Mulching film residue monitoring method and system based on multi-source data fusion

Abstract

The invention discloses a mulching film residue monitoring method and system based on multi-source data fusion, and belongs to the technical field of chemical detection, and the method comprises the following steps: monitoring a farmland region at different observation track points according to a preset navigation route based on an unmanned aerial vehicle to obtain low-altitude remote sensing observation data and a low-altitude remote sensing monitoring image, the plastic film mulching state of each block is obtained through analysis, and the type of each block is obtained through statistics; based on the type of each block and / or low-altitude remote sensing observation data analysis, remote sensing sampling frequency adjustment is executed until remote sensing monitoring is completed, and a farmland global remote sensing image is obtained; based on remote sensing observation data analysis of a farmland global remote sensing image, a sampling density execution quota is determined, sampling points are arranged, mulching film residue monitoring is carried out on each sampling point, and a mulching film residue value of a farmland area is obtained. The technical problem that the accuracy of a mulching film residue monitoring result is affected due to unreasonable arrangement of sampling points for mulching film residue detection in the prior art is solved.

Description

Technical Field

[0001] This invention relates to the field of chemical detection technology, and in particular to a method and system for monitoring residual plastic film based on multi-source data fusion. Background Technology

[0002] Monitoring residual plastic film refers to the technical process of detecting and assessing the distribution and degree of residual plastic film in farmland soil and surface at the end of the agricultural production cycle or during the plastic film recycling stage. Existing methods for monitoring residual plastic film typically focus on ground sampling and experimental analysis. This involves selecting several fixed sampling points within the farmland area, sampling the topsoil, sorting the samples to separate residual plastic film, and then obtaining relevant parameters through steps such as counting the number of pieces, weighing, rinsing, drying, and constant weight treatment. Based on these parameters, the amount of residual plastic film or residual pollution indicators can be calculated. Alternatively, low-altitude imaging of the farmland area can be performed using image sensors mounted on drones, and the percentage of area covered by plastic film in the images can be analyzed to obtain a quantitative result of the amount of residual plastic film on the farmland surface within the measurement area.

[0003] For example, Chinese invention patent CN104007039B discloses a method for monitoring the residual pollution coefficient of plastic film in topsoil, including: step a, selecting sampling points for sampling; step b, sample detection and processing; step b1, sample coarse sorting to remove soil clods and wet soil; step b2, sample fine sorting; step b3, counting the number of residual film pieces; step b4, counting the weight of residual film; step b6, rinsing the residual film; step b7, drying the residual film; step b8, fine sorting again; step b9, constant weight of residual film; step c, calculating the residual pollution coefficient of plastic film; step c1, calculating the residual amount; step c2, calculating the residual pollution coefficient.

[0004] The above-mentioned technology has at least the following technical problems:

[0005] Existing technologies mainly involve obtaining soil samples from the topsoil at fixed sampling points for residual film statistics, or remotely measuring residual film on the farmland surface using uniform image processing parameters. However, the layout of sampling points or the selection of image acquisition areas usually adopt preset rules or fixed methods. But there are differences in the spatial differences in film coverage, the non-uniformity of residual film distribution, and the differences in film types between different farmland areas. Therefore, the number, location, and sampling density of sampling points lack targeted configuration. At this time, if the same sampling or measurement strategy is still used under different farmland conditions, it is easy to make the obtained data unable to fully reflect the actual residual film status, thus affecting the representativeness and accuracy of the residual film monitoring results. Summary of the Invention

[0006] To address the technical problem in existing technologies where the accuracy of plastic film residue monitoring results is affected by unreasonable sampling point placement, this invention provides a method and system for monitoring plastic film residue based on multi-source data fusion. The technical solution is as follows:

[0007] On the one hand, a method for monitoring residual plastic film mulch based on multi-source data fusion is provided. This method includes: using a UAV to monitor farmland areas at different observation trajectory points according to a preset flight route, obtaining low-altitude remote sensing observation data and low-altitude remote sensing monitoring images, and analyzing the plastic film coverage status of each block to statistically determine the block types, including covered areas, bare areas, and coverage fluctuation areas. The low-altitude remote sensing observation data is used to characterize the fracture characteristics of the plastic film. Based on the analysis of each block type and / or low-altitude remote sensing observation data, the remote sensing sampling frequency is adjusted until remote sensing monitoring is completed to obtain a remote sensing image of the entire farmland area. Based on the remote sensing observation data analysis of the remote sensing image of the entire farmland area, the sampling density is determined and quotas are implemented, and sampling points are deployed. Plastic film residue monitoring is performed at each sampling point to obtain the residual value of plastic film in the farmland area. The residual value of plastic film is a statistical measure of the residual status formed based on the plastic film residue monitoring results of the sampling points.

[0008] On the other hand, a multi-source data fusion-based monitoring system for residual plastic film is provided. This system includes: a block assessment module, a data acquisition and adjustment module, and a residue monitoring module. The block assessment module is used to monitor farmland areas at different observation trajectory points based on a UAV following a preset flight route, obtain low-altitude remote sensing observation data and low-altitude remote sensing monitoring images, analyze the plastic film coverage status of each block, and statistically determine the block type. The data acquisition and adjustment module is used to adjust the remote sensing sampling frequency based on the analysis of each block type and / or low-altitude remote sensing observation data until the remote sensing monitoring is completed and a remote sensing image of the entire farmland area is obtained. The residue monitoring module is used to analyze the remote sensing observation data of the remote sensing image of the entire farmland area, determine the sampling density to implement the quota, deploy sampling points, and perform plastic film residue monitoring at each sampling point to obtain the plastic film residue value of the farmland area.

[0009] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0010] 1. The method for monitoring residual plastic film based on multi-source data fusion provided by this invention acquires low-altitude remote sensing observation data and low-altitude remote sensing monitoring images from different observation trajectory points using UAVs, and performs alignment analysis and consistency determination on the plastic film coverage status of the same block under different observation trajectories, thereby accurately distinguishing covered areas, bare areas, and areas with fluctuating coverage. Based on this, remote sensing sampling frequency adjustment, sampling density quota determination, and sampling point layout are executed in a coordinated manner, effectively solving the technical problem in the prior art where the accuracy of plastic film residue monitoring results is affected by unreasonable sampling point layout.

[0011] 2. This invention, after identifying the coverage fluctuation area, determines the block fluctuation area based on the corresponding area change of the coverage fluctuation area in different low-altitude remote sensing monitoring images, combined with the positive difference of the coverage fluctuation area after angle correction, and further obtains the comprehensive influence value of the coverage area. Based on the comprehensive influence value of the coverage area, hierarchical response processing is performed to dynamically obtain the first constraint factor of remote sensing observation and increase the remote sensing sampling frequency, thereby realizing targeted and intensified monitoring of areas with unstable mulch film coverage. This effectively solves the problems of fixed remote sensing monitoring frequency and difficulty in timely capturing mulch film breakage and migration changes in the existing technology.

[0012] 3. By introducing low-altitude remote sensing observation data, the remote sensing coverage integrity of each block is obtained through comprehensive analysis. When the remote sensing coverage integrity is lower than the threshold, a second constraint factor for remote sensing observation is obtained to increase the remote sensing sampling frequency. This enables joint constraint monitoring of the degree of mulch film breakage and coverage continuity, thereby improving the integrity and reliability of mulch film residue information in the remote sensing images of the entire farmland area. This effectively solves the problem of insufficient identification of areas with missing coverage caused by relying solely on a single image feature in existing technologies.

[0013] 4. After obtaining remote sensing images of the entire farmland area, the initial sampling density quota is matched based on the integrity of remote sensing coverage. The sampling density quota rules for non-degradable and degradable mulch films are applied according to the type of mulch film. At the same time, the sampling density quota is proportionally allocated based on the proportion of covered area, exposed area, and covered area fluctuation area, and sampling points are randomly deployed. This achieves a coordinated matching between sampling intensity and the risk level of mulch film residue and mulch film type, effectively solving the problem of distortion in residue assessment and early warning results caused by the failure to distinguish between mulch film type and block characteristics in existing technologies. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 A macroscopic flowchart of the method for monitoring residual plastic film based on multi-source data fusion provided in the embodiments of this application;

[0016] Figure 2 A flowchart illustrating the overall process of the method for monitoring residual plastic film based on multi-source data fusion provided in this application embodiment;

[0017] Figure 3A flowchart illustrating the collaborative processing of remote sensing sampling frequency adjustment and full-domain image construction in the multi-source data fusion-based method for monitoring residual plastic film provided in this application embodiment;

[0018] Figure 4 A schematic diagram of the structure of a plastic film residue monitoring system based on multi-source data fusion provided in an embodiment of this application;

[0019] Figure 5 A schematic diagram of mulch film identification in a multi-source data fusion-based mulch film residue monitoring system provided in an embodiment of this application;

[0020] Figure 6 A schematic diagram of the front of the same block of the mulch film residue monitoring system based on multi-source data fusion provided in the embodiments of this application;

[0021] Figure 7 This is a schematic diagram of the rear section of the same block of the mulch film residue monitoring system based on multi-source data fusion provided in the embodiments of this application. Detailed Implementation

[0022] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0023] In the embodiments of the present invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that when the distinction is not emphasized, their intended meanings are consistent.

[0024] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0025] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0026] In the first embodiment of this solution, as Figure 1 As shown, Figure 1This is a macroscopic flowchart of a multi-source data fusion-based method for monitoring residual plastic film in an embodiment of this application. The method includes the following steps: using a drone to monitor farmland areas at different observation trajectory points along a preset flight route, obtaining low-altitude remote sensing observation data and low-altitude remote sensing monitoring images, analyzing the plastic film coverage status of each block, and statistically determining the block types, including covered areas, bare areas, and coverage fluctuation areas. The low-altitude remote sensing observation data is used to characterize the fracture characteristics of the plastic film. Based on the analysis of each block type and / or the low-altitude remote sensing observation data, the remote sensing sampling frequency is adjusted until remote sensing monitoring is completed, obtaining a remote sensing image of the entire farmland area. Based on the remote sensing observation data analysis of the remote sensing image of the entire farmland area, the sampling density is determined and quotas are implemented, and sampling points are deployed. Plastic film residue monitoring is performed at each sampling point to obtain the residual value of the plastic film in the farmland area. The residual value of the plastic film is a statistical measure of the residual status formed based on the plastic film residue monitoring results of the sampling points.

[0027] In this embodiment, it should be noted that there are two types of mulch films: biodegradable mulch films and non-biodegradable mulch films. The non-biodegradable mulch films include 0.015 mm thickened high-strength mulch films, 0.012 mm thickened high-strength mulch films, and ordinary PE mulch films. Four types of fully biodegradable mulch films were tested: mulch films T4, T5, T6, and T7. There were four treatments, each with three replicates. The treatment numbers and names are shown in Table 1.

[0028] Table 1 is a statistical table of the naming of each treatment in the experiment.

[0029]

[0030] In the above, (I), (II) and (III) represent the number of repetitions.

[0031] The degradation status of the mulch film for each treatment is shown in Table 2. Treatment T7 entered the induction period first, with a faster degradation rate, and entered each stage first. Treatments T4 and T5 entered the induction period first, followed by treatment T6, which had the slowest degradation rate and entered the induction period last. Treatment T4 took a longer time to enter the cracking period, while treatments T5, T6, and T7 took a shorter time to enter the cracking period. The time interval between the cracking period and the large cracking period was roughly the same for each treatment.

[0032] Table 2 Survey on the Degradation Status of Mulch Film

[0033]

[0034] like Figure 2 As shown, Figure 2This is an overall flowchart of the method for monitoring residual plastic film based on multi-source data fusion provided in this application embodiment. It involves using a drone to perform remote sensing monitoring of farmland areas, collecting remote sensing images and observation data; performing block alignment processing on the collected remote sensing images to determine the corresponding identical observation blocks in each remote sensing image; extracting plastic film coverage features within the identical observation blocks and performing consistency judgment on the coverage features of each observation block in different remote sensing images; when coverage features are inconsistent, the corresponding observation block is marked as an unstable coverage block; when coverage features are consistent, the stable coverage state of the corresponding observation block is determined; based on the unstable coverage blocks and stable coverage states, the farmland area is divided into coverage area types; based on this, the remote sensing sampling frequency is adjusted according to the coverage area type, and a remote sensing image of the entire farmland area is generated; the remote sensing coverage integrity is evaluated based on the remote sensing image of the entire farmland area, and then the sampling density quota is determined; sampling points are deployed according to the sampling density quota, and plastic film residue detection is performed at each sampling point, finally outputting the monitoring results of plastic film residue in the farmland area.

[0035] Furthermore, the types of each block are obtained. Specifically, alignment analysis is performed on low-altitude remote sensing images of farmland areas at different observation trajectory points to identify corresponding identical blocks in each low-altitude remote sensing image. The mulch film coverage status within the same block is then identified using feature recognition, yielding results including mulch film coverage and uncovered mulch film. Based on these feature recognition results, a comparison is made. If the feature recognition results for a certain block within the same area are inconsistent across different low-altitude remote sensing images, the mulch film coverage status of that block is marked as pending; otherwise, the mulch film coverage status is determined based on the feature recognition results. The mulch film coverage status corresponding to the block is determined, thereby obtaining the mulch film coverage status of each block; the mulch film coverage status includes fully covered, uncovered, and undetermined coverage; based on the isomorphic correspondence between the mulch film coverage status of each block and each block type (wherein the correspondence is: blocks with fully covered mulch film correspond to covered areas, blocks with uncovered mulch film correspond to bare areas, and blocks with undetermined coverage correspond to covered fluctuating areas), each block type is determined, and each block type includes covered areas, bare areas, and covered fluctuating areas.

[0036] In this embodiment, it should be noted that farmland area refers to the entire land area, and block refers to a small part of farmland area. Therefore, farmland area includes several blocks, and the mulch film status in each block is the same.

[0037] The feature identification results are obtained by performing feature identification on the mulch film coverage status within the same block. The specific method is as follows: Spectral feature analysis of farmland areas is performed using drones to analyze its spectral reflectance characteristics and spatial texture features. Pixels within the block are analyzed and processed. Mulch film covered areas are characterized by high reflectance and low texture continuity in the visible or near-infrared bands, while bare soil or crop areas without mulch film exhibit discontinuous reflectance characteristics and obvious texture changes. By comprehensively judging the pixel reflectance intensity distribution, texture continuity, and regional connectivity within the block, when the proportion of pixels in the block that meet the mulch film coverage characteristics reaches the preset judgment condition, the block is determined to be covered by mulch film; otherwise, it is determined to be uncovered by mulch film. Thus, the feature identification results of mulch film coverage or uncovered mulch film are obtained.

[0038] In existing low-altitude remote sensing monitoring technology, farmland remote sensing monitoring images acquired from different observation trajectory points often have spatial translation, rotation, or scale differences due to factors such as changes in UAV attitude, slight fluctuations in flight altitude, and heading deviations. In order to achieve cross-time comparative analysis of the same farmland area, it is usually necessary to perform spatial alignment processing on remote sensing monitoring images acquired from different observation trajectory points.

[0039] Alignment of low-altitude remote sensing images of farmland areas at different observation trajectory points is achieved through the following method: Based on the positioning and attitude measurement systems onboard UAVs, spatial consistency processing is performed on low-altitude remote sensing images acquired at different times. This process includes three consecutive steps: initial spatial consistency processing, establishment of cross-trajectory block correspondence, and geometric correction. First, initial spatial consistency processing is performed on the remote sensing images based on the position and attitude information synchronously recorded by the UAV during remote sensing. The position information includes longitude, latitude, and flight altitude, while the attitude information includes heading angle, pitch angle, and roll angle. Based on the position information, the remote sensing images are mapped to their corresponding locations in geographic space. By combining the attitude and position information, the remote sensing images acquired at different observation trajectory points are uniformly mapped to the same geographic coordinate reference system, eliminating the overall offset caused by differences in flight paths and attitude changes. Subsequently, based on the initial spatial homogenization process, image elements with spatial stability are extracted within the farmland area, including plot boundaries, field ridge orientation, road outlines, or fixed surface texture structures. These image elements are then matched in remote sensing images from different observation trajectory points to determine the correspondence between the same farmland area and the block level (different blocks corresponding to the same farmland area are considered as a group of correspondences). Geometric correction processing is then performed on the remote sensing images. By performing local translation, rotation, or scale fine-tuning within the block area, local deviations that still exist after the initial spatial homogenization process are corrected. This ensures that remote sensing images acquired from different observation trajectory points are aligned at the block level within a unified spatial reference frame, providing a consistent spatial basis for subsequent comparison and judgment of mulch film coverage status.

[0040] In farmland plastic film residue monitoring, the distribution of plastic film on the farmland surface typically exhibits significant spatial heterogeneity, with substantial differences in film coverage continuity, damage levels, and exposure conditions across different locations. If the farmland is not divided into blocks and the entire farmland is treated as a homogeneous area for uniform sampling and testing, the sampling results are easily dominated by areas with high coverage or high exposure, thus reducing the ability of the plastic film residue monitoring results to accurately represent the true state of the farmland. By performing block-level analysis of farmland areas based on multi-time-lapse low-altitude remote sensing images and dividing the blocks into covered areas, exposed areas, and areas with fluctuating coverage, it is possible to effectively distinguish between blocks with relatively stable film coverage and those with frequently changing conditions. This also provides a basis for subsequently adopting differentiated sampling density configurations and detection strategies for different block types. In particular, areas with fluctuating coverage, where the film coverage fluctuates at different times, can be given greater attention in subsequent sampling and testing by being separately identified, preventing these blocks from being overlooked due to their unstable state.

[0041] Furthermore, the remote sensing sampling frequency adjustment is performed, including: based on the analysis of each block type, when a block type of a certain block is a coverage fluctuation area, the corresponding area of ​​the coverage fluctuation area in each low-altitude remote sensing monitoring image is obtained and recorded as the area of ​​each coverage fluctuation area; the block fluctuation area is determined based on the analysis of each coverage fluctuation area area; a comprehensive analysis is performed based on the area of ​​each coverage fluctuation area and the block fluctuation area to obtain the comprehensive influence value of the coverage area area; a graded response processing is performed based on the comprehensive influence value of the coverage area area to obtain the first constraint factor of remote sensing observation; the current remote sensing sampling frequency is increased based on the first constraint factor of remote sensing observation (the specific increase processing method is: add the product of the current remote sensing sampling frequency and the first constraint factor of remote sensing observation to obtain the execution value of the remote sensing sampling frequency); the execution value of the remote sensing sampling frequency is obtained, and the remote sensing sampling frequency update configuration is performed to complete the remote sensing sampling frequency adjustment; when there is no coverage fluctuation area, the current remote sensing sampling frequency is maintained.

[0042] In this embodiment, a comprehensive analysis is performed based on the area of ​​each coverage fluctuation zone and the area of ​​block fluctuation to obtain the comprehensive impact value of the coverage area. The specific method is as follows: the area of ​​each coverage fluctuation zone is averaged to obtain the average area of ​​the coverage fluctuation zone; the average area of ​​the coverage fluctuation zone is divided by the sample value of the area of ​​the coverage fluctuation zone (a pre-set reference value) to obtain the processed value of the area of ​​the coverage fluctuation zone; the area of ​​block fluctuation is divided by the sample value of the area of ​​block fluctuation (a pre-set reference value) to obtain the processed value of the area of ​​block fluctuation; the processed values ​​of the area of ​​coverage fluctuation zone and the processed values ​​of the area of ​​block fluctuation are multiplied by their respective weights (including the weight of the area of ​​coverage fluctuation zone and the weight of the area of ​​block fluctuation, wherein the weight of the area of ​​coverage fluctuation zone and the weight of the area of ​​block fluctuation are added together to obtain the comprehensive impact value of the coverage area. The area weights of the cover fluctuation zone and the block fluctuation zone can be obtained through hierarchical response and mapping matching. The specific method is as follows: Obtain the pre-stored cover area weight mapping rule table, which is obtained by fusing historical sample data (historical sample data is the statistical data of the historical plastic film residue concentration ratio of the sampling points corresponding to the cover fluctuation zone and the block fluctuation zone when conducting historical plastic film residue analysis, where the higher the historical plastic film residue concentration, the higher the weight of the corresponding area). By comparing the current cover fluctuation zone area processing value and the block fluctuation zone area processing value with the historical sample data in the corresponding cover area weight mapping rule table, the historical sample data that is closest to the current cover fluctuation zone area processing value or the block fluctuation zone area processing value is selected as the historical control sample data, and a set of weights corresponding to the historical control sample data is used as the cover fluctuation zone area weight and the block fluctuation zone area weight. If the historical control sample data is not unique, the weights obtained after averaging are used as the cover fluctuation zone area weight and the block fluctuation zone area weight.

[0043] A hierarchical response processing method is used based on the comprehensive impact value of the covered area to obtain the first constraint factor for remote sensing observation. Specifically, the method involves acquiring pre-stored gradient intervals of the comprehensive impact of each covered area and comparing them with the comprehensive impact value of the covered area. If the comprehensive impact value of the covered area falls within a certain gradient interval, the first constraint reference factor for remote sensing observation corresponding to that gradient interval is obtained as the first constraint factor for remote sensing observation. This first constraint reference factor is derived from historical low-altitude remote sensing monitoring data and historical data on the evolution of residual plastic film. Specifically, during historical monitoring, for farmland areas, the covered area, the area of ​​the covered area fluctuation zone, and their changes over time are continuously recorded, and corresponding plastic film residue detection results are collected simultaneously. Statistical analysis is performed on the relationship between the historical comprehensive impact value of the covered area and the rate of plastic film damage expansion, the magnitude of residue changes, and the effectiveness of subsequent sampling. This evaluates the actual gain effect of increasing the remote sensing sampling frequency on capturing changes in plastic film status under different covered area change intensities, thereby forming the first constraint reference factor for remote sensing observation used to regulate remote sensing observation behavior. The comprehensive impact value of the covered area is divided into multiple gradient intervals instead of being analyzed linearly because when the comprehensive impact value of the covered area is within the same gradient interval, the fluctuation intensity, damage evolution trend, and spatial variation characteristics of the mulch film coverage status within that area are generally similar, and the differences in remote sensing observation needs are not significant. A unified first constraint reference factor for remote sensing observation can meet the monitoring accuracy requirements. However, when the comprehensive impact value of the covered area spans different gradient intervals, the severity of mulch film coverage changes and the information acquisition needs differ significantly, requiring corresponding adjustments to the observation intensity. This achieves stability in the adjustment of the remote sensing sampling frequency.

[0044] By statistically analyzing the spatial area of ​​coverage fluctuation zones in low-altitude remote sensing images at different times, the stability of the mulch film coverage status in a given area over time can be reflected. When the area of ​​a coverage fluctuation zone within a given area is large, it indicates that the area exhibits frequent changes in mulch film coverage over consecutive observation periods, typically corresponding to localized damage, tearing, or displacement of residual film. For example, after mulch film damage, the exposed position and coverage pattern of some remaining film can change rapidly due to wind, tillage disturbance, or surface micro-airflows. This results in the same area appearing as alternating "covered-exposed" states in different remote sensing images, thus being identified as a coverage fluctuation zone.

[0045] By statistically analyzing the spatial area of ​​mulch cover fluctuation zones in low-altitude remote sensing images at different times, the stability of mulch cover status in the corresponding area over time can be objectively reflected. This allows for the identification of key areas where mulch cover status frequently changes due to damage, tearing, overturning, or disturbance caused by wind and cultivation. Compared to covered or bare areas, the mulch cover status within mulch cover fluctuation zones exhibits significant time-varying and uncertainties. If a fixed sampling frequency is used, insufficient observation time may lead to the omission of details regarding mulch cover status changes, thus affecting the representativeness of subsequent sampling point deployment. By dynamically increasing the remote sensing sampling frequency when mulch cover fluctuation zones are detected, more dense observation data can be obtained within a shorter time scale, enhancing the ability to characterize the mulch cover evolution process in such areas. This allows for a more complete recording and identification of mulch cover status changes, facilitating the rational deployment of subsequent sampling points.

[0046] Furthermore, the block fluctuation area is determined by the following method: obtaining the area of ​​each coverage fluctuation area and the corresponding shooting tilt angle, performing angle correction processing on each coverage fluctuation area based on the shooting tilt angle to obtain the corrected coverage fluctuation area; performing range processing on each coverage fluctuation area (subtracting the minimum value from the maximum value of each coverage fluctuation area to obtain the maximum range of coverage fluctuation area), and using the maximum range of coverage fluctuation area as the block fluctuation area.

[0047] In this embodiment, the area of ​​each coverage fluctuation zone can be obtained through the attitude information acquisition system of the low-altitude remote sensing carrier. Specifically, within the same block, the coverage status of the remote sensing monitoring images at continuous observation times is determined, the pixels determined to be coverage fluctuation zones are extracted, and the spatially adjacent pixels with consistent change characteristics are aggregated to obtain the mulch film coverage status and form coverage fluctuation zones. Based on the spatial resolution parameters of the image (including the actual distance between the center points of adjacent pixels on the ground, the length scale of a single pixel under the ground projection, and the ground coverage area corresponding to a single pixel), the number of pixels in the coverage fluctuation zone is converted into spatial area, thereby obtaining the area of ​​each coverage fluctuation zone.

[0048] The shooting tilt angle can be obtained through the attitude information acquisition system of the low-altitude remote sensing carrier. Specifically, while the low-altitude remote sensing platform is performing image acquisition, attitude parameter data is collected simultaneously. The attitude parameters include at least pitch angle, roll angle, and heading angle. The angle between the optical axis of the remote sensing device and the ground normal is used as the shooting tilt angle. The shooting tilt angle of each remote sensing monitoring image is bound and stored with the corresponding coverage fluctuation area to form a correspondence of "coverage fluctuation area - shooting tilt angle".

[0049] Angle correction processing is performed as follows: Using orthophoto imaging as the baseline imaging condition, the shooting tilt angle is used as a constraint parameter for area correction. Based on geometric projection relationships, tilt projection compensation is applied to the area of ​​each coverage fluctuation zone, restoring it from the image plane projection area to the equivalent ground area. When the shooting tilt angle increases, an angle correction coefficient (which is the shooting tilt angle itself) is introduced to scale and correct the original coverage fluctuation zone area, thereby eliminating the area magnification or compression effect caused by tilted shooting. The corrected coverage fluctuation zone area ensures that the fluctuation area of ​​the block reflects the actual change in the mulch film coverage state, rather than a spurious change caused by differences in imaging angle.

[0050] By introducing the shooting tilt angle during the determination of the area of ​​the coverage fluctuation zone and performing angle correction processing on the area of ​​the coverage fluctuation zone, the interference of low-altitude remote sensing imaging attitude changes on the area statistics results is effectively eliminated, enabling the area of ​​the coverage fluctuation zone to truly reflect the variation range of the mulch film coverage status over time. Based on this, by performing range processing on each corrected area of ​​coverage fluctuation zone, the degree of coverage change at different observation times within the block is centrally characterized, resulting in a representative area of ​​the block's fluctuation, thereby avoiding the amplified impact of single-moment anomalies or occasional disturbances on the judgment results.

[0051] Furthermore, to obtain a remote sensing image of the entire farmland area, the specific method is as follows: statistically analyze the low-altitude remote sensing monitoring images before and after the remote sensing sampling frequency adjustment, and jointly label them as each low-altitude remote sensing monitoring image; stitch the low-altitude remote sensing monitoring images together, and when a certain low-altitude remote sensing monitoring image contains a coverage fluctuation area, select the coverage fluctuation area corresponding to the smallest coverage fluctuation area as the reference coverage fluctuation area, and then stitch this reference coverage fluctuation area with the other low-altitude remote sensing monitoring images to obtain a remote sensing image of the entire farmland area; when there is no coverage fluctuation area, directly stitch the low-altitude remote sensing monitoring images to obtain a remote sensing image of the entire farmland area.

[0052] In this embodiment, the low-altitude remote sensing images are stitched together. The specific method is as follows: after completing the spatial registration and block consistency correction of each low-altitude remote sensing image, the low-altitude remote sensing images are stitched together based on a unified geographic coordinate reference system. Specifically, according to the coverage range of each low-altitude remote sensing image in geographic space, the overlapping area between adjacent images is determined. Within the overlapping area, block boundaries and stable surface textures are used as stitching constraints to perform translation and micro-scale rotation correction on the images, so that adjacent images achieve spatial continuity within the overlapping area. When there are multiple overlapping areas, the images are fused sequentially according to the coverage order, and finally all low-altitude remote sensing images are stitched together into a remote sensing image of the entire farmland area, thereby ensuring spatial continuity and coverage integrity.

[0053] The coverage fluctuation area corresponding to the minimum coverage fluctuation area is selected as the baseline coverage fluctuation area because: the area of ​​the coverage fluctuation area reflects the spatial range of changes in the mulch film coverage status of the same block in remote sensing observations at different times. The minimum coverage fluctuation area corresponds to the situation where the mulch film still exists (it may only exist partially or completely), but appears as locally exposed at individual observation times only due to wind disturbance, rolling, or instantaneous displacement. Therefore, selecting the coverage fluctuation area corresponding to the minimum coverage fluctuation area as the baseline coverage fluctuation area for stitching can prioritize the preservation of the most complete spatial representation of mulch film coverage information and the least interference from exposure during image fusion. This avoids misjudging short-term exposure or transient occlusion as long-term residual film loss due to selecting images with larger coverage fluctuation areas, thereby reducing the risk of mulch film being mistakenly missed and improving the authenticity and completeness of the spatial distribution representation of mulch film in remote sensing images of farmland.

[0054] Further, determining the sampling density execution quota includes the following steps: obtaining the remote sensing coverage integrity of the farmland area and matching it to obtain the initial sampling density quota; obtaining the mulch film type, determining the corresponding sampling density quota rule based on the mulch film type, adjusting the sampling density quota, and determining the sampling density execution quota; the mulch film type includes non-degradable mulch film and degradable mulch film; the sampling density quota rule includes non-degradable sampling density quota rule and degradable sampling density quota rule.

[0055] Determining the sampling density execution quota includes the following steps: obtaining the initial sampling density quota for the farmland area; obtaining the type of mulch film, determining the corresponding sampling density quota rule based on the mulch film type (adjusting the sampling density quota according to different rules for different mulch film types, the purpose of which is to better adapt to the mulch film type to obtain the sampling density quota), adjusting the sampling density quota, and determining the sampling density execution quota; mulch film types include non-degradable mulch film and degradable mulch film; sampling density quota rules include non-degradable sampling density quota rules and degradable sampling density quota rules.

[0056] In this embodiment, as Figure 3 As shown, Figure 3The flowchart below illustrates the collaborative processing of remote sensing sampling frequency adjustment and full-area image construction in the multi-source data fusion-based method for monitoring residual plastic film provided in this application. The process involves acquiring remote sensing images and extracting plastic film coverage features from these images; calculating the remote sensing coverage integrity based on these features and comparing it with a preset coverage integrity threshold; increasing the remote sensing sampling frequency and re-acquiring remote sensing images to update the plastic film coverage features when the remote sensing coverage integrity meets the preset requirements; determining an effective remote sensing image set and generating a full-area remote sensing image of farmland based on this set; and determining a sampling density quota and deploying plastic film residue sampling points accordingly.

[0057] It should be noted that adjusting the sampling density quota and determining the sampling density execution quota involves analyzing non-degradable mulch film based on remote sensing observation data from farmland-wide remote sensing images. The specific method is as follows: Obtain the mulch film thickness and perform correlation analysis to obtain the tensile integrity coefficient, which characterizes the risk level of mulch film breakage during mechanical recycling; obtain cover evolution impact parameters and analyze them to obtain the cover evolution impact index, which characterizes the degree of mulch film aging. Cover evolution impact parameters include mulch film burial depth, mulch film coverage duration, and maximum temperature difference within the coverage period; compare the cover evolution impact index with its threshold and combine it with the tensile integrity coefficient to obtain a recycling impact factor set (specifically, when the cover evolution impact index is above the threshold, a second recycling impact constraint factor is obtained based on the cover evolution impact index matching; at this time, the recycling impact factor set is specifically the first recycling impact factor). The set of recycling impact factors includes a first constraint factor and a second constraint factor for recycling impact. When the cover evolution impact index is less than the cover evolution impact index threshold, the recycling impact factor set specifically refers to the first constraint factor for recycling impact. Based on the recycling impact factor set, the preset initial sampling density quota is corrected to obtain the sampling density execution quota for non-degradable mulch film. (Specifically, if the recycling impact factor set consists of the first constraint factor and the second constraint factor for recycling impact, the initial sampling density quota is multiplied by the first constraint factor and the second constraint factor for recycling impact to obtain the corrected sampling density quota, i.e., the sampling density execution quota for non-degradable mulch film. If the recycling impact factor set consists of the first constraint factor for recycling impact, the initial sampling density quota is multiplied by the first constraint factor for recycling impact to obtain the corrected sampling density quota, i.e., the sampling density execution quota for non-degradable mulch film.) The recycling impact factor set specifically includes either the first constraint factor for recycling impact or the second constraint factor for recycling impact.

[0058] The recovery influence first constraint factor is obtained by acquiring the preset gradient intervals of each tensile integrity coefficient and the corresponding recovery influence first constraint reference factor, and comparing them with the current tensile integrity coefficient. If the tensile integrity coefficient is within a preset tensile integrity coefficient gradient interval, the recovery influence first constraint reference factor corresponding to that tensile integrity coefficient gradient interval is acquired as the recovery influence first constraint factor.

[0059] The method for obtaining the second constraint factor of the recovery influence is as follows: obtain the preset gradient interval of each coverage evolution influence index and the second constraint reference factor of the recovery influence corresponding to each coverage evolution influence index interval, and compare it with the current coverage evolution influence index. If the coverage evolution influence index is within a preset gradient interval of the coverage evolution influence index, then obtain the second constraint reference factor of the recovery influence corresponding to the gradient interval of the coverage evolution influence index as the second constraint factor of the recovery influence.

[0060] The tensile integrity coefficient, which characterizes the risk level of mulch film breakage during mechanical recycling, is obtained based on the thickness of the mulch film. The specific method is as follows: The nominal or measured thickness of the mulch film in the farmland area to be tested is obtained and used as an input feature. This thickness is then matched and analyzed with historical data on the mechanical recycling process of mulch film. Historical data includes records of breakage frequency, continuous breakage length, and complete recycling ratio for mulch films of different thicknesses under the same or similar recycling conditions (such as consistent recycling speed, traction method, and recycling machine structure). Subsequently, data samples whose difference between the mulch film thickness and the current mulch film thickness is within a preset thickness tolerance range are selected from the historical data. Based on the proportion of the length of the unbroken recycling section of the mulch film in the selected samples to the total recycling length, a statistical calculation is performed to obtain the historical tensile stability level for the corresponding thickness range. This historical tensile stability level is then normalized to obtain the tensile integrity coefficient, which characterizes the risk level of mulch film breakage during mechanical recycling. A larger tensile integrity coefficient indicates higher tensile stability and lower breakage risk for the mulch film of that thickness under the same recycling conditions.

[0061] The cover evolution impact index, used to characterize the degree of aging of plastic film, is obtained by the following method: A preset standard set of cover evolution impacts is obtained, which includes the calibration values ​​of plastic film burial depth (M0), plastic film coverage duration (F0), and maximum temperature difference within the coverage period (W0). The plastic film burial depth (M1), plastic film coverage duration (F1), and maximum temperature difference within the coverage period (W1) are divided by their corresponding calibration values ​​to obtain the treatment values ​​for each cover evolution impact. Each treatment value includes the treatment value of plastic film burial depth (M1 / M0), the treatment value of plastic film coverage duration (F1 / F0), and the treatment value of maximum temperature difference within the coverage period (W1 / W0). Each treatment value is multiplied by its corresponding cover evolution impact weight and then summed to obtain the cover evolution impact index. The weights of the influence of cover evolution include the weight of the soil burial depth of the mulch film (σ1), the weight of the duration of mulch film coverage (σ2), and the weight of the maximum temperature difference within the coverage period (σ3).

[0062] The weights for the soil burial depth of the plastic film, the duration of plastic film coverage, and the maximum temperature difference within the coverage period can be obtained from a database. For example, pre-stored historical monitoring data of farmland plastic film can be retrieved, which records the aging results of plastic film under different soil burial depths. Based on the historical soil burial depth range, the historical soil burial depth can be divided into multiple soil burial depth gradient intervals, and the variation of the degree of aging of the plastic film within each soil burial depth gradient interval can be statistically analyzed. Based on the relative contribution ratio of the degree of aging of the plastic film to the overall aging impact within each soil burial depth gradient interval, a pre-set weight can be assigned to each soil burial depth gradient interval. The corresponding weights for the mulch film burial depth are determined by classifying the mulch film burial depth of the current farmland area into its corresponding burial depth gradient interval. The mulch film burial depth weight corresponding to this gradient interval is then selected as the mulch film burial depth weight. The weights for the duration of mulch film coverage and the maximum temperature difference within the coverage period are obtained in the same way as the mulch film burial depth weight. They are all based on historical data of the corresponding parameters. First, a parameter gradient interval is formed, and then the corresponding weight is determined based on the contribution of each gradient interval to the degree of mulch film aging. Finally, the corresponding weight is selected according to the gradient interval in which the current parameter is located.

[0063] The sampling density quotas are set differently based on the type of mulch film. This is primarily because non-degradable and degradable mulch films differ significantly in their residual morphology, fragmentation patterns, and environmental responses. Non-degradable mulch films are more prone to fragmentation and concealment during long-term field exposure, exhibiting high spatial dispersion and significant local variations. Insufficient sampling density can easily lead to underestimation of residual amounts. In contrast, the morphological changes of degradable mulch films during degradation are highly correlated with time, temperature, and humidity conditions, resulting in staged and regional characteristics in their residual state. Matching sampling density quotas to different mulch film types helps to adjust sampling intensity in a targeted manner, ensuring that sampling point placement better aligns with different mulch film residue mechanisms, thereby improving the representativeness and accuracy of mulch film residue monitoring results.

[0064] Matching the initial sampling density quota based on the remote sensing coverage integrity of farmland areas can reflect the sufficiency of plastic film distribution and the level of potential residue risk in farmland on an overall scale. When the remote sensing coverage integrity is large (above the cover evolution influence index threshold), it indicates that the plastic film coverage or residue in farmland is extensive and continuous. Sparse sampling is difficult to accurately reflect the true residue status. Therefore, it is necessary to constrain the initial sampling density quota by recovering the second constraint factor to obtain a corrected sampling density quota. This allows for the spatial deployment of more sampling points to obtain more comprehensive and detailed information on plastic film residue. When the remote sensing coverage integrity is small (below the cover evolution influence index threshold), the corrected sampling density quota can be obtained without combining the recovery of the second constraint factor.

[0065] Further, the sampling point layout includes the following steps: Proportional processing is performed on the covered area, bare area, and covered fluctuation area of ​​the farmland region to obtain the area proportion of each block type, including the proportion of covered area, bare area, and covered fluctuation area; based on the area proportion of each block type, the sampling density execution quota is proportionally allocated (specifically, the area proportion of each block type is multiplied by the corresponding sampling density execution quota to obtain the sampling density execution quota for each area type), and the sampling density execution quota for each block type is obtained, thereby randomly distributing sampling points (specifically, randomly distributing sampling points within the corresponding block type).

[0066] In this embodiment, if a certain area type (such as no coverage fluctuation area) does not exist among the covered area, bare area, and covered fluctuation area of ​​farmland, then that type will not be proportionally processed; that is, only the other existing area types will be proportionally processed.

[0067] For example, if the area proportions of covered areas, exposed areas, and areas with fluctuating coverage are 80%, 15%, and 5% respectively, then when allocating the sampling density execution quota proportionally, the sampling density execution quota for each area type will be obtained by multiplying this proportion by the sampling density execution quota.

[0068] Furthermore, it also includes determining whether to issue an early warning based on the residual value of plastic film in farmland areas. The specific method is as follows: determine the corresponding plastic film early warning rule based on the type of plastic film in the farmland area (the plastic film early warning rule is used to determine whether an early warning is needed, and different plastic film types require different early warning rules). The corresponding plastic film early warning rules include early warning rules for non-degradable plastic film and early warning rules for degradable plastic film; compare and analyze the residual value of plastic film in the farmland area with the corresponding early warning rules to determine whether to issue an early warning. If the residual value of plastic film in the farmland area meets the corresponding early warning rules, an early warning is issued; otherwise, no early warning is issued.

[0069] In this embodiment, a comparative analysis is performed based on the residual value of mulch film in farmland areas and the corresponding mulch film early warning rules. Specifically, when the mulch film type is non-degradable mulch film, if the residual value of non-degradable mulch film is above the first threshold for mulch film residue, a mulch film residue early warning reminder is issued; otherwise, no mulch film residue early warning reminder is issued.

[0070] By distinguishing between non-degradable and degradable mulch films and setting corresponding early warning rules for each, early warning judgments can accurately reflect the essential differences in environmental behavior, residual forms, and risk evolution paths of different types of mulch films. Non-degradable mulch films are stable in farmland for a long time, with strong residue accumulation and high removal difficulty. Their environmental risks are mainly manifested in the continuous accumulation of residues and long-term impacts on the topsoil structure. Therefore, adopting stricter early warning rules with residue limit control as the core is conducive to early identification of high residue risk areas and triggering intervention measures. On the other hand, degradable mulch films undergo a gradual decomposition process under natural conditions, and their residual state is significantly affected by time, climate, and surface environment. The phased fluctuations in residue values ​​do not necessarily correspond to long-term risks. Therefore, by setting early warning rules that match their degradation characteristics, unnecessary false alarms caused by short-term residue changes can be avoided.

[0071] like Figure 4 As shown, Figure 4This is a schematic diagram of the structure of the mulch film residue monitoring system based on multi-source data fusion provided in this application embodiment. The mulch film residue monitoring system based on multi-source data fusion provided in this application embodiment includes: a block evaluation module, a data acquisition and adjustment module, and a residue monitoring module. The block evaluation module is used to monitor farmland areas at different observation trajectory points based on a UAV following a preset flight route, obtaining low-altitude remote sensing observation data and low-altitude remote sensing monitoring images, and analyzing the mulch film coverage status of each block to statistically determine the block type. The data acquisition and adjustment module is used to adjust the remote sensing sampling frequency based on the block type and / or low-altitude remote sensing observation data analysis until remote sensing monitoring is completed, obtaining a remote sensing image of the entire farmland area. The residue monitoring module is used to determine the sampling density and implement quotas based on the remote sensing observation data analysis of the farmland area remote sensing image, and to deploy sampling points, performing mulch film residue monitoring at each sampling point to obtain the mulch film residue value of the farmland area.

[0072] like Figure 5 , Figure 6 and Figure 7 As shown, Figure 5 This is a schematic diagram of the mulch film identification system based on multi-source data fusion provided in the embodiments of this application. Figure 6 This is a schematic diagram of the front of the same block of the mulch film residue monitoring system based on multi-source data fusion provided in the embodiments of this application. Figure 7 This is a schematic diagram of the rear of the same block in the multi-source data fusion-based monitoring system for residual plastic film provided in this application embodiment. It can identify plants, land and block types (such as bare areas), and can determine the same area of ​​the two images by analyzing the features of the two images.

[0073] The second embodiment, based on the first embodiment, performs spatial resolution adjustment. If a block is identified as a coverage fluctuation zone, spatial resolution adjustment is performed; otherwise, the current spatial resolution is maintained. Higher spatial resolution means more pixels per unit area, making details such as mulch film edges and fracture textures easier to identify. The specific method for increasing spatial resolution is as follows: the current spatial resolution is added to the product of the spatial resolution and the first constraint factor of remote sensing observation to obtain the adjusted spatial resolution, and then a spatial resolution update configuration is performed to complete the spatial resolution increase adjustment.

[0074] By adaptively enhancing the spatial resolution of remote sensing imagery, the number of effective pixels per unit area is significantly increased, enabling a more refined characterization of microstructural features such as mulch film edge morphology, damage cracks, rolled-up wrinkles, and exposed patches. Compared to fixed-resolution imaging, this allows for a shift from coarse-grained assessment to fine-grained identification of mulch film residue. High spatial resolution imagery enhances the discernibility of structural differences within areas of cover fluctuation, thereby improving the spatial accuracy and interpretability of mulch film residue monitoring results. This facilitates refined monitoring and analysis of complex mulch film damage areas.

[0075] The third embodiment, based on the first embodiment, can also adjust the remote sensing sampling frequency in the following ways: acquire low-altitude remote sensing observation data for each block and analyze the remote sensing coverage integrity of each block. The low-altitude remote sensing observation data includes the coverage projection area, fracture feature length, and fracture extension width. Based on the remote sensing coverage integrity analysis of each block, if the remote sensing coverage integrity of a certain block is above a preset remote sensing coverage integrity threshold, maintain the current remote sensing sampling frequency. If the remote sensing coverage integrity of a certain block is less than the remote sensing coverage integrity threshold, perform hierarchical response processing based on the remote sensing coverage integrity to obtain a second constraint factor for remote sensing observation. Based on this second constraint factor, increase the current remote sensing sampling frequency to obtain an execution value for the remote sensing sampling frequency (the execution value is obtained by adding the product of the current remote sensing sampling frequency and the second constraint factor for remote sensing observation). Then, perform a remote sensing sampling frequency update configuration to complete the remote sensing sampling frequency adjustment. If the remote sensing coverage integrity of each block is above the remote sensing coverage integrity threshold, maintain the current remote sensing sampling frequency.

[0076] The remote sensing coverage integrity of each block is obtained as follows: A preset low-altitude remote sensing observation data set is acquired, which includes sample values ​​of the coverage projection area (S0), the fault feature length limit value (A0), and the fault extension width limit value (B0). The coverage projection area (S1), fault feature length (A1), and fault extension width (B1) are divided by their respective low-altitude remote sensing observation data sets to obtain processed low-altitude remote sensing observation values. Each processed value includes the coverage projection area processed value (S1 / S0), the fault feature length processed value (A1 / A0), and the fault extension width processed value (B1 / B0). Each processed value is then factored into its corresponding low-altitude remote sensing observation weight (including the coverage projection area weight, fault feature length weight, and fault extension width weight), multiplied, and summed to obtain the remote sensing coverage integrity. The remote sensing coverage integrity of each block is obtained similarly.

[0077] It should be noted that the coverage projection area represents the coverage area of ​​the mulch film (which can be obtained by obtaining the number of pixels and then converting it into area), and the fracture feature length represents the number of pixels where the mulch film is not covered (by connecting the two pixels with the farthest uncovered state of the mulch film, obtaining the pixel line, counting the number of pixels on the line, and then converting the number of pixels into length, the fracture feature length can be obtained). The fracture propagation width is specifically obtained by drawing several perpendicular lines from the line connecting the pixels, obtaining each perpendicular line, and selecting the maximum length among the perpendicular lines as the fracture propagation width.

[0078] The weights for low-altitude remote sensing observations, including the weight of the coverage projection area, the weight of the fault feature length, and the weight of the fault extension width, can be obtained from a database. The specific method is as follows: retrieve pre-stored historical low-altitude remote sensing monitoring data, which records historical monitoring data under different coverage projection area conditions; divide the historical coverage projection area into multiple coverage projection area gradient intervals based on the range of historical coverage projection area values, and statistically analyze the variation in the accuracy of identifying residual plastic film within each gradient interval; pre-set a corresponding coverage projection area weight for each gradient interval based on the relative contribution ratio of each gradient interval to the effectiveness of low-altitude remote sensing observations; determine the interval affiliation of the current low-altitude remote sensing observation's coverage projection area with the gradient interval, identify the corresponding gradient interval, and select the coverage projection area weight corresponding to that gradient interval as the coverage projection area weight in the low-altitude remote sensing observations. The methods for obtaining the weights of fracture feature length and fracture propagation width are the same as those for obtaining the weights of the covered projection area. Both are based on historical remote sensing monitoring data of the corresponding parameters. First, a parameter gradient interval is formed, and then the corresponding weights are determined according to the contribution of each gradient interval to the identification of residual plastic film and the interpretation of fractures. The corresponding weights are selected according to the gradient interval in which the current parameter is located. It should be noted that the weights of fracture feature length and fracture propagation width are negative numbers.

[0079] A hierarchical response processing based on remote sensing coverage integrity is used to obtain the second constraint factor for remote sensing observation. Specifically, the method involves acquiring pre-stored gradient intervals for each remote sensing coverage integrity level and the corresponding second constraint reference factor. The remote sensing coverage integrity is compared with each gradient interval. If the remote sensing coverage integrity falls within a preset gradient interval, the corresponding second constraint reference factor is obtained as the second constraint factor. It should be noted that the second constraint reference factor originates from historical remote sensing monitoring data and historical measured data of residual plastic film. Specifically, during historical monitoring, the remote sensing coverage integrity of different farmland areas is recorded over a long period, and the plastic film residue detection results for the corresponding time periods are acquired simultaneously. Statistical analysis of historical remote sensing coverage integrity and historical residue changes is conducted to summarize the actual impact of increasing the remote sensing sampling frequency on residue monitoring effectiveness at different coverage integrity levels, thereby forming the second constraint reference factor for adjusting observation behavior. Dividing remote sensing coverage integrity into multiple gradient intervals is beneficial because regions within the same gradient interval exhibit relatively small differences in mulch film stability, damage expansion trends, and residual variation magnitudes, resulting in consistent overall observation requirements. Therefore, uniformly configuring a single second constraint reference factor for remote sensing observation within the same gradient interval can accurately reflect the observation adjustment intensity requirements of that interval, ensuring the objectivity and reproducibility of parameter sources while avoiding frequent adjustments and data redundancy caused by excessive subdivision.

[0080] The remote sensing coverage integrity is obtained by analyzing the coverage projection area, fracture feature length, and fracture propagation width. This is because the interaction between these parameters is taken into account. For example, the coverage projection area is used to characterize the overall coverage level of the mulch film in the block; the fracture feature length reflects the degree of spatial continuity disruption after the mulch film is damaged within the block, and the larger the value, the worse the stability of the mulch film coverage; the fracture propagation width reflects the degree of erosion of the effective coverage area of ​​the mulch film by the damaged area, and an increase in its value will further weaken the positive effect of the coverage projection area on the integrity. Therefore, under the combined effect of the three, even if the coverage projection area is large, if the fracture feature length and fracture propagation width increase significantly, it will still lead to a decrease in the remote sensing coverage integrity.

[0081] The fourth embodiment, based on the first and third embodiments, performs remote sensing sampling frequency adjustment, which can also be achieved through the following steps: When a certain block has a coverage fluctuation area and the remote sensing coverage integrity of a certain block is less than the remote sensing coverage integrity threshold, the current remote sensing sampling frequency is increased to obtain an execution value for the remote sensing sampling frequency, and the remote sensing sampling frequency update configuration is executed to complete the remote sensing sampling frequency adjustment; otherwise, the current remote sensing sampling frequency is maintained. The specific method for obtaining the execution value for the remote sensing sampling frequency is as follows: Based on the first constraint factor and the second constraint factor of remote sensing observation, the current remote sensing sampling frequency is increased to obtain the execution value for the remote sensing sampling frequency (the product of the current remote sensing sampling frequency and the first constraint factor of remote sensing observation is added to obtain the first execution value of the remote sensing sampling frequency; the product of the first execution value of the remote sensing sampling frequency and the second constraint factor of remote sensing observation is added to obtain the execution value of the remote sensing sampling frequency).

[0082] In the fifth embodiment, based on the first embodiment, the biodegradable mulch film is analyzed based on remote sensing observation data of the whole farmland remote sensing image to determine the sampling density execution quota. The specific method is to use the initial sampling density quota as the sampling density execution quota of the biodegradable mulch film.

[0083] Sampling points are deployed based on the sampling density quota of biodegradable mulch film. The specific method is as follows: the area of ​​the covered area, the area of ​​the bare area, and the area of ​​the covered area fluctuation area of ​​the farmland are processed proportionally to obtain the area proportion of each block type; the sampling density quota is proportionally allocated based on the area proportion of each block type to obtain the sampling density quota of each block type, and then the sampling points are randomly deployed.

[0084] When the mulch film is biodegradable, a comparative analysis is performed based on the residual mulch film value in the farmland area and the corresponding mulch film warning rules. The specific method is as follows: obtain the current surface performance response value of the farmland area, analyze and obtain the mulch film residue threshold adjustment coefficient, and adjust the preset mulch film residue threshold based on the mulch film residue threshold adjustment coefficient (specifically: add the current mulch film residue threshold to the product of the mulch film residue threshold and the mulch film residue threshold adjustment coefficient to obtain the adjusted mulch film residue threshold, i.e., the second mulch film residue threshold). If the residual mulch film value of the biodegradable mulch film is above the second mulch film residue threshold, a mulch film residue warning reminder is issued; otherwise, no mulch film residue warning reminder is issued.

[0085] The adjustment coefficient for residual plastic film is obtained by matching the surface performance response value of the current farmland area. The specific method is as follows: obtain the preset gradient interval of surface performance response value and the reference coefficient for adjustment of residual plastic film corresponding to the gradient interval of surface performance response value, and compare it with the surface performance response value. If the surface performance response value is within a preset gradient interval of surface performance response value, then the reference coefficient for adjustment of residual plastic film corresponding to the gradient interval of surface performance response value is used as the adjustment coefficient for residual plastic film.

[0086] Random sampling point deployment includes: deploying sampling points based on a quota for the sampling density of the biodegradable mulch film. Specifically, this involves: proportionally processing the area of ​​the covered area, the area of ​​the bare area, and the area of ​​the area with fluctuating coverage in the farmland region to obtain the area proportion of each block type. This area proportion includes the proportion of the covered area, the area of ​​the bare area, and the area of ​​the area with fluctuating coverage. Based on the area proportion of the bare area, a comparison is made with a preset threshold. If the area proportion of the bare area is above the threshold, then the sampling density quota of the biodegradable mulch film is proportionally allocated based on the area proportion of each block type, thus obtaining the sampling density quota for each block type, and randomizing the sampling points accordingly. If the area proportion of the bare area is less than the threshold, then based on the area proportion of the farmland region... The remote sensing coverage integrity of the field area is matched to obtain the division scale. Based on the division scale (division length and division width), each block type is further divided into blocks (the previously divided blocks are further subdivided without re-dividing the original blocks). This yields each residual assessment unit (specifically, each block type is divided into blocks according to the division scale, thus dividing each block into smaller blocks, which are the residual assessment units). The land surface performance parameters (including average land surface temperature and land surface area) of each residual assessment unit are obtained, and the land surface performance response value of each residual assessment unit is analyzed. The sampling density quota for each residual assessment unit is then determined, and sampling points are randomly deployed.

[0087] The average surface temperature can be obtained by detecting infrared temperature sensors, using the temperature of the set and center point of the residual assessment units as the average surface temperature. The surface area is obtained by analyzing the number of pixels in the residual assessment units and then performing area conversion.

[0088] The classification scale is obtained by matching the remote sensing coverage integrity of farmland areas. The specific method is as follows: retrieve pre-stored historical remote sensing data of farmland, which records the historical remote sensing coverage integrity and the corresponding farmland spatial classification scale; divide the historical remote sensing coverage integrity into intervals according to the coverage integrity value range to form multiple coverage integrity gradient intervals, and pre-associate a type of farmland spatial classification scale for each coverage integrity gradient interval; determine the interval affiliation between the current farmland area's remote sensing coverage integrity and the coverage integrity gradient interval, determine the coverage integrity gradient interval to which it belongs, and select the farmland spatial classification scale corresponding to the coverage integrity gradient interval as the classification scale of the current farmland area.

[0089] The surface performance response values ​​for each residue assessment unit are obtained as follows: A surface performance control set is acquired, including a surface mean temperature control value (DT0) and a surface area control value (DS0). The surface mean temperature (DT1) and surface area (DS1) are divided by the surface performance control set values ​​to obtain the surface performance treatment values ​​for each unit. Each surface performance treatment value includes a surface mean temperature treatment value (DT1 / DT0) and a surface area treatment value (DS1 / DS0). Each surface performance treatment value is then multiplied by its corresponding surface performance weight and summed to obtain the surface performance response value. This yields the surface performance response value for each residue assessment unit, where the surface performance weights include a surface mean temperature weight (τ1) and a surface area weight (τ2).

[0090] The weights for average surface temperature and surface area can be obtained from a database. For example, pre-stored historical residual assessment data can be retrieved, which includes the results of changes in plastic film residue under different average surface temperature conditions. Based on the range of historical average surface temperature values, the historical average surface temperature is divided into multiple average surface temperature gradient intervals, and the statistical characteristics of plastic film residue levels within each average surface temperature gradient interval are statistically analyzed. Based on the differences in the degree of influence of each average surface temperature gradient interval on plastic film residue changes, the average surface temperature weight corresponding to each average surface temperature gradient interval is determined. The average surface temperature of the current residue assessment unit is compared with the average surface temperature gradient interval to determine the interval affiliation, and the average surface temperature weight corresponding to its gradient interval is selected as the average surface temperature weight of the residue assessment unit. The method for obtaining the surface area weight is the same as that for obtaining the average surface temperature weight. Both are based on the historical data of the corresponding parameters to divide the gradient intervals, and the influence contribution of each gradient interval on plastic film residue changes is predetermined. The current weight is directly selected from the gradient interval where the parameter is located.

[0091] When the proportion of exposed areas is high, allocating sampling density directly based on the area proportion of each block type is beneficial for quickly covering large-scale exposed areas and improving the overall understanding of the diffusion status of biodegradable mulch film residues. When the proportion of exposed areas is low, a further division scale is introduced based on the integrity of remote sensing coverage, refining the original blocks into residue assessment units. This improves spatial resolution without damaging the original block structure, allowing sampling density to be dynamically allocated based on more refined spatial units. By combining surface performance parameters such as average surface temperature and surface area of ​​the residue assessment units, a sampling density execution quota is constructed, enabling targeted deployment of sampling points in areas where the coverage status does not change significantly but potential residue risks exist. This enhances the ability to identify localized residues, uneven decomposition, and hidden residues of biodegradable mulch film.

[0092] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for monitoring residual plastic film based on multi-source data fusion, characterized in that, Includes the following steps: Based on the UAV monitoring the farmland area at different observation trajectory points according to the preset flight route, low-altitude remote sensing observation data and low-altitude remote sensing monitoring images are obtained, and the mulch film coverage status of each block is analyzed and the block types are statistically determined. The block types include covered areas, bare areas and covered fluctuating areas. The low-altitude remote sensing observation data is used to characterize the fracture characteristics of the mulch film. Based on the analysis of data from each block type and / or low-altitude remote sensing observation, the remote sensing sampling frequency is adjusted until remote sensing monitoring is completed and a remote sensing image of the entire farmland is obtained. Based on the remote sensing observation data analysis of the entire farmland remote sensing image, the sampling density quota is determined and sampling points are set up. The residual plastic film is monitored at each sampling point to obtain the residual value of plastic film in the farmland area. The residual value of plastic film is a statistical quantity of residual status formed based on the residual plastic film monitoring results of the sampling points.

2. The method for monitoring residual plastic film based on multi-source data fusion as described in claim 1, characterized in that: The specific method for obtaining each block type is as follows: Alignment analysis is performed on low-altitude remote sensing images of farmland at different observation trajectory points to identify the corresponding blocks in each low-altitude remote sensing image, and feature identification is performed on the mulch film coverage status within the same blocks to obtain feature identification results, which include mulch film coverage and mulch film uncovering. Based on the feature recognition results, a comparison is made. If the feature recognition results of a certain block within the same block are inconsistent in the low-altitude remote sensing monitoring images, the mulch film coverage status of that block is marked as an undetermined coverage status. Otherwise, the mulch film coverage status of that block is determined based on the feature recognition results, thereby obtaining the mulch film coverage status of each block. The mulch film covering status includes fully covered, uncovered, and pending coverage status. Based on the isomorphic correspondence between the mulch film coverage status of each block and the block type, the block types are determined, including covered areas, bare areas, and coverage fluctuation areas.

3. The method for monitoring residual plastic film based on multi-source data fusion as described in claim 1, characterized in that: The process of adjusting the remote sensing sampling frequency includes: Based on the analysis of each block type, when there is a block type that is a coverage fluctuation area, the corresponding area of ​​the coverage fluctuation area in each low-altitude remote sensing monitoring image is obtained and recorded as the area of ​​each coverage fluctuation area. The block fluctuation area is determined based on the analysis of the area of ​​each coverage fluctuation area. Based on the comprehensive analysis of the coverage fluctuation area and the block fluctuation area, the comprehensive impact value of the coverage area is obtained. Based on the comprehensive impact value of the coverage area, a graded response processing is performed to obtain the first constraint factor of remote sensing observation. Based on the first constraint factor of remote sensing observation, the current remote sensing sampling frequency is increased to obtain the execution value of the remote sensing sampling frequency. Then, the remote sensing sampling frequency update configuration is executed to complete the adjustment of the remote sensing sampling frequency. If there is no coverage fluctuation area, the current remote sensing sampling frequency will be maintained.

4. The method for monitoring residual plastic film based on multi-source data fusion as described in claim 3, characterized in that: The specific method for determining the fluctuation area of ​​the block is as follows: The area of ​​each coverage fluctuation zone and the shooting tilt angle corresponding to each coverage fluctuation zone area are obtained. An angle correction process is performed on the area of ​​each coverage fluctuation zone based on the shooting tilt angle to obtain the corrected area of ​​each coverage fluctuation zone. The area of ​​each coverage fluctuation zone is processed by range processing to obtain the maximum difference value of the coverage fluctuation zone area, which is then used as the block fluctuation area.

5. The method for monitoring residual plastic film based on multi-source data fusion as described in claim 1, characterized in that: The process of adjusting the remote sensing sampling frequency also includes: Low-altitude remote sensing observation data of each block is acquired and analyzed to obtain the remote sensing coverage integrity of each block. The low-altitude remote sensing observation data includes the coverage projection area, fracture feature length and fracture extension width. Based on the remote sensing coverage integrity analysis of each block, if the remote sensing coverage integrity of a certain block is above the preset remote sensing coverage integrity threshold, the current remote sensing sampling frequency will be maintained. When the remote sensing coverage integrity of a certain block is less than the remote sensing coverage integrity threshold, a hierarchical response processing is performed based on the remote sensing coverage integrity to obtain the second constraint factor of remote sensing observation. Based on the second constraint factor of remote sensing observation, the current remote sensing sampling frequency is increased to obtain the execution value of the remote sensing sampling frequency, and the remote sensing sampling frequency update configuration is executed to complete the adjustment of the remote sensing sampling frequency. When the remote sensing coverage integrity of each block is above the remote sensing coverage integrity threshold, the current remote sensing sampling frequency is maintained.

6. The method for monitoring residual plastic film based on multi-source data fusion as described in claim 1, characterized in that: The specific method for obtaining the remote sensing image of the entire farmland area is as follows: Statistical analysis of low-altitude remote sensing images before and after the adjustment of remote sensing sampling frequency, and joint labeling of them as individual low-altitude remote sensing images; The low-altitude remote sensing images are stitched together. When a low-altitude remote sensing image contains a coverage fluctuation area, the coverage fluctuation area corresponding to the smallest area of ​​the coverage fluctuation area is selected as the reference coverage fluctuation area. Then, the reference coverage fluctuation area is stitched together with the other low-altitude remote sensing images to obtain a remote sensing image of the entire farmland. When there is no coverage fluctuation area, the low-altitude remote sensing monitoring images are directly stitched together to obtain a remote sensing image of the entire farmland area.

7. The method for monitoring residual plastic film based on multi-source data fusion as described in claim 1, characterized in that: The determination of sampling density and quota execution includes the following steps: Obtain the remote sensing coverage completeness of farmland areas and perform matching to obtain the initial sampling density quota; Obtain the type of mulch film, determine the corresponding sampling density quota rule based on the mulch film type, adjust the sampling density quota, and determine the sampling density execution quota. The types of mulch film include non-degradable mulch film and degradable mulch film; The sampling density quota rules include non-degradable sampling density quota rules and degradable sampling density quota rules.

8. The method for monitoring residual plastic film based on multi-source data fusion as described in claim 7, characterized in that: The sampling point layout includes the following steps: The area ratios of covered areas, bare areas, and areas with fluctuating coverage in farmland regions are processed proportionally to obtain the area proportions of each block type, including the area proportions of covered areas, bare areas, and areas with fluctuating coverage. Based on the area proportion of each block type, the sampling density execution quota is proportionally allocated to obtain the sampling density execution quota for each block type, and sampling points are then randomly deployed.

9. The method for monitoring residual plastic film based on multi-source data fusion as described in claim 1, characterized in that: This also includes determining whether to issue an early warning based on the residual value of plastic film in farmland areas. The specific method is as follows: The corresponding early warning rules for mulch film are determined based on the type of mulch film in the farmland area. The corresponding early warning rules for mulch film include early warning rules for non-degradable mulch film and early warning rules for degradable mulch film. The system compares and analyzes the residual value of plastic film in farmland areas with the corresponding early warning rules to determine whether to issue an early warning. If the residual value of plastic film in farmland areas meets the corresponding early warning rules, an early warning is issued; otherwise, no early warning is issued.

10. A system for monitoring residual plastic film as described in any one of claims 1-9, characterized in that, include: Block evaluation module, data acquisition and adjustment module, and residual monitoring module; The block assessment module is used to monitor farmland areas at different observation trajectory points based on the UAV following a preset flight route, obtain low-altitude remote sensing observation data and low-altitude remote sensing monitoring images, analyze the mulch film coverage status of each block, and statistically determine the type of each block. The data acquisition and adjustment module is used to perform remote sensing sampling frequency adjustment based on the analysis of each block type and / or low-altitude remote sensing observation data, until remote sensing monitoring is completed and a remote sensing image of the entire farmland is obtained. The residual monitoring module is used to analyze remote sensing observation data based on remote sensing images of the entire farmland area, determine the sampling density quota, deploy sampling points, perform residual monitoring of plastic film at each sampling point, and obtain the residual value of plastic film in the farmland area.

Images