after passing through the cleaning device is monitored by the sensor to identify the ratio of impurities.

[0034] By adopting the above technical solution, this technical solution solves the problem that it is difficult to accurately measure the impurity content rate in pepper harvesting operations. Utilizing the near-infrared spectrum sensor to quickly identify and classify substances based on the reflection characteristic differences of the molecular structures of substances in the near-infrared band, combined with the automated weight distribution system to monitor the weight change after cleaning, effectively overcomes the difficulties of traditional methods in distinguishing peppers from impurities such as weeds and soil, realizes the accurate quantification of the impurity content rate, provides reliable impurity content rate data for the evaluation of pepper harvesting quality, and improves the accuracy and scientificity of operation quality evaluation.

[0035] Further, the specific method for collecting the trajectory data of the harvester, the trajectory and path data of the drone in step (1) is as follows:

[0036] (5-1) The method for collecting the harvester trajectory data is as follows:

[0037] (5-1-1) Equipment installation and parameter setting. Install the second RTK-GPS positioning module on the harvester, set its data collection frequency, for example, collect data once per second. This system can obtain the longitude and latitude coordinates of the harvester in the geographical coordinate system (WGS84 coordinate system) in real time, and at the same time record the corresponding timestamp t, as well as the driving speed v of the harvester. Each data point collected is represented as (lon i , lat i , t i , v i ), where i represents the collection order;

[0038] (5-1-2) Data recording and storage. During the harvesting operation of the harvester, the second RTK-GPS positioning module continuously collects data and stores these data in the local storage device to form a trajectory point sequence {(lon1, lat1, t1, v1), (lon2, lat2, t2, v2),..., (lon n , lat n , t n , v n )};

[0039] (5-1-3) Data transmission and reception. Adopt a stable wireless transmission technology (such as 4G, 5G or a dedicated frequency band) to transmit the trajectory point data collected by the second RTK-GPS positioning module on the harvester to the drone ground control station in real time. After the ground control station receives the data, it analyzes and verifies the data to ensure the integrity and accuracy of the data, and stores it in the local database for subsequent processing;

[0040] (5-2) The method for processing and generating the trajectory and path data of the drone is as follows:

[0041] (5-2-1) Coordinate system conversion, converting the longitude and latitude coordinates received in the WGS84 coordinate system into coordinates (x i , y i ) in the plane rectangular coordinate system. The converted trajectory points are represented as (x i , y i , t i , v i );

[0042] (5-2-2) Calculate the angle change. For three adjacent trajectory points P i-1 (x i-1 , y i-1 ), P i (x i , y i ), P i+1 (x i+1 , yi +1 ), calculate the vectors and . Calculate the cosine value i of their included angle α through the vector dot product formula, and then obtain the angle α i through the inverse trigonometric function;

[0043] (5-2-3) Classification and judgment. Set the angle threshold θ (such as 45°). When |α i | > θ, it is determined that the segment where the point P i is located is turning data; when |α i | ≤ θ, it is determined as straight-line data;

[0044] (5-2-4) Linear fitting (for straight-line data). For the classified straight-line data point set {(x j1 , y j1 , t j1 , v j1 ), (x j2 , y j2 , t j2 , v j2 ),..., (x jm , y jm , t jm , v jm )} (j represents the index of the straight-line data points), use the least squares method for linear fitting. Assume the straight-line equation is y = kx + b. According to the least squares principle, solve the system of equations to obtain the slope k and intercept b of the straight line, thereby determining the straight-line equation;

[0045] (5-2-5) Straight-line segment generation: For the fitted straight line, a series of points are evenly selected on the straight line as the flight path points of the UAV according to the flight parameters of the UAV (such as flight speed, safety distance, etc.); for example, points are successively taken on the straight line at a fixed distance d (such as 5 meters). Let the starting point be (x start , y start ), then the subsequent points can be calculated by the formula (x next , y next ) = (x start + dcosβ, y start + dcosβ), where β is the inclination angle of the straight line and tanβ = k;

[0046] (5-2-6) Turning segment generation: For the turning data points, according to the minimum turning radius R of the UAV, an arc is drawn with the turning point as the center and R as the radius for path transition; a series of points on the arc are determined through geometric calculations as the flight path points to ensure the UAV can fly smoothly at the turning; the path points of the straight-line segment and the turning segment are connected in sequence to form a complete UAV flight trajectory.

[0047] By adopting the above technical solution, the problem of being unable to accurately obtain the operation trajectories and path data of the harvester and the UAV in the pepper harvesting operation is solved. By installing RTK-GPS positioning modules on the harvester and the UAV and standardizing the data acquisition, storage, transmission, and processing processes, the accurate recording and conversion of information such as their positions and speeds are realized, ensuring the integrity and accuracy of the trajectory data, providing key basic data support for subsequent operation quality evaluation, image acquisition and matching, etc., and improving the reliability and effectiveness of the entire operation monitoring and evaluation system.

[0048] Further, the specific method of the image stitching technology in step (3) is as follows:

[0049] (6-1) Through the RTK-GPS positioning system, the geographical coordinates of each image acquisition point can be accurately obtained;

[0050] (6-2) Image acquisition: The UAV takes X images, and each image contains a part of the operation area and has accurate RTK positioning information;

[0051] (6-3) Feature extraction and matching: By extracting the feature points in the images and using the matching algorithm (SIFT), the matching points between adjacent images are found;

[0052] (6-4) Image registration and stitching: According to the matching feature points, the transformation matrix of each image is calculated using the homography transformation, and all the images are synthesized into a complete operation area image of Y1 meters × Y2 meters;

[0053] (6-5) Image fusion: fuse the joints to eliminate seams and form a seamless image of the work area.

[0054] By adopting the above technical solution, the problem of difficulty in integrating pepper harvesting scene images into a complete and seamless image was solved, providing a comprehensive and accurate visual data foundation for the subsequent pepper harvesting operation quality analysis based on the overall image, thereby improving the accuracy and effectiveness of the evaluation.

[0055] Furthermore, the specific method of the visual recognition algorithm described in step (4) is as follows: the ground pepper fruits are quickly identified and located in the image through a neural network, and the image can be processed in real time for the quality assessment of pepper harvesting operations, and the fallen peppers can be accurately identified and located, and their number and distribution can be calculated. By predicting the bounding box and classification information of each area, accurate detection results are provided to help monitor and evaluate the quality of harvesting operations.

[0056] By adopting the above technical solution, the problem of difficulty in quickly and accurately identifying and counting the number and distribution of peppers dropped on the ground during pepper harvesting operations was solved, providing key data support for accurately evaluating the quality of pepper harvesting operations, and helping to promptly discover operational problems and take improvement measures.

[0057] Furthermore, the loss rate evaluation method described in step (5) is as follows: the loss rate is evaluated by calculating the number of peppers dropped per unit area, and the loss rate is divided into three levels: high, medium and low. The number of crops dropped per unit area (such as the number of pepper grains) is used as the evaluation standard, and the loss rate level is divided according to the number of drops, among which, high loss rate: the number of peppers dropped per unit area is greater than X1 grains / square meter; medium loss rate: the number of peppers dropped per unit area is X2-X1 grains / square meter; low loss rate: the number of peppers dropped per unit area is less than X2 grains / square meter; wherein X1>X2.

[0058] By adopting the above technical solution, the problem of difficult quantitative evaluation of crop drop loss rate in pepper harvesting operation is solved. Based on the number of peppers dropped per unit area, the loss rate is divided into three levels: high, medium and low. This overcomes the shortcomings of traditional evaluation that lacks clear quantitative indicators and grade divisions, provides a clear basis for intuitive and accurate judgment of harvesting operation quality, and facilitates timely adjustment of operation strategies to reduce losses.

[0059] In summary, the present invention includes at least one of the following beneficial technical effects:

[0060] (1) High-precision positioning and data acquisition: Through the RTK-GPS positioning module, the system can obtain high-precision geographic coordinate data of drones and harvesters in real time, ensuring that the collected images and sensor data have accurate spatial positioning information. This provides a reliable foundation for subsequent image stitching, data analysis, and operation quality assessment.

[0061] (2) Multi-dimensional data collection and analysis: The system integrates an airborne detection system for drones and an airborne detection system for harvesters, capable of simultaneously collecting various data such as images of the operation scene, chili damage rate, and impurity rate. Through multi-dimensional data collection and analysis, the system can comprehensively evaluate the quality of the harvesting operation and provide more comprehensive decision-making support.

[0062] (3) Real-time data processing and transmission: The system realizes real-time data transmission and communication between various devices through a data communication module, ensuring the synchronous operation of system information. The real-time data processing ability enables the system to detect problems in a timely manner during the operation and provide optimization suggestions, thereby improving the operation efficiency and quality.

[0063] (4) Image stitching and visual recognition technology: The system adopts advanced image stitching technology and machine vision algorithms, which can stitch multiple images of the operation scene into a complete image of the operation area and quickly identify and locate ground chili fruits through a neural network. This not only improves the efficiency of image processing but also ensures the accuracy of image analysis.

[0064] (5) Intelligent evaluation and decision-making support: The system can perform intelligent processing and analysis on the collected data through an operation quality information processing module and a data storage and analysis module, generate an operation quality report, and provide decision-making support for subsequent optimization. This helps farmers or operators adjust the operation plan according to the evaluation results, optimize equipment configuration, and reduce crop losses.

[0065] (6) Loss rate classification evaluation: The system introduces an evaluation index for the crop dropping loss rate and divides the loss rate into three levels: high, medium, and low. By quantitatively evaluating the crop loss rate, the system can help users more intuitively understand the operation quality and take corresponding improvement measures, thereby reducing crop losses and increasing economic benefits.

[0066] (7) Automation and intelligent operation: The system reduces the need for manual intervention through automated sensors and intelligent algorithms, improving the automation level of the operation. This not only reduces labor costs but also reduces human errors and improves the accuracy and consistency of the operation.

[0067] (8) Long-term data storage and trend analysis: The system has a data storage and analysis function, capable of storing historical data for a long time and conducting trend analysis. By monitoring long-term data, the system can help users discover potential problems in the operation and provide data support for future operation planning. Description of the Drawings

[0068] Figure 1 It is a diagram of the overall system composition of the present invention.

[0069] Figure 2 This is the flowchart for UAV identification in the present invention.

[0070] Figure 3 This is the overall flowchart for operation quality evaluation in the present invention. Specific implementation manners

[0071] The present invention will be further clarified below in conjunction with the accompanying drawings and specific implementation manners. It should be understood that these implementation manners are only used to illustrate the present invention patent and not to limit the scope of the present invention patent. After reading the present invention patent, various equivalent forms of modification by those skilled in the art fall within the scope defined by the appended claims of the present application.

[0072] Embodiment

[0073] Overview of the embodiment

[0074] This embodiment details the application of the airborne sensing comprehensive evaluation system in actual chili harvesting operations through specific parameters and operation steps. Suppose there is a chili planting area of 10 hectares. The system, through the collaborative work of the UAV and the harvester, collects data in real time and conducts operation quality assessment.

[0075] System composition and parameter setting

[0076] UAV airborne detection system:

[0077] Imaging acquisition module: Equipped with a 4K high-definition camera with a resolution of 3840×2160 and a frame rate of 30fps, capable of capturing the details of the chili surface.

[0078] The first satellite RTK-GPS positioning module: The positioning accuracy is ±1 cm, and the data update frequency is 10 Hz, ensuring the accuracy of the UAV flight path and image acquisition.

[0079] Harvester airborne detection system:

[0080] Damage rate sensor: Adopts a high-resolution camera (1080p) and an infrared sensor, capable of monitoring the damage situation of chili in real time, and the damage rate detection accuracy is ±0.5%.

[0081] Impurity rate sensor: Adopts a near-infrared spectroscopy sensor with a wavelength range of 900 - 1700 nm, capable of quickly identifying chili, weeds, and soil, and the impurity rate detection accuracy is ±1%.

[0082] The second satellite RTK-GPS positioning module: The positioning accuracy is ±2 cm, and the data update frequency is 5 Hz, recording the trajectory data of the harvester in real time.

[0083] Data transmission and processing unit:

[0084] Data communication module: Adopts 5G wireless transmission technology with a data transmission rate of 1Gbps to ensure real-time data transmission.

[0085] Job quality information processing module: Adopts a high-performance processor (such as Intel Xeon E5-2680) with a processing speed of 2.5GHz, capable of real-time processing of a large amount of image and sensor data.

[0086] Data storage and analysis module: Equipped with 1TB of SSD storage, capable of storing at least 3 months of job data and supporting long-term trend analysis.

[0087] Specific implementation steps

[0088] Equipment installation and parameter setting:

[0089] Install the second RTK-GPS positioning module on the harvester, set the data collection frequency to once per second, and obtain the longitude, latitude coordinates, timestamp, and driving speed of the harvester in real time.

[0090] Install the first RTK-GPS positioning module and imaging acquisition module on the UAV to ensure that the UAV can obtain high-precision geographic coordinate data and collect high-quality job scene images.

[0091] Data collection and transmission:

[0092] Harvester trajectory data collection: During the operation of the harvester, the second RTK-GPS positioning module continuously collects trajectory data. Assuming the driving speed of the harvester is 5km / h and data is collected once per second, the sequence of data points collected is:

[0093] {(lon1, lat1, t1, v1), (lon2, lat2, t2, v2),..., (lonn, latn, tn, vn)}

[0094] Where lon and lat are longitude and latitude coordinates, t is the timestamp, and v is the driving speed.

[0095] UAV image collection: After the harvesting operation is completed, the UAV conducts a tracing flight along the operation path of the harvester to collect job scene images. Assuming the flight altitude of the UAV is 50 meters, the flight speed is 10m / s, and one image is collected every 5 seconds, a total of 200 images are collected.

[0096] Image stitching and visual recognition:

[0097] Image stitching: Through the RTK-GPS positioning system, accurately obtain the geographical coordinates of each image acquisition point. Use the SIFT algorithm to extract the feature points in the images, and stitch 200 images into a complete image of the operation area through homography transformation. The size of the image is 100 meters × 100 meters.

[0098] Visual recognition: Use a convolutional neural network (CNN) to process the stitched image, identify and locate the fallen chili fruits. Assuming that the number of fallen chilies per square meter is 10, the system can accurately identify and calculate the total number of fallen ones.

[0099] Loss rate evaluation and report generation:

[0100] Loss rate evaluation: Calculate the crop drop loss rate based on the number of fallen chilies per unit area. Assuming that the number of fallen chilies per square meter is 10, the loss rate is divided into:

[0101] High loss rate: The number of fallen chilies per unit area > 15 grains per square meter;

[0102] Medium loss rate: The number of fallen chilies per unit area is 5 - 15 grains per square meter;

[0103] Low loss rate: The number of fallen chilies per unit area < 5 grains per square meter;

[0104] According to the calculation results, the system determines that the loss rate of the current operation area is a medium loss rate.

[0105] Report generation: The operation quality information processing module generates an operation quality report based on the vehicle-mounted sensor data and UAV image data. The report content includes:

[0106] Chili damage rate: 2.5%;

[0107] Impurity content rate: 3.8%;

[0108] Crop drop loss rate: Medium loss rate (10 grains per square meter);

[0109] Operation efficiency: Harvest 5 hectares per hour;

[0110] Environmental impact: Low.

[0111] Data storage and trend analysis:

[0112] Data storage: All collected data (including trajectory data, image data, damage rate, impurity content rate, etc.) is stored in the data storage and analysis module. The storage capacity is 1TB, which can store at least 3 months of operation data.

[0113] Trend Analysis: Through the analysis of historical data, the system found that the chili damage rate has gradually decreased in the past three months, the impurity content has remained stable, and the crop dropping loss rate has increased. The system recommends optimizing the operating speed of the harvester and the working parameters of the cleaning device to further reduce the loss rate.

[0114] System Advantages and Effects

[0115] High-precision Positioning and Data Acquisition: Through the RTK-GPS positioning module, the system can obtain the high-precision geographical coordinate data of the unmanned aerial vehicle and the harvester in real time, ensuring that the collected images and sensor data have accurate spatial positioning information.

[0116] Multi-dimensional Data Acquisition and Analysis: The system can simultaneously collect various data such as images of the operation scene, chili damage rate, and impurity content to comprehensively evaluate the quality of the harvesting operation.

[0117] Real-time Data Processing and Transmission: Through the 5G wireless transmission technology, the system realizes real-time data transmission and communication between various devices, ensuring the synchronous operation of the system information.

[0118] Intelligent Evaluation and Decision Support: The system can perform intelligent processing and analysis on the collected data, generate an operation quality report, and provide decision support for subsequent optimization.

[0119] Automation and Intelligent Operation: Through automated sensors and intelligent algorithms, the system reduces the need for manual intervention and improves the automation level of the operation.

[0120] This embodiment details the application of the airborne sensing comprehensive evaluation system in actual chili harvesting operations through specific parameters and operation steps. Through technical means such as high-precision positioning, multi-dimensional data acquisition, real-time data processing, and intelligent evaluation, the system significantly improves the quality evaluation efficiency and accuracy of chili harvesting operations, provides strong technical support for farmers and operators, helps reduce crop losses, improve operation efficiency, and optimize resource allocation.

Claims

1. A comprehensive evaluation system for pepper harvesting quality, characterized in that: The evaluation system includes: The UAV airborne detection system consists of an imaging acquisition module and a first satellite RTK-GPS positioning module; The imaging acquisition module is used to obtain image information of the operation scene after the harvesting operation; The first satellite RTK-GPS positioning module enables the drone to obtain high-precision geographic coordinate data, provide accurate geographic location and flight altitude information, and ensure that the collected images have accurate spatial positioning information, which is convenient for subsequent stitching and analysis; The harvester onboard detection system is composed of an onboard sensor unit, a harvester positioning unit and a data transmission and processing unit in parallel; The vehicle-mounted sensor unit is composed of a damage rate sensor, a foreign matter rate sensor, and a vehicle-mounted data communication bus; The damage rate sensor is used to detect whether the peppers are damaged during the harvesting process and provide damage rate information; The impurity sensor is used to detect non-target substances such as weeds and soil mixed in the harvest, so as to obtain the impurity information of the operation; The data communication bus is used to transmit the damage rate and impurity rate information to the data transmission and processing module in real time; The harvester positioning unit is composed of a second satellite RTK-GPS positioning module, through which the harvester obtains accurate geographic coordinates in real time, so as to track and evaluate the quality of the harvesting operation; The data transmission and processing unit is composed of a data communication module, an operation quality information processing module, and a data storage and analysis module in parallel; The data communication module is responsible for data transmission and communication between various devices in the entire system, and transmits the data stream between the imaging acquisition module, the vehicle-mounted sensor unit and the operation quality information processing module in real time to ensure the synchronization of system information; The operation quality information processing module is responsible for receiving and processing the image data from the drone and the vehicle-mounted sensor data, processing and analyzing the data, generating an operation quality report, and providing decision support for subsequent optimization; The data storage and analysis module is responsible for the data storage of the entire system, the analysis of historical data and the monitoring of long-term trends, ensuring the integrity and security of the data, providing support for subsequent analysis, and helping to optimize operation plans and equipment configurations.

2. A comprehensive evaluation method for pepper harvesting operation quality, characterized in that: The following steps are involved: (1) During the pepper harvesting process, the damage rate and impurity rate of peppers are obtained through the vehicle-mounted sensor unit, and the trajectory data of the harvester and the trajectory and path data of the drone are collected through the satellite RTK-GPS positioning module; (2) After the harvesting operation is completed, the UAV flies into the operation area and patrols the historical path along the operation path of the harvester, using the imaging acquisition module to collect images of the operation scene at different positioning points; (3) Match the work scene image with the work trajectory through the positioning information and work trajectory of the RTK-GPS positioning module, and stitch multiple work scene images into a complete work area image through image stitching technology; (4) using a machine vision algorithm to process the spliced ​​images, identify the fallen peppers in the images, and calculate the number of fallen peppers per unit area; (5) Introducing a crop drop loss rate evaluation index, which calculates the loss rate based on the number of peppers dropped per unit area and divides the loss rate into three levels: high, medium, and low; (6) Based on the evaluation results, the operation quality evaluation information is generated through the vehicle-mounted sensor data and the drone image data, and the operation quality is uploaded to the data storage and analysis module.

3. The method for comprehensive evaluation of pepper harvesting quality according to claim 2, characterized in that: The specific method for obtaining the pepper damage rate in step (1) is as follows: Through the imaging acquisition module and image visual processing algorithm, the surface and internal damage of the pepper are monitored and analyzed in real time, and the damaged area is identified by combining the machine learning model, so as to accurately judge the damage rate of the pepper during the harvesting process; through the multi-angle acquisition of the imaging acquisition module, the surface details of the pepper are captured, especially under different lighting conditions, to ensure the quality and clarity of the image. In order to fully obtain the different angle information of the pepper and avoid missing the damaged area due to the shooting angle problem, the system can be configured with multiple cameras to shoot the pepper from different perspectives. By rotating or moving the platform, a full-scale inspection of the surface and interior of the pepper can be achieved. The acquired images need to undergo a series of preprocessing, such as denoising, contrast adjustment, edge enhancement, etc., to improve the recognition of the damaged area. The damage recognition algorithm uses the convolutional neural network model in deep learning to train and optimize the algorithm specifically for pepper damage detection. Through a large amount of labeled data, the model can identify the cracks, dents, color changes and other damage features on the surface of the pepper. According to the image data collected by the camera and combined with the algorithm recognition results, the damage rate of the pepper is calculated.

4. The method for comprehensive evaluation of pepper harvesting quality according to claim 2, characterized in that: The specific method for obtaining the impurity content of pepper in step (1) is as follows: Near-infrared spectral sensors and automated weight distribution systems are used. The near-infrared spectral sensors can quickly identify and classify the harvested materials, and capture the chemical composition characteristics of different substances through reflection spectroscopy. The molecular structures of substances such as peppers, weeds, and soil have different reflection characteristics in the near-infrared band. Reflection spectroscopy is used to distinguish these substances. Combined with the automated weight distribution system, the sensor monitors the weight change of the material after passing through the cleaning device and identifies the proportion of impurities.

5. The method for comprehensive evaluation of pepper harvesting quality according to claim 2, characterized in that: The specific method of collecting the trajectory data of the harvester, the trajectory and path data of the drone in step (1) is as follows: (5-1) The method for collecting harvester trajectory data is as follows: (5-1-1) Equipment installation and parameter setting: Install the second RTK-GPS positioning module on the harvester, set its data collection frequency, obtain the longitude and latitude coordinates of the harvester in the geographic coordinate system in real time, and record the corresponding timestamp t and the driving speed v of the harvester. The data points collected each time are expressed as (lon i ,lat i , t i , v i ), where i represents the order of collection; (5-1-2) Data recording and storage. During the harvesting operation, the second RTK-GPS positioning module continuously collects data and stores the data in a local storage device to form a trajectory point sequence {(lon1, lat1, t1, v1), (lon2, lat2, t2, v2), ..., (lon n ,lat n , t n , v n )}; (5-1-3) Data transmission and reception: using stable wireless transmission technology, the track point data collected by the second RTK-GPS positioning module on the harvester is transmitted to the UAV ground control station in real time. After receiving the data, the ground control station analyzes and verifies the data to ensure its integrity and accuracy, and stores it in the local database for subsequent processing; (5-2) The processing and generation method of the trajectory and path data of the drone is as follows: (5-2-1) Coordinate system conversion: convert the received longitude and latitude coordinates in the WGS84 coordinate system into coordinates in the plane rectangular coordinate system (x i ,y i ), the transformed trajectory point is expressed as (x i ,y i , t i , v i ); (5-2-2) Calculate the angle change for three adjacent trajectory points P i-1 (x i-1 ,y i-1 ), P i (x i ,y i ), P i+1 (x i+1 , +1 ), calculate the vector and Calculate their angle α through the vector dot product formula i The cosine value of Then use the inverse trigonometric function to get the angle α i ; (5-2-3) Classification judgment, set the angle threshold θ, when |α i |>θ, the decision point P i The segment is the turning data; when |α i When |≤θ, it is determined to be straight data; (5-2-4) Linear fitting: For the straight line data, for the classified straight line data point set {(x j1 ,y j1 , t j1 , v j1 ), (x j2 ,y j2 , t j2 , v j2 ),...,(x jm ,y jm , t jm , v jm )}, where j represents the index of the straight line data point. The least square method is used for straight line fitting. The straight line equation is y=kx+b. According to the principle of least square method, by solving the equation group Get the slope k and intercept b of the straight line, and thus determine the equation of the straight line; (5-2-5) Straight line segment generation. For the fitted straight line, according to the flight parameters of the UAV, a series of points are uniformly selected on the straight line as the flight path points of the UAV; points are selected on the straight line in sequence at a fixed distance d, and the starting point is set as (x start ,y start ), then the subsequent points are calculated by the formula (x next ,y next )=(x start +dcosβ,y start +dcosβ), where β is the inclination angle of the straight line, tanβ=k; (5-2-6) Turning segment generation: For turning data points, according to the minimum turning radius R of the drone, an arc is drawn with the turning point as the center and R as the radius to perform path transition; a series of points on the arc are determined through geometric calculations as flight path points to ensure that the drone can fly smoothly at the turn; the path points of the straight segment and the turning segment are connected in sequence to form a complete drone flight trajectory.

6. The method for comprehensive evaluation of pepper harvesting quality according to claim 2, characterized in that: The specific method of the image stitching technology in step (3) is as follows: (6-1) Through the RTK-GPS positioning system, the geographic coordinates of each image acquisition point can be accurately obtained; (6-2) Image collection: The drone takes X images, and each image contains a part of the work area and has accurate RTK positioning information; (6-3) Feature extraction and matching: extracting feature points from images and using matching algorithms to find matching points between adjacent images; (6-4) Image registration and stitching: Based on the matched feature points, the transformation matrix of each image is calculated using homography transformation, and all images are synthesized into a complete work area image of Y1m × Y2m; (6-5) Image fusion: fuse the joints to eliminate seams and form a seamless image of the work area.

7. The method for comprehensive evaluation of pepper harvesting quality according to claim 2, characterized in that: The specific method of the visual recognition algorithm described in step (4) is as follows: quickly identify and locate ground pepper fruits in the image through a neural network, process images in real time for the quality assessment of pepper harvesting operations, accurately identify and locate fallen peppers, calculate their number and distribution, and provide accurate detection results by predicting the bounding box and classification information of each area, thereby helping to monitor and evaluate the quality of harvesting operations.

8. The method for comprehensive evaluation of pepper harvesting quality according to claim 2, characterized in that: The evaluation method of the loss rate described in step (5) is as follows: the loss rate is evaluated by calculating the number of peppers dropped per unit area, and the loss rate is divided into three levels: high, medium and low. The number of crops dropped per unit area is used as the evaluation standard, and the loss rate level is divided according to the number of drops, among which, high loss rate: the number of peppers dropped per unit area is greater than X1 grains / square meter; medium loss rate: the number of peppers dropped per unit area is X2-X1 grains / square meter; low loss rate: the number of peppers dropped per unit area is less than X2 grains / square meter; wherein X1>X2.

Images