A power transmission line ring water protection monitoring method based on image fusion

Abstract

The present application belongs to the technical field of environmental protection monitoring of power transmission line, and relates to a kind of environmental protection monitoring method of power transmission line based on image fusion, which obtains image data of monitoring area and sensing data based on tower coordinate system by image acquisition equipment and tower sensor group respectively;According to the first monitoring time period, the first area image of the first index area corresponding to the environmental protection index is determined, wherein the image data is used to determine the plant state feature and the plant distribution feature, and the sensing data is the hydrological soil feature that causes the feature change;The second area image in the second monitoring time period is fused with the first area image to determine the image deviation and its corresponding plant state trend and plant distribution trend.The present application realizes the spatial coupling of image and sensing data through the tower coordinate system, can quantitatively determine the plant state and distribution trend, and improves the monitoring accuracy and efficiency of environmental protection.

Description

Technical Field

[0001] This invention relates to the field of environmental and water conservation monitoring technology for power transmission lines, specifically to a method for environmental and water conservation monitoring of power transmission and transformation lines based on image fusion. Background Technology

[0002] Transmission and transformation line projects are characterized by their wide geographical reach, complex terrain, and long construction periods, and their construction and operation may have multifaceted impacts on the ecological environment along the route. Against this backdrop, how to implement efficient and accurate environmental and water conservation monitoring during the construction phase of transmission and transformation projects has become a pressing technical problem for power grid companies.

[0003] Currently, environmental and water conservation monitoring for power transmission and transformation projects mainly relies on manual methods such as inspections, on-site monitoring, and field investigations. Environmental supervision methods include document review, inspections, on-site monitoring, witnessing, meetings, monitoring, and field investigations; water and soil conservation monitoring methods include stake observation, runoff plot observation, and stepped sediment collection meter measurement. These methods are primarily manual, consuming significant human and material resources, and lack process documentation. Supervision only controls a limited number of points in the process, failing to grasp the overall environmental impact throughout the construction phase. Furthermore, environmental protection acceptance work typically begins after the main project is completed, by which time the construction and supervision units have already left the site. Major environmental problems that arise during construction often cannot be fully restored, affecting the acceptance of the main project.

[0004] In recent years, some studies have attempted to introduce technologies such as satellite remote sensing and drones into the field of environmental and water conservation monitoring. However, the aforementioned existing technologies still have the following shortcomings: First, remote sensing image data and ground sensor data are usually treated as parallel data sources and simply superimposed. This fails to establish a causal hierarchical framework of "image data representing plant status / distribution - sensor data representing driving causes," making it difficult to accurately distinguish whether the driving factors of vegetation change are natural growth, soil erosion, or human destruction, resulting in distorted analysis of the causes of environmental and water conservation indicators.

[0005] Secondly, the use of GPS latitude and longitude coordinates or geodetic coordinate systems for spatial registration fails to fully utilize the advantages of power transmission and transformation line towers as inherent spatial nodes to construct a dedicated coordinate system, making it difficult to achieve precise spatial coupling between above-ground vegetation distribution and groundwater hydrological characteristics.

[0006] Third, existing methods, such as satellite remote sensing, drones, and on-site monitoring, operate independently, lacking a progressive and coordinated regulatory mechanism that moves from general to specific.

[0007] Fourth, existing methods mainly achieve change detection, that is, to determine whether vegetation has changed, but it is difficult to further determine the direction and trend of the change, and cannot provide sufficient decision-making basis for the formulation of environmental protection and water conservation measures.

[0008] In summary, existing monitoring technologies for environmental protection and water conservation in power transmission and transformation projects mainly suffer from the following problems: the lack of a causal relationship framework between data sources makes it difficult to explain the driving factors of vegetation change; the lack of a dedicated spatial benchmark system for power transmission and transformation projects makes it difficult to achieve precise coupling between vegetation distribution and hydrological characteristics; the lack of a progressive linkage mechanism among various monitoring methods results in low monitoring efficiency; and the monitoring results can only reflect whether there is a change, making it difficult to quantify the trend direction. Summary of the Invention

[0009] This invention provides a method for monitoring environmental protection and water conservation in power transmission and transformation lines based on image fusion, aiming to solve the technical problems in existing monitoring of environmental protection and water conservation in power transmission and transformation projects, such as the lack of causal correlation of data sources, the lack of dedicated benchmarks for spatial registration, the lack of progressive linkage of monitoring methods, and the lack of trend quantification of monitoring results.

[0010] A method for monitoring the environmental protection and water conservation of power transmission and transformation lines based on image fusion is applied to a monitoring system for the environmental protection and water conservation of power transmission and transformation lines, the system comprising: The image acquisition device is configured to connect to the edge computing terminal via a wireless data transmission link, and is used to acquire multispectral image data of the monitoring area during the monitoring period. The tower sensor array is configured to interconnect with the edge computing terminal via an Internet of Things (IoT) communication protocol to collect sensing data based on the tower coordinate system. The edge computing terminal is configured to connect to the central server cluster via a wired or wireless network, and is used to perform denoising and coordinate correction on the image data; The central server cluster is configured to be interconnected with the storage array via a high-speed data bus for performing cross-time period image fusion and trend determination. The method includes the following steps: S1: Acquire image data of the monitoring area within the monitoring period and acquire sensing data based on the tower coordinate system; wherein, the monitoring period includes the first monitoring period and the second monitoring period; the tower coordinate system is composed of the above-ground vegetation distribution network and the groundwater variation network, and the above-ground vegetation distribution network and the groundwater variation network are coupled. S2: Based on the first monitoring time period, determine the first index value corresponding to at least one environmental and water conservation index associated with vegetation status, and extract the first region image of the first index region to which the first index value belongs; wherein, the image data is used to determine the plant status characteristics corresponding to the first index value and to determine the plant distribution characteristics corresponding to the first index region, and the sensing data is the hydrological and soil characteristics that cause changes in plant status characteristics and plant distribution characteristics; at the same time, based on the second monitoring time period, extract the second region image corresponding to the environmental and water conservation index; S3: Perform spatiotemporal alignment before image fusion, and set spatial reference parameters using the positioning data of the tower coordinate system; based on the spatial reference parameters, compare the second region image with the first region image to determine the pixel deviation; S4: Based on the pixel deviation, the second region image corresponding to the environmental and water conservation index within the second monitoring time period is fused with the first region image to generate a fusion difference map. The image deviation is determined based on the fusion difference map, and the plant status trend and plant distribution trend are determined based on the image deviation.

[0011] Furthermore, the image data includes satellite remote sensing image data and UAV aerial photography data; the satellite remote sensing image data is acquired by the satellite ground receiving station and received through the central server cluster; The monitoring method employs a three-tiered, progressive regulatory framework to oversee the monitored area: Level 1: Conduct a comprehensive survey of the monitoring area using satellite remote sensing to obtain satellite remote sensing images of the entire area and identify areas with suspected environmental and water conservation problems; Level 2: Based on the suspected environmental and water conservation problem areas, conduct detailed investigations of key areas using drones to obtain drone aerial images of key areas and confirm whether environmental and water conservation problems exist in the suspected problem areas. Level 3: Based on the areas identified as having environmental and water conservation issues in Level 2, supplementary verification of these areas will be conducted through on-site inspections. The scope of the detailed investigation of key areas at the second level is dynamically triggered and determined by the results of the overall survey at the first level, while the scope of the on-site verification at the third level is dynamically triggered and determined by the results of the detailed investigation at the second level.

[0012] Furthermore, image data and perceptual data are acquired, including: During the monitoring period, a reference frame is acquired. The reference frame is a target frame image of the monitoring area that has no missing features during the first monitoring period. The target frame image is configured as the starting frame within the monitoring time period and is used as an input frame for subsequent processing. When the target frame image is used as an input frame, the quantitative data of the starting frame is used as a reference, and incremental data is superimposed to obtain the forward monitoring frame image of the monitoring area. The incremental data is the measured change of environmental and water conservation indicators between two adjacent frames. The first monitoring time period includes a starting frame and at least one forward monitoring frame image representing the real-time value of at least one environmental and water conservation index. The forward monitoring frame images are arranged in chronological order, and the last forward monitoring frame image in the first monitoring time period is the initial frame image of the environmental and water conservation index corresponding to the second monitoring time period.

[0013] Furthermore, during image data acquisition, the calibrated reflectivity is determined based on the target frame image; When a forward monitoring frame image is received, it is determined whether the real-time reflectance of the forward monitoring frame image conforms to the calibrated reflectance. Among them, when the difference between the real-time reflectance and the calibrated reflectance is within the preset threshold range, it is determined to meet the calibrated reflectance, and the forward monitoring frame images are arranged in sequence. When the difference between the real-time reflectance and the calibrated reflectance exceeds the preset threshold range, the CMOS gain and aperture value of the image acquisition device for the forward monitoring frame image are adjusted according to the RGB mean value of the target frame image, and the image is re-acquired until the difference between the brightness value of the re-acquired forward monitoring frame image and the brightness value of the target frame image is within the preset brightness tolerance range. Then, the forward monitoring frame images are arranged sequentially.

[0014] Furthermore, the first indicator area is defined, including: When the edge computing terminal receives the image data, it divides the image into 200×200 pixel grid units, and each grid unit is superimposed with the sensing data at the corresponding location to generate a feature matrix. Based on a preset vegetation coverage threshold, vegetation areas are determined in each grid cell; When the sensing data of consecutive adjacent grid cells can all be used to calculate the first index value, the corresponding grid cell is classified as the first index region; wherein, the first index value is the value of the environmental and water conservation index associated with the vegetation status.

[0015] Furthermore, the first region image is enhanced by filtering with the first filtering band to form a first bandpass filtered image, and then filtered with the second filtering band to form a second bandpass filtered image; based on the response difference between the first bandpass filtered image and the second bandpass filtered image, the visible adjacency boundary between the vegetated area and the non-vegetated area is extracted.

[0016] Furthermore, the spatiotemporal alignment operation and pixel offset determination in S3 specifically include: Spatial reference parameters are set using the positioning data from the tower coordinate system; Based on spatial reference parameters, the vegetation features of the second region image and the first region image are compared pixel by pixel to determine the pixel deviation. Based on pixel deviation, determine the trend of plant status and plant distribution: for the same pixel location, determine the trend of plant status improvement or deterioration based on the positive or negative value of the vegetation index difference; determine the trend of plant distribution expansion or contraction based on the increase or decrease of the number of pixels in the vegetation coverage area.

[0017] Furthermore, prior to image fusion, the following steps are also included: Binarized vegetation distribution maps of the first and second region images are determined respectively, and pixel determination criteria for vegetation regions are set. Based on the pixel determination criteria, morphological processing is used to extend the vegetation outline on the binary vegetation distribution map, and predefined structural elements are used to enhance the outline. The enhanced outlines are overlaid onto the georeferencing base map to analyze the area characteristics and morphological properties of the regions with outline differences. The vegetation change category is determined based on the differences in morphological characteristics, and the fusion anomaly area is determined based on the vegetation change category.

[0018] Furthermore, fusing the second region image with the first region image includes: Using the tower coordinates as the reference point, construct a geometrically registered vector layer of the vegetation restoration area; Based on the vector layers, calculation channels corresponding to different image deviation indices are constructed.

[0019] Furthermore, the image acquisition equipment is a multispectral UAV or a fixed-point long-term image monitoring device, and the tower sensor group includes a soil moisture sensor, a runoff velocity sensor, and a rainfall sensor; Edge computing terminals are installed in substations or inspection stations along the line, while the central server cluster is deployed in the power grid company's data center. The method further includes: Based on plant status trends, plant distribution trends, and image deviations, an environmental and water conservation monitoring report is generated. When the deviation of the plant state determined according to the plant state trend exceeds the preset alarm threshold, or when the proportion of pixels with the attenuation distribution trend in the plant distribution trend exceeds the preset alarm threshold, a data acquisition frequency adjustment command is sent to the edge computing terminal.

[0020] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention realizes the spatial coupling and causal relationship between image data and sensing data by constructing an environmental and water conservation monitoring architecture based on the tower coordinate system. Specifically, by positioning the sensing data as hydrological and soil characteristics that cause changes in plant state characteristics and plant distribution characteristics, a two-layer analytical architecture of "image characterization results and sensing characterization causes" is established, which can accurately distinguish the driving factors of vegetation change and improve the accuracy of environmental and water conservation index analysis.

[0021] (2) By establishing a tower coordinate system coupled with the aboveground vegetation distribution network and the underground hydrological variation network, and taking the transmission and transformation line towers as the spatial reference, this invention achieves accurate spatial registration of aboveground vegetation distribution characteristics and underground hydrological characteristics, eliminates spatial positioning errors under conventional coordinate systems, and improves the consistency of multi-source data fusion.

[0022] (3) This invention generates a fusion difference map by fusing the first area image during the first monitoring period with the second area image during the second monitoring period. The image deviation is determined based on the fusion difference map, and the plant status trend and plant distribution trend are determined based on the positive and negative values ​​and magnitude of the image deviation. This achieves an improvement in the monitoring dimension from "whether it changes" to "how it changes", providing a quantitative basis for the formulation of environmental protection and water conservation measures.

[0023] (4) The present invention uses a three-level progressive regulatory framework to dynamically determine the monitoring scope of the next level based on the monitoring results of the previous level, thereby achieving precise supervision from coarse to fine and from general to specific. This avoids the problems of repeated or missed monitoring in traditional methods, improves monitoring efficiency, and reduces labor costs.

[0024] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and drawings.

[0025] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0026] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram illustrating the steps of a water and environmental protection monitoring method for power transmission and transformation lines based on image fusion in an embodiment of the present invention. Figure 2 This is a schematic diagram of the architecture of a power transmission and transformation line environmental protection and water conservation monitoring system according to an embodiment of the present invention; Figure 3 This is a flowchart illustrating the three-level progressive regulatory architecture in an embodiment of the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0028] Example 1: This example provides a method for monitoring the environmental protection and water conservation of power transmission and transformation lines based on image fusion. This method is applied to a monitoring system for the environmental protection and water conservation of power transmission and transformation lines. Figure 2 As shown, the power transmission and transformation line environmental protection and water conservation monitoring system includes: image acquisition equipment, tower sensor group, edge computing terminal and central server cluster.

[0029] The image acquisition equipment is configured to connect to an edge computing terminal via a wireless data transmission link to collect multispectral image data of the monitored area during the monitoring period. The image acquisition equipment can be a multispectral drone or a fixed-point long-term image monitoring device. When using a multispectral drone, the drone carries a multispectral camera and flies along the transmission line according to a planned route to collect high-resolution multispectral image data. When using a fixed-point long-term image monitoring device, a fixed multispectral camera is installed near the tower or at key locations along the line to periodically collect image data of the monitored area.

[0030] The tower sensor array is configured to interconnect with edge computing terminals via an IoT communication protocol to collect sensing data based on the tower coordinate system. The tower sensor array includes a soil moisture sensor, a runoff velocity sensor, and a rainfall sensor. The soil moisture sensor is buried at a predetermined depth below the ground surface around the tower to monitor soil moisture content in real time; the runoff velocity sensor is installed on surface runoff channels near the tower to monitor surface runoff velocity; and the rainfall sensor monitors the amount of rainfall in the area.

[0031] Edge computing terminals are configured to connect to the central server cluster via wired or wireless networks to perform denoising and coordinate correction on image data. These terminals are installed in substations or inspection stations along the route and possess data storage and preliminary computing capabilities. When the image acquisition equipment is a multispectral UAV, mobile terminal devices can be used as edge computing terminals to achieve real-time processing of on-site data.

[0032] The central server cluster is configured to interconnect with the storage array via a high-speed data bus to perform cross-time period image fusion and trend determination. Deployed within the power grid company's data center, the central server cluster possesses high-performance computing capabilities and large-capacity storage capacity.

[0033] Based on the above system, such as Figure 1 As shown, the monitoring method in this embodiment includes the following steps: S1: Acquire image data of the monitored area during the monitoring period and acquire sensing data based on the tower coordinate system.

[0034] The monitoring period includes a first monitoring period and a second monitoring period. The first monitoring period is the baseline period, used to establish the initial state baseline for environmental and water conservation monitoring; the second monitoring period is the comparison period, used for comparative analysis with the first monitoring period. The first monitoring period can be selected during a period when vegetation growth is relatively stable, such as the initial growth period of vegetation in spring of a certain year; the second monitoring period can be selected during the same phenological period of another year to eliminate the natural differences in vegetation growth caused by seasonal changes.

[0035] The tower coordinate system consists of a surface vegetation distribution network and a groundwater variation network, which are coupled. The surface vegetation distribution network is constructed by extracting spatial features such as vegetation cover, density, and health from image data; the groundwater variation network is constructed by extracting spatial features such as soil moisture, groundwater runoff direction, and water level changes from the tower sensor array. After spatial registration based on the tower coordinates, a quantitative correlation model (e.g., the functional relationship between vegetation cover and groundwater depth) is established to ensure that the vegetation status and hydrological parameters at the same geographic point can be accurately correlated.

[0036] In this embodiment, the tower coordinate system is a dedicated coordinate system constructed using the towers of the transmission and transformation line as spatial reference points. The tower coordinate system is a geometric coordinate system constructed using the towers of the transmission and transformation line as spatial references; a network of above-ground vegetation distribution and a network of underground hydrological changes are established and coupled within this tower coordinate system. All image data and sensing data are spatially registered using the precise GPS coordinates of the towers and transformed to this coordinate system, eliminating spatial offset errors between different data sources.

[0037] In this embodiment, the aboveground vegetation distribution network refers to a data network composed of vegetation spatial distribution information extracted from image data, including spatial distribution characteristics such as vegetation coverage, vegetation type, vegetation density, and vegetation health. The groundwater variation network refers to a data network composed of hydrological and soil parameters extracted from the tower sensor array, including spatial distribution characteristics such as soil moisture distribution, runoff direction, and water level changes. The coupling of the aboveground vegetation distribution network and the groundwater variation network means that the two networks are spatially superimposed and correlated in the tower coordinate system to establish a quantitative response relationship between vegetation status and hydrological and soil parameters.

[0038] S2: Based on the first monitoring time period, determine the first index value corresponding to at least one environmental and water conservation index associated with vegetation status, and extract the first region image of the first index region to which the first index value belongs.

[0039] The image data is used to determine the plant state characteristics corresponding to the first indicator value and the plant distribution characteristics corresponding to the first indicator region. Plant state characteristics include indicators reflecting plant growth status such as vegetation coverage, leaf chlorophyll content, and vegetation health index; plant distribution characteristics include spatial location, area, and boundary shape of the vegetation-covered area, reflecting the spatial distribution of plants. Simultaneously, based on the second monitoring time period, images of the second region corresponding to the environmental and water conservation indicators are extracted.

[0040] Sensing data refers to hydrological and soil characteristics that trigger changes in plant state and distribution. These characteristics include parameters reflecting hydrological conditions and soil status, such as soil moisture, runoff velocity, rainfall, and soil water content. In this invention, sensing data is explicitly positioned as the "causal interpretation layer," meaning that image data represents the "monitoring results" (what the plant state and distribution are), while sensing data represents the "driving causes" (why these changes occur), thus establishing a two-layer analytical framework of "image representation results—sensing data representation causes."

[0041] In this embodiment, environmental protection and water conservation indicators refer to specific quantifiable indicators used to evaluate the impact of power transmission and transformation line projects on the ecological environment along the line. Environmental protection and water conservation indicators include at least one of the following: area of ​​landslide and debris disposal area, vegetation restoration area, and vegetation coverage. The first indicator value refers to the specific value of the environmental protection and water conservation indicator within the first monitoring period, for example, the vegetation coverage within the first monitoring period is 70%. The first indicator area refers to the spatial area corresponding to the first indicator value; that is, within this area, the environmental protection and water conservation indicator is taken as the first indicator value.

[0042] S3: Perform spatiotemporal alignment before image fusion, and set spatial reference parameters using the positioning data of the tower coordinate system; based on the spatial reference parameters, compare the second region image with the first region image to determine the pixel deviation.

[0043] Spatiotemporal alignment includes two aspects: temporal alignment and spatial alignment. Temporal alignment refers to calibrating the second region image with the first region image in the time dimension to ensure that the images from the two time phases correspond to the same season or phenological period, eliminating differences in natural vegetation growth caused by seasonal changes. Spatial alignment refers to accurately registering the second region image with the first region image in the spatial dimension based on the positioning data of the tower coordinate system, eliminating spatial offsets caused by factors such as image acquisition angle, platform attitude, and terrain undulations.

[0044] Spatial reference parameters include the tower base center coordinates, image projection parameters, and pixel ground resolution. Based on these spatial reference parameters, the second region image and the first region image are strictly aligned to the same tower coordinate system, eliminating offsets caused by terrain or equipment errors.

[0045] Based on spatiotemporal alignment, the vegetation features of the second region image and the first region image are compared pixel by pixel to determine pixel deviation. Vegetation features include Normalized Difference Vegetation Index (NDVI), percentage of green pixels, etc. Pixel deviation can be the difference in NDVI value, the difference in the number of green pixels, etc.

[0046] In this embodiment, the Normalized Difference Vegetation Index (NDVI) is a quantitative indicator reflecting the growth status of vegetation. The calculation formula is: NDVI = (NIR - RED) / (NIR + RED), where NIR is the near-infrared reflectance and RED is the red reflectance. NDVI values ​​range from -1 to 1. Healthy vegetation typically has an NDVI value above 0.3; a higher NDVI value indicates healthier and denser vegetation cover. In environmental and water conservation monitoring, the temporal changes in NDVI can reflect the recovery or degradation trend of vegetation.

[0047] In this embodiment, pixel deviation refers to the difference in pixel values ​​between the second region image and the first region image at the same geographical location. The sign of the pixel deviation reflects the direction of vegetation change; a positive value indicates that the vegetation condition at that location has improved (enhanced), while a negative value indicates that the vegetation condition at that location has deteriorated (decreased). By statistically analyzing the distribution of global pixel deviation, the intensity distribution of overall vegetation change can be determined.

[0048] S4: Based on the pixel deviation, the second region image corresponding to the environmental and water conservation index within the second monitoring time period is fused with the first region image to generate a fusion difference map. The image deviation is determined based on the fusion difference map, and the plant status trend and plant distribution trend are determined based on the image deviation.

[0049] Image fusion refers to the process of overlaying and comparing a second region image with a first region image at the pixel level to generate a fused difference map. In this embodiment, the image fusion adopts a pixel-level fusion method, which involves performing difference or ratio calculations on each pixel of the two images in the same coordinate system. Before fusion, it is necessary to ensure that the two images have undergone radiometric and atmospheric corrections to eliminate radiometric differences caused by different lighting and atmospheric conditions.

[0050] Image bias refers to the difference in pixel values ​​at various locations in a fused difference image, including biases in vegetation cover and leaf chlorophyll content. The magnitude of image bias reflects the extent of changes in vegetation condition.

[0051] Plant status trends refer to the direction of change in a plant's physiological state, including trends of improvement (increased vegetation health, increased chlorophyll content) and trends of deterioration (decreased vegetation health, reduced chlorophyll content). Plant distribution trends refer to the direction of change in the spatial distribution range of plants, including trends of expansion and contraction.

[0052] Based on the sign and magnitude of image deviations, both plant condition trends and plant distribution trends can be determined simultaneously. For example, an increase in NDVI value indicates improved plant condition, while a decrease in NDVI value indicates deterioration; an increase in vegetation cover pixels indicates expanded plant distribution, while a decrease in vegetation cover pixels indicates shrunken plant distribution.

[0053] In this embodiment, the groundwater variation network refers to a data network composed of surface and groundwater parameters extracted from the tower sensor group. The hydrological parameters include, but are not limited to, information that can reflect the hydrological conditions of the monitoring area, such as soil moisture, surface runoff velocity, rainfall, and groundwater level changes.

[0054] In this embodiment, the data storage and retrieval process for each step includes: Data format: Image data is multispectral raster data, including four bands: red, green, blue, and near-infrared, with each band stored as a 16-bit integer; Sensing data is time-series numerical data, including soil moisture (%), runoff velocity (m / s), and rainfall (mm), stored in floating-point format; Tower coordinate system is spatial reference data, including the coordinate origin, axial direction, and projection parameters of each tower.

[0055] Storage method: Image data is stored in the distributed file system of the central server cluster, and an index directory is established according to the monitoring time period and tower number; Sensing data is stored in the time series database of the central server cluster, and an index is established according to the tower number and timestamp; Tower coordinate system parameters are stored in the relational database of the central server cluster.

[0056] Reading process: After receiving image data uploaded by the image acquisition device, the edge computing terminal sends the image data to the central server cluster via a wireless data transmission link; the tower sensor group uploads the sensing data to the edge computing terminal in real time via the Internet of Things communication protocol, and the edge computing terminal cleans the sensing data and forwards it to the central server cluster; the central server cluster reads the image data and sensing data from the storage array via a high-speed data bus for subsequent processing.

[0057] The main technical contribution of this embodiment lies in solving the technical problems of distinguishing the driving factors of vegetation change and quantifying trends in environmental and water conservation monitoring of power transmission and transformation projects. By constructing a tower coordinate system coupled with the aboveground plant distribution network and the underground hydrological change network, a two-layer analytical architecture of "image representation results and perception representation causes" is established. This elevates the perception data from parallel data sources to a causal explanation layer, solving the technical problem in the prior art of distinguishing whether vegetation change is caused by natural growth, soil erosion, or human destruction. Through pixel-level fusion and deviation analysis of regional images in two time periods, the monitoring dimension is improved from "whether it changes" to "how it changes," enabling quantitative determination of plant status trends and distribution trends, and providing a quantitative basis for the formulation of environmental and water conservation measures.

[0058] Example 2: Based on Example 1, this example further specifies that the image data includes satellite remote sensing image data and UAV aerial photography data. The satellite remote sensing image data is acquired by a satellite ground receiving station and received through the central server cluster; for example... Figure 3 As shown, the monitoring method employs a three-tiered progressive regulatory framework to oversee the monitored area.

[0059] Level 1: Conduct a comprehensive survey of the monitored area using satellite remote sensing to acquire full-line satellite remote sensing imagery and identify areas suspected of environmental and water conservation issues. The satellite remote sensing imagery utilizes multispectral and panchromatic images acquired by high-resolution remote sensing satellites, such as the domestic Gaofen-2 satellite and the US SkySat satellite. Satellite remote sensing offers advantages such as a large detection range, rapid data acquisition, strong repeatability, and independence from terrain and regional limitations, enabling rapid surveys of the entire transmission and transformation line. Areas suspected of environmental and water conservation issues refer to areas initially identified based on satellite remote sensing imagery that may have environmental and water conservation problems. These potential problems include: abnormal expansion of construction disturbance areas, failure to restore temporary land occupation, signs of landslides and debris spills, and vegetation damage. Identification of suspected problem areas can be achieved through a combination of automatic interpretation and human-computer interactive visual interpretation.

[0060] Level Two: Based on the areas suspected of environmental and water conservation issues, a detailed investigation of key areas is conducted using drones to acquire aerial photographs of these areas and confirm the existence of environmental and water conservation problems. Vertical take-off and landing (VTOL) drones or multi-rotor drones are used. Compared to satellite remote sensing, drone aerial photographs have higher spatial resolution, reaching below 0.1m, and can identify content that satellite remote sensing cannot, such as engineering measures like drainage ditches and temporary barriers. The detailed investigation of key areas includes: the extent of tower foundation disturbance, the implementation status of engineering measures such as slope protection and drainage ditches, the implementation of temporary measures, the situation of slope erosion and debris disposal in hilly areas, and the status of topsoil stripping, preservation, and utilization.

[0061] Level 3: Based on the areas identified as having environmental and water conservation issues in Level 2, supplementary verification is conducted through on-site inspections. On-site inspections involve comparing the specific power transmission and transformation project's environmental impact report (form) and its approval, and the water and soil conservation plan report (form) and its approval, combined with satellite remote sensing and drone monitoring to identify environmental and water conservation issues. On-site inspections must be recorded using photographs, videos, written descriptions, and forms, and archived according to tower numbers. The scope of the detailed investigation in Level 2 is dynamically triggered by the results of the Level 1 comprehensive survey, and the scope of the on-site inspection in Level 3 is dynamically triggered by the results of the Level 2 detailed investigation.

[0062] In this embodiment, the "dynamic triggering determination" is implemented as follows: After the first-level comprehensive survey is completed, the central server cluster automatically generates a list of suspected problem areas according to the preset suspected problem judgment rules. This list contains the spatial coordinates and problem type of each suspected problem area. This list serves as the task input for the second-level UAV detailed survey. The central server cluster automatically plans the UAV flight path based on the area coordinates in the list to determine the key detailed survey area. After the second-level UAV detailed survey is completed, the system generates a list of confirmed problem areas based on the survey results. This list serves as the task input for the third-level on-site verification to determine the on-site verification points and verification content.

[0063] In this embodiment, suspected problem areas refer to spatial areas identified based on automatic or visual interpretation of satellite remote sensing images that may have violations of environmental protection and water conservation regulations or inadequate environmental protection and water conservation measures. The criteria for determining suspected problem areas include: construction disturbance exceeding the scope of prevention and control responsibilities, failure to restore temporary land occupation as required, signs of landslides or debris spills, and abnormally low vegetation cover.

[0064] In this embodiment, the data storage and retrieval process for each step includes: Data formats: Satellite remote sensing images are multispectral and panchromatic fusion images with a spatial resolution better than 1m; UAV aerial images are visible light images with a spatial resolution better than 0.1m; suspected problem areas are vector surface data; environmental and water conservation problem confirmation records are structured data.

[0065] Storage method: Satellite remote sensing images and UAV aerial images are stored in a distributed file system, indexed by monitoring level and monitoring time period; suspected problem areas and problem confirmation records are stored in a spatial database.

[0066] Reading process: The central server cluster reads satellite remote sensing images from the distributed file system to conduct a comprehensive survey and generate a list of suspected problem areas; based on this list, it reads drone aerial images of the corresponding areas from the distributed file system for detailed investigation; based on the detailed investigation results, it generates a list of on-site verification tasks for use by on-site verification personnel.

[0067] The main technical contribution of this embodiment is to solve the technical problem of low regulatory efficiency in environmental and water conservation monitoring of power transmission and transformation projects: by constructing a three-level progressive regulatory framework of "satellite full-line survey, drone key detailed survey, and on-site supplementary verification", and making the monitoring scope of the next level dynamically triggered by the monitoring results of the previous level, it realizes precise supervision "from coarse to fine, from surface to point". This not only avoids the problem of repeated monitoring or missed monitoring caused by the independent operation of satellites, drones and on-site in traditional methods, but also greatly reduces the workload of unnecessary on-site verification by focusing at each level, thereby improving regulatory efficiency.

[0068] Example 3: Based on Example 1, this example further specifies the specific implementation method for acquiring image data and perception data.

[0069] During the monitoring period, a reference frame is acquired. The reference frame is a target frame image of the monitoring area that has no missing features during the first monitoring period.

[0070] The selection of reference frames is achieved through an image quality assessment algorithm. This algorithm eliminates blurry, occluded, or missing frames, selecting only those with clear vegetation cover, no cloud cover, and no sensor anomalies as reference frames. These reference frames are highly reliable and can serve as a quantitative benchmark for all subsequent monitoring frames.

[0071] The target frame image is configured as the starting frame within the monitoring period and serves as an input frame for subsequent processing. The starting frame marks the beginning of the monitoring cycle and provides quantitative baseline data. When the target frame image is used as an input frame, incremental data is superimposed on the quantitative data of the starting frame to obtain a forward monitoring frame image of the monitoring area; the incremental data represents the measured changes in environmental and water conservation indicators between adjacent frames.

[0072] The first monitoring period includes a starting frame and at least one forward monitoring frame image representing the real-time value of at least one environmental and water conservation index. The forward monitoring frame images are arranged in chronological order, and the last forward monitoring frame image in the first monitoring period is the initial frame image of the corresponding environmental and water conservation index in the second monitoring period.

[0073] Specifically, the first monitoring period includes a starting frame and multiple forward monitoring frames acquired chronologically, forming a continuous sequence of "starting frame, monitoring frame 1, monitoring frame 2, monitoring frame n". The last forward monitoring frame of the first monitoring period is directly used as the initial frame of the second monitoring period, replacing the actual image of the first frame of the second monitoring period in traditional methods. This benchmark consistency mechanism for cross-period analysis results in a lower calculation deviation when comparing the actual monitoring data of the second monitoring period with the initial frame, achieving reverse calibration and improving the accuracy of image selection for subsequent periods.

[0074] In this embodiment, the reference frame refers to an image frame selected as a reference standard within the monitoring period. The reference frame has the following characteristics: no cloud cover, no sensor noise, clear vegetation cover areas, and complete and identifiable ground features. The quality of the reference frame directly affects the quantitative analysis accuracy of all subsequent monitoring frames.

[0075] In this embodiment, the target frame image refers to an image frame configured as both a start frame and an input frame. When the target frame image serves as the start frame, it provides initial quantitative data for the monitoring cycle; when it serves as the input frame, it acts as a reference input for the generation of subsequent forward monitoring frames. The target frame image plays the role of a "static reference".

[0076] In this embodiment, incremental data refers to the change in environmental and water conservation indicators actually acquired by the image acquisition device between two adjacent frames, such as the measured difference in vegetation coverage between two adjacent monitoring sessions.

[0077] In this embodiment, the forward monitoring frame image refers to the monitoring frame image generated by superimposing incremental data on the quantitative data of the starting frame. The forward monitoring frame images are arranged in chronological order, reflecting the continuous change process of environmental and water conservation indicators within the monitoring period. Incremental data refers to the change in environmental and water conservation indicators between two adjacent frames, such as a daily change of 0.1% in vegetation cover or a decrease of 0.2% in soil moisture.

[0078] In this embodiment, the data storage and retrieval process for each step includes: Data format: The reference frame and target frame are raster image data; the forward monitoring frame is raster image data, with each frame accompanied by a timestamp and incremental data record; the incremental data is time-series numerical data.

[0079] Storage method: Reference frames and target frames are stored in the "Baseline Image" directory of the distributed file system; forward monitoring frames are stored in the "Time Series Monitoring" directory and numbered in chronological order; incremental data is stored in the time series database.

[0080] Reading process: The central server cluster reads the target frame image as the starting frame from the distributed file system; reads incremental data from the time series database; overlays the starting frame with the incremental data to generate the forward monitoring frame image; stores the forward monitoring frame image in chronological order, and marks the last frame as the initial frame of the second monitoring time period.

[0081] The main technical contribution of this embodiment is to solve the technical problem of inconsistent benchmarks in cross-time period image comparison: by configuring the target frame image to have the dual roles of "starting frame" and "input frame", benchmark consistency in time-series monitoring is achieved; by using the last frame of the forward monitoring frame of the first monitoring time period as the initial frame of the second monitoring time period, the benchmark deviation caused by the lack or abnormality of the first frame features in the cross-time period comparison in the traditional method is eliminated, and the accuracy of cross-time period image fusion is improved.

[0082] Example 4: Based on Example 3, this example further defines the specific implementation method of reflectivity calibration and dynamic correction in image data acquisition.

[0083] During image data acquisition, the calibrated reflectivity is determined based on the target frame image.

[0084] The calibration reflectance is based on the target frame image without missing features. The average reflectance of vegetation, soil, water bodies, and other land cover features is extracted as the calibration reflectance to establish an absolute reference for multi-time period image comparison. Specifically, sample pixels from vegetation areas, bare soil areas, and water bodies are selected in the target frame image, and the average reflectance of each area in each band is calculated as the calibration reflectance of that land cover type in that band.

[0085] When a forward monitoring frame image is received, it is determined whether the real-time reflectivity of the forward monitoring frame image conforms to the calibrated reflectivity.

[0086] The determination method is as follows: Select sample pixels from vegetation areas, bare soil areas, and water bodies in the forward monitoring frame image that are at the same geographical location as the target frame image, and calculate the real-time average reflectance; compare the real-time reflectance with the calibrated reflectance band by band and calculate the difference.

[0087] Specifically, when the difference between the real-time reflectance and the calibrated reflectance is within a preset threshold range, it is determined to meet the calibrated reflectance, and the forward monitoring frame images are sorted. The preset threshold range can be set according to the actual monitoring accuracy requirements; for example, a reflectance difference within ±5% is considered to meet the calibrated reflectance.

[0088] When the difference between the real-time reflectance and the calibrated reflectance exceeds the preset threshold range, the CMOS gain and aperture value of the image acquisition device for the forward monitoring frame image are adjusted according to the RGB mean of the target frame image, and the image is re-acquired until the difference between the brightness value of the re-acquired forward monitoring frame image and the brightness value of the target frame image is within the preset brightness tolerance range. Then the forward monitoring frame images are sorted.

[0089] Specifically, when the real-time reflectivity does not conform to the calibrated reflectivity, the average RGB value of each band of the target frame image is calculated as the target brightness value; the brightness of the forward monitoring frame is dynamically corrected by adjusting the CMOS gain and aperture value of the image acquisition device; the brightness change of the forward monitoring frame is monitored in real time until the difference between its brightness value and the brightness value of the target frame image is within the preset brightness tolerance range, and then the corrected forward monitoring frame images are sorted.

[0090] In this embodiment, the calibrated reflectance refers to the standard reflectance value determined based on the target frame image, which serves as the radiation reference for all subsequent monitoring frames. The calibrated reflectance reflects the ground object reflection characteristics of the monitoring area under ideal conditions and is a benchmark reference for eliminating differences in radiation between images from different time periods.

[0091] In this embodiment, for the same land cover type, under the same illumination conditions, there is a linear mapping relationship between RGB brightness values ​​and reflectance values ​​of each band. When the difference between the RGB brightness values ​​of the forward monitoring frame image and the target frame image is brought within a preset brightness tolerance range by adjusting the CMOS gain and aperture value, it can be determined that the reflectance of each band of the forward monitoring frame image has returned to the allowable deviation range of the calibrated reflectance.

[0092] In this embodiment, real-time reflectance refers to the actual reflectance value of ground objects in the forward monitoring frame image. Due to changes in factors such as lighting conditions, atmospheric conditions, and sensor status, the real-time reflectance may deviate from the calibrated reflectance.

[0093] In this embodiment, CMOS gain refers to the signal amplification factor of the CMOS sensor in the image acquisition device. Increasing the CMOS gain can improve image brightness, but it will also amplify noise; decreasing the CMOS gain can reduce brightness, but it may lose weak signals. Aperture value refers to the size of the lens aperture, which affects the amount of light entering the lens. Adjusting the aperture value can change the image brightness, but it will also affect the depth of field and resolution.

[0094] In this embodiment, the preset threshold range refers to the allowable deviation range used to determine whether the real-time reflectance meets the calibrated reflectance. Setting the preset threshold range requires balancing monitoring accuracy and equipment adjustment capabilities. A threshold that is too small will cause a large number of normal images to be judged as unqualified and trigger correction, increasing the processing burden; a threshold that is too large will cause radiation deviations to not be corrected in time, affecting the accuracy of subsequent analysis. The preset brightness tolerance range refers to the allowable deviation range used to determine whether the brightness of the forward monitoring frame image has been adjusted to be consistent with the target frame image.

[0095] In this embodiment, the data storage and retrieval process for each step includes: Data format: Calibrated reflectance is a floating-point array, organized by band and ground feature type; real-time reflectance is a floating-point array; CMOS gain and aperture values ​​are integers or floating-point numbers; RGB mean values ​​are three integer values.

[0096] Storage method: The calibrated reflectance is stored in a relational database and associated with the target frame image ID; the real-time reflectance is temporarily stored in the memory of the edge computing terminal; the CMOS gain and aperture value are stored in the configuration file of the image acquisition device.

[0097] Reading process: The edge computing terminal extracts sample pixels from the forward monitoring frame image and calculates the real-time reflectance; it reads the calibrated reflectance from the database; it compares the real-time reflectance with the calibrated reflectance, and if it exceeds the preset threshold range, it sends an adjustment command to the image acquisition device. The image acquisition device adjusts the CMOS gain and aperture value according to the command and re-acquires the image until the brightness meets the requirements.

[0098] The main technical contribution of this embodiment is to solve the technical problem of inconsistent radiometric values ​​in images across multiple time periods: by determining the calibrated reflectance based on the target frame image without feature loss, an absolute radiometric reference for comparing images across multiple time periods is established; by comparing the real-time reflectance with the calibrated reflectance, radiometric deviations are automatically identified; and by dynamically adjusting the CMOS gain and aperture value, radiometric correction of the forward monitoring frame image is achieved, ensuring that images in different monitoring time periods have visual and radiometric consistency, providing high-quality input data for subsequent pixel-level fusion.

[0099] Example 5: Based on Example 1, this example further defines the method for determining the first indicator region.

[0100] When the edge computing terminal receives image data, it divides the image into 200×200 pixel grid units, and each grid unit is superimposed with the sensing data at the corresponding location to generate a feature matrix.

[0101] Based on a preset vegetation coverage threshold, vegetation areas are determined in each grid cell.

[0102] Image segmentation involves dividing a complete remote sensing image into regular grid units of a fixed pixel size (200×200 pixels). Each grid unit corresponds to a ground area of ​​approximately 40m×40m, depending on the image resolution. The 200×200 pixel grid size was determined based on a combination of accuracy requirements and computational efficiency for environmental and water conservation monitoring of power transmission and transformation projects: according to the typical size of the disturbance range of tower foundations in environmental and water conservation monitoring of power transmission and transformation projects and the spatial resolution of the image, a 200×200 pixel grid size can cover the typical disturbance range of a single tower, while also considering computational efficiency. This standardized segmentation method eliminates the subjectivity of manual segmentation.

[0103] Each grid cell is overlaid with the corresponding location's sensing data to generate a feature matrix. The feature matrix is ​​constructed as follows: extract the pixel values ​​of all pixels within the grid cell in each band, and calculate statistical measures such as the mean and standard deviation; simultaneously, acquire the sensing data of the corresponding geographical location of the grid cell; combine the image feature statistics with the sensing data to form a multidimensional feature matrix. The feature matrix can simultaneously reflect vegetation status and environmental factors.

[0104] When the sensing data of consecutively adjacent grid cells can all yield a first index value, the corresponding grid cell is classified as a first index region. The first index value is the value of an environmental and water conservation index associated with vegetation status.

[0105] Continuous adjacent grid cells refer to a set of grid cells that are spatially adjacent and without any breaks. Only when all continuously adjacent grid cells meet the condition that "the perceived data can be used to calculate the first index value" are these grid cells classified as the first index region.

[0106] "The sensing data can calculate the first indicator value" means that based on the sensing data of the grid cell, combined with a preset calculation model or empirical formula, the specific value of the environmental and water conservation indicator can be calculated. For example, when the first indicator value is "soil moisture ≥ 20% and vegetation survival", this condition can only be met if the sensing data of the grid cell can support the calculation of this indicator.

[0107] In this embodiment, a grid cell refers to a basic analysis unit obtained by dividing the remote sensing image into segments with a fixed pixel size. The 200×200 pixel grid size ensures sufficient spatial resolution to identify vegetation changes around the tower base while avoiding an increase in computational load due to overly fine segmentation.

[0108] In this embodiment, the feature matrix refers to a multi-dimensional data matrix generated by overlaying image features and perceived data. Each row of the feature matrix corresponds to a grid cell, and each column corresponds to a feature dimension. The feature matrix serves as the data foundation for subsequent indicator region classification.

[0109] In this embodiment, consecutive adjacent grid cells refer to a set of grid cells that are adjacent in spatial topology. The determination of consecutive adjacency adopts the eight-neighborhood rule, that is, a grid cell is considered to be adjacent to grid cells in the eight directions above, below, left, right, upper left, upper right, lower left, and lower right.

[0110] In this embodiment, the data storage and retrieval process for each step includes: Data format: grid cells are regular blocks of raster data; feature matrix is ​​a two-dimensional floating-point array; the first index region is vector surface data.

[0111] Storage method: Grid cell data is stored in a distributed file system and indexed by row and column numbers; feature matrices are stored in a columnar database for easy and fast querying by feature dimension; vector data of the first index region is stored in a spatial database.

[0112] Reading process: After receiving image data, the edge computing terminal segments the image according to a size of 200×200 pixels; for each grid cell, it reads the corresponding location's sensing data from the database; it overlays the image features and sensing data to generate a feature matrix; it traverses all grid cells, identifies the set of grid cells that are continuously adjacent and whose sensing data can all be used to calculate the first index value, and classifies them as the first index region.

[0113] The main technical contribution of this embodiment is to solve the technical problem of insufficient consistency of cause in the determination of the index region: by dividing the image into standardized grid units and superimposing them with the sensing data to generate a feature matrix, pixel-level spatial coupling between the image domain and the sensing domain is realized; by using the dual screening rule that "the sensing data of continuously adjacent grid units can all be used to calculate the first index value", the first index region is ensured to have both spatial continuity and consistency of cause, avoiding misjudgment of the region due to local anomalies, and providing a high-quality reference region for subsequent dual-time period image fusion.

[0114] Example 6: Based on Example 1, this example further defines the bandpass filter region division in the first region image.

[0115] The first region image is enhanced by filtering with the first filtering band to form a first bandpass filtered image, and then filtered with the second filtering band to form a second bandpass filtered image. Based on the response difference between the first bandpass filtered image and the second bandpass filtered image, the visible adjacency boundary between the vegetated area and the non-vegetated area is extracted.

[0116] The first filtering band uses the red-edge to near-infrared band (680-750nm), which has high reflectivity to healthy vegetation and can effectively enhance vegetation information. After filtering by the first filtering band, healthy vegetation areas are highlighted in the image, while non-vegetation areas are dark.

[0117] The second filtering band uses the visible light band (500-600nm, i.e., the green light band), which has a good ability to distinguish bare soil and non-vegetated areas. Through filtering with the second filtering band, the characteristics of bare soil and non-vegetated areas are enhanced.

[0118] Based on the differences in the responses of two filtered images to different land cover types, the visible adjacency boundary between vegetated and non-vegetated areas is extracted through edge detection. This means that the boundary between vegetated and non-vegetated areas can be clearly distinguished on the image. This visible adjacency boundary facilitates subsequent region segmentation and contour extraction.

[0119] In this embodiment, bandpass filtering refers to a filtering process that allows only electromagnetic waves within a specific wavelength range to pass through. The first bandpass-filtered image is the image generated after processing with the first filtering band, which highlights the distribution of healthy vegetation. The second bandpass-filtered image is the image generated after processing with the second filtering band, which highlights the distribution of bare soil and non-vegetation areas.

[0120] In this embodiment, the visible adjacency boundary refers to the boundary line between vegetated and non-vegetated areas extracted by edge detection based on the response difference between the first and second bandpass filtered images. The formation of the visible adjacency boundary is due to the significant differences in reflectance among different land cover types at different wavelengths. Healthy vegetation has high reflectance in the near-infrared band and low reflectance in the visible light band, while bare soil is the opposite. By selecting an appropriate filtering band, this difference can be amplified, making the boundaries between different land cover types clearer in the image.

[0121] In this embodiment, the data storage and retrieval process for each step includes: Data format: The first and second bandpass filtered images are raster image data, with the same spatial range and pixel size as the first region image; the visible adjacent boundaries are vector line data.

[0122] Storage method: Bandpass filtered images are stored in a distributed file system; visual adjacency boundaries are stored in a spatial database.

[0123] Reading process: The central server cluster reads the first region image from the distributed file system; filters of the first and second filtering bands are applied respectively to generate the first bandpass filtered image and the second bandpass filtered image; edge detection is performed based on the response difference between the two filtered images to extract the visible adjacent boundaries.

[0124] The main technical contribution of this embodiment is to solve the technical problem of blurred boundaries between vegetated and non-vegetated areas: by using different filtering bands to filter and enhance the image of the first region, the vegetated and non-vegetated areas show significant reflectance differences under different bands; based on the response differences of the two filtered images, the visible adjacent boundaries are extracted, achieving clear segmentation between vegetated and non-vegetated areas, providing accurate regional boundaries for subsequent pixel-level comparison and trend determination.

[0125] Example 7: Based on Example 1, this example further defines the specific implementation method of spatiotemporal alignment operation and pixel deviation determination.

[0126] Perform spatiotemporal alignment and determine pixel offset, including the following sub-steps: Spatial reference parameters are established using positioning data from the tower coordinate system. These parameters include the tower base center coordinates, image projection parameters, and pixel ground resolution. The spatial reference parameters are set based on the precise GPS coordinates of each tower of the transmission line, ensuring that all image data are compared within the same spatial reference frame.

[0127] Based on spatial reference parameters, the vegetation features of the second region image and the first region image are compared pixel by pixel to determine the pixel deviation. On the basis of spatiotemporal alignment, each pixel position in the second region image and the first region image is compared one by one. The vegetation features compared include Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EDI), and Green Chlorophyll Index (HCCI). The pixel deviation is calculated as follows: Deviation value = Pixel value of second region image - Pixel value of first region image.

[0128] Based on pixel deviation, the trends in plant status and distribution are determined: for the same pixel location, the sign of the difference in vegetation index indicates whether the plant status is improving or deteriorating; the increase or decrease in the number of pixels in the vegetation-covered area indicates whether the plant distribution is expanding or shrinking. Plant distribution trends include plant enhancement and decay trends. A plant enhancement trend refers to a trend where the pixel deviation is greater than or equal to the number of pixels in the vegetation-covered area, indicating an expansion of the vegetation distribution range or an increase in vegetation density; a plant decay trend refers to a trend where the pixel deviation is less than or equal to the number of pixels in the vegetation-covered area, indicating a shrinking of the vegetation distribution range or a decrease in vegetation density. In this embodiment, pixel deviation refers to the difference in pixel values ​​between the second region image and the first region image at the same geographical location. The sign of the pixel deviation reflects the direction of vegetation change: for the same pixel location, a positive pixel deviation indicates an increase in the vegetation index at that location, i.e., an improvement in plant status; a negative pixel deviation indicates a decrease in the vegetation index at that location, i.e., a deterioration in plant status. By statistically analyzing the distribution of global pixel deviations, the overall plant distribution trend can be determined.

[0129] Specifically, a positive pixel deviation indicates that the vegetation characteristic value of that pixel location in the second monitoring period is higher than that in the first monitoring period, meaning the plant condition at that location has improved, representing a trend of plant enhancement. A negative pixel deviation indicates that the vegetation characteristic value of that pixel location in the second monitoring period is lower than that in the first monitoring period, meaning the plant condition at that location has deteriorated, representing a trend of plant decline. When the pixel deviation is zero or close to zero, it indicates that the vegetation condition at that location has not changed significantly.

[0130] By statistically analyzing the distribution of global pixel deviations, the overall plant distribution trend in the monitored area can be quantitatively assessed.

[0131] In this embodiment, spatial reference parameters refer to a set of parameters used to unify image data from different time phases and sources under the same spatial reference frame. Spatial reference parameters ensure the precise spatial correspondence between the second region image and the first region image, which is a prerequisite for pixel-level comparison.

[0132] In this embodiment, vegetation features refer to quantitative indicators extracted from remote sensing images that reflect the vegetation growth status, including but not limited to the Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EDI), and Green Chlorophyll Index (RCI). In this embodiment, the plant condition improvement trend refers to a trend of improved plant physiological status determined by a positive difference in the vegetation index, while the plant condition deterioration trend refers to a trend of worsening plant physiological status determined by a negative difference in the vegetation index. The plant distribution expansion trend refers to a trend of expanded plant spatial distribution range determined by an increase in the number of pixels in the vegetation-covered area, while the plant distribution shrinkage trend refers to a trend of reduced plant spatial distribution range determined by a decrease in the number of pixels in the vegetation-covered area. The plant enhancement distribution trend and the plant decay distribution trend are two mutually exclusive trend types based on the positive or negative determination of pixel deviation. The plant enhancement distribution trend indicates an expansion of vegetation coverage or an increase in vegetation density, usually signifying ecological restoration or environmental improvement; the plant decay distribution trend indicates a shrinkage of vegetation coverage or a decrease in vegetation density, usually signifying ecological damage or environmental degradation.

[0133] In this embodiment, the data storage and retrieval process for each step includes: Data format: Spatial baseline parameters are configuration files in JSON format; vegetation feature values ​​are floating-point matrices; pixel deviations are floating-point matrices; plant status trends and plant distribution trends are categorical raster data.

[0134] Storage method: Spatial baseline parameters are stored in a relational database; vegetation feature values ​​and pixel deviations are stored in a distributed file system and saved in GeoTIFF format; plant status trends and plant distribution trends are stored in a spatial database.

[0135] Reading process: The central server cluster reads spatial reference parameters from the database; the second region image and the first region image are registered and resampled based on the spatial reference parameters; vegetation feature values ​​and pixel deviations are calculated pixel by pixel; and plant status trend classification map and plant distribution trend classification map are generated according to the positive and negative values ​​of the pixel deviations.

[0136] The main technical contribution of this embodiment lies in solving the technical problem of difficulty in quantifying the direction of vegetation change: by setting spatial reference parameters through the positioning data of the tower coordinate system, accurate spatial alignment of images from multiple time periods is ensured; by comparing vegetation feature values ​​pixel by pixel to determine pixel deviation, the quantification of vegetation change is achieved; by mapping the positive or negative of pixel deviation to the trend of improvement or deterioration of plant status, and mapping the increase or decrease of the number of pixels in the vegetation coverage area to the trend of expansion or contraction of plant distribution, the monitoring dimension is improved from "whether it changes" to "how it changes", providing a quantitative basis for the formulation of environmental protection and water conservation measures.

[0137] Example 8: Based on Example 1, this example further specifies the binarization and morphological contour enhancement processing before image fusion.

[0138] Before image fusion, the following sub-steps are also included: Binary vegetation distribution maps of the first and second region images are determined separately, and pixel identification criteria for vegetation regions are established. A binary vegetation distribution map is a binary image that separates vegetation regions from non-vegetation regions. Pixel identification criteria for vegetation regions are established, for example: pixels with an NDVI value greater than 0.3 are identified as vegetation pixels, and pixels with an NDVI value less than or equal to 0.3 are identified as non-vegetation pixels. Using this identification criterion, the first and second region images are converted into binary vegetation distribution maps, respectively.

[0139] Based on the pixel determination criteria, morphological processing is used to expand the vegetation outline on the binary vegetation distribution map, and predefined structural elements are used to enhance the outline.

[0140] Morphological processing includes two basic operations: dilation and erosion. Dilation expands the boundaries of vegetation areas and fills in small holes within them; erosion shrinks the boundaries of vegetation areas and removes isolated noisy pixels. By combining dilation and erosion, vegetation outlines can be smoothed, eliminating subtle noise interference.

[0141] Predefined structuring elements are templates used in morphological processing to define the shape and size of a neighborhood. Structuring elements can be circular, square, or cross-shaped, and their size can be determined based on image resolution and vegetation patch size. Using predefined structuring elements for contour enhancement can make vegetation outlines smoother and more continuous.

[0142] The enhanced contour lines are overlaid onto the georegistration base map to analyze the area characteristics and morphological properties of areas with contour differences. The georegistration base map is a reference map that has undergone geocoding and coordinate correction. By overlaying the vegetation contour lines of the enhanced first and second region images onto the same georegistration base map, the differences in vegetation distribution between the two imagery periods can be visually compared.

[0143] Contour difference regions refer to areas where the vegetation contours differ between the first and second region images, including newly added vegetation areas, areas where vegetation has disappeared, and boundary offset areas. Area features refer to the size of the contour difference region, which can be calculated by multiplying the number of pixels by the pixel area. Morphological characteristics refer to the shape features of the contour difference region, including aspect ratio, roundness, and boundary complexity.

[0144] The vegetation change category is determined based on the differences in morphological characteristics, and corresponding fusion anomaly areas are identified based on these categories. The morphological characteristics of the areas with different outlines can determine the vegetation change category. For example, areas with increased area and regular boundaries may indicate artificial vegetation restoration; areas with decreased area and irregular boundaries may correspond to vegetation degradation, and the specific reasons (such as natural erosion or human damage) need further confirmation through on-site verification; areas with offset boundaries may be due to construction disturbance. Fusion anomaly areas need to be marked for subsequent on-site verification.

[0145] In this embodiment, the binarized vegetation distribution map refers to a binary image in which vegetated areas are marked as 1 and non-vegetated areas as 0. Binarization is a key step in converting continuously changing vegetation indices into discrete vegetation / non-vegetation classifications.

[0146] In this embodiment, morphological processing refers to mathematical operations based on image morphology, including dilation, erosion, opening, and closing operations. Dilation can expand the boundary of vegetation areas outward, filling small holes inside the areas; erosion can shrink the boundary of vegetation areas inward, removing isolated noise points.

[0147] In this embodiment, the predefined structural element refers to a template of fixed shape and size used in morphological processing. The choice of structural element affects the effect of morphological processing; larger structural elements can eliminate more noise, but may also make the vegetation outline overly smooth; smaller structural elements can retain more detail, but the noise reduction effect is limited.

[0148] In this embodiment, the contour difference region refers to the spatial region where the vegetation contour lines differ between the two binary vegetation distribution maps. The contour difference region is the basis for analyzing vegetation change categories and identifying fusion anomaly regions.

[0149] In this embodiment, the data storage and retrieval process for each step includes: Data format: The binary vegetation distribution map is binary raster data; the vegetation outline is vector line data; the outline difference area is vector surface data, with attached area and morphological attribute.

[0150] Storage method: The binary vegetation distribution map is stored in a distributed file system; the vegetation outline and the outline difference region are stored in a spatial database.

[0151] Reading process: The central server cluster reads the first and second region images from the distributed file system; calculates NDVI and applies pixel criteria to generate a binary vegetation distribution map; applies morphological processing to extend and enhance the vegetation outline; overlays the outline onto the georegistration base map; analyzes the area characteristics and morphological properties of the outline difference areas; determines the vegetation change category based on morphological properties and marks fusion anomaly areas.

[0152] The main technical contribution of this embodiment lies in solving the technical problem of automatic identification of vegetation change categories: by using binarization processing to convert continuously changing vegetation indices into discrete vegetation / non-vegetation classifications, the complexity of the analysis is reduced; by using morphological processing to expand and enhance vegetation outlines, noise interference is eliminated, making the outlines clearer and more continuous; by analyzing the area characteristics and morphological properties of the outline difference areas, the automatic determination of vegetation change categories and the automatic marking of abnormal areas are realized, providing accurate target areas for subsequent on-site verification.

[0153] Example 9: Based on Example 1, this example further defines the specific implementation method of fusing the second region image and the first region image.

[0154] The second region image is fused with the first region image, including the following sub-steps: Using the tower coordinates as the reference point, a geometrically registered vector layer of the vegetation restoration area is constructed. Geometric registration refers to the precise spatial alignment of the second region image and the first region image within the same tower coordinate system. Using the tower coordinates as the reference point and the precise GPS coordinates of the tower as control points, geometric correction and registration are performed on the two images to ensure that pixels at the same geographical location correspond precisely in the two images.

[0155] A vegetation restoration area vector layer is a layer that extracts the spatial distribution information of vegetation restoration areas from geometrically registered image data and represents it in vector format. The vegetation restoration area vector layer contains information such as the spatial location, boundary shape, and area of ​​each vegetation restoration patch.

[0156] Based on the vector layers, calculation channels corresponding to different image deviation indices are constructed. These indices include: vegetation cover deviation, NDVI deviation, chlorophyll content deviation, and vegetation area deviation. Each deviation index corresponds to an independent calculation channel.

[0157] A computational channel refers to the computational process and data processing path designed for a specific deviation index. Each computational channel includes: data input, deviation calculation method, and deviation output. By constructing multiple parallel computational channels, various image deviation indices can be calculated simultaneously, enabling multi-dimensional and comprehensive vegetation change analysis. The calculation results of different channels are independent and do not interfere with each other, facilitating subsequent comprehensive analysis and cross-validation.

[0158] In this embodiment, geometric registration refers to the process of spatially aligning image data acquired from different time phases and different sensors within the same geographic coordinate system. The accuracy of geometric registration directly affects the accuracy of pixel-level comparison. Using tower coordinates as the reference point for geometric registration fully leverages the advantage of transmission line towers as inherent spatial nodes, enabling high-precision registration.

[0159] In this embodiment, the vegetation restoration area vector layer refers to a spatial distribution layer of the vegetation restoration area represented in vector format. Vector layers have the advantages of precise positioning and attribute storage, facilitating subsequent spatial analysis and decision support.

[0160] In this embodiment, a calculation channel refers to an independent calculation process designed for a specific image deviation index. Each calculation channel includes four stages: data reading, index calculation, deviation analysis, and result output. Multi-channel parallel computing can fully utilize the parallel computing capabilities of the central server cluster and improve processing efficiency.

[0161] In this embodiment, the data storage and retrieval process for each step includes: Data format: Geometric registration parameters are affine transformation matrices or polynomial transformation coefficients; the vector layer of the vegetation restoration area is in Shapefile or GeoJSON format; the input of each calculation channel is a specific band or index of the two phases of images, and the output is a deviation raster map and statistical reports.

[0162] Storage method: Geometric registration parameters are stored in a relational database and associated with tower numbers; the vector layer of vegetation restoration area is stored in a spatial database; the input data and output results of each calculation channel are stored in a distributed file system, with subdirectories created according to channel type.

[0163] Reading process: The central server cluster reads geometric registration parameters and coordinates of each tower from the database; performs geometric registration of the two phases of images based on the tower coordinates; extracts vegetation restoration areas from the registered images and generates a vector layer; based on the vector layer, starts multiple calculation channels to calculate various deviation indicators in parallel; summarizes the calculation results of each channel and generates a comprehensive monitoring report.

[0164] The main technical contribution of this embodiment is to solve the technical problems of parallel computing of multiple indicators and result fusion: by constructing a geometrically registered vector layer of vegetation restoration area with tower coordinates as the reference point, the accurate correspondence of multi-source data under the same spatial reference frame is ensured; by constructing multiple computing channels corresponding to different image deviation indicators, parallel computing and data isolation of multi-dimensional deviation indicators are realized, thereby improving computing efficiency and the reliability of results.

[0165] Example 10: Based on Example 1, this example further defines the specific equipment type and closed-loop feedback mechanism of the system.

[0166] The image acquisition equipment is a multispectral UAV or a fixed-point long-term image monitoring device. The tower sensor group includes soil moisture sensors, runoff velocity sensors, and rainfall sensors.

[0167] Edge computing terminals are installed at substations or inspection stations along the power line. The central server cluster is deployed in the power grid company's data center.

[0168] The method further includes: Based on plant status trends, plant distribution trends, and image deviations, an environmental and water conservation monitoring report is generated. The report includes the following: monitoring period, monitoring area, monitoring values ​​for each environmental and water conservation indicator, plant status trend analysis, plant distribution trend analysis, image deviation statistics, a list of suspected problem areas, and rectification suggestions. The monitoring report is generated in PDF or Word format and can be viewed and archived by all participating units.

[0169] In this embodiment, the plant state deviation amplitude refers to the quantified value of plant state changes determined based on image deviation, such as vegetation cover deviation, leaf chlorophyll content deviation, etc. This deviation amplitude is used to characterize the strength of plant state trends.

[0170] When the deviation of plant status determined based on the plant status trend exceeds a preset alarm threshold, or when the percentage of pixels exhibiting a decaying distribution trend in the plant distribution trend exceeds a preset alarm threshold, a data acquisition frequency adjustment command is sent to the edge computing terminal. The preset alarm threshold can be set according to environmental and water conservation management requirements. For example, an alarm is triggered when the deviation of plant status (such as a decrease in vegetation coverage) decreases by more than 10%, or when the percentage of pixels exhibiting a decaying distribution trend in plants exceeds 20%.

[0171] The data acquisition frequency adjustment command refers to the instruction sent by the central server cluster to the edge computing terminal to adjust the data acquisition frequency of the image acquisition device. When an alarm is triggered, the central server cluster can instruct the edge computing terminal to increase the image acquisition frequency to enhance monitoring of the problem area; when the environmental and water conservation situation returns to normal, it can instruct to reduce the acquisition frequency to save resources.

[0172] In this embodiment, the environmental and water conservation monitoring report is a decision support document for environmental and water conservation managers. The monitoring report presents monitoring results and trend analysis in a visual manner, making it easy for managers to quickly grasp the status of environmental and water conservation.

[0173] In this embodiment, the preset alarm threshold is a pre-set critical value of the environmental and water conservation index used to trigger an alarm. The setting of the alarm threshold needs to comprehensively consider factors such as environmental and water conservation management requirements, monitoring accuracy, and false alarm rate.

[0174] In this embodiment, the data acquisition frequency adjustment command is a control command issued by the central server cluster to the edge computing terminal, used to dynamically adjust the data acquisition frequency. This closed-loop feedback mechanism realizes a complete closed loop of "monitoring, analysis, alarm, and response," improving the intelligence level of environmental and water conservation monitoring.

[0175] In this embodiment, the data storage and retrieval process for each step includes: Data format: Environmental and water conservation monitoring reports are PDF or Word documents; preset alarm thresholds are numerical parameters stored in configuration files; data acquisition frequency adjustment commands are structured commands.

[0176] Storage method: Environmental protection and water conservation monitoring reports are stored in the "Monitoring Reports" directory of the distributed file system; preset alarm thresholds are stored in a relational database; data acquisition frequency adjustment instructions are transmitted through a message queue and recorded in the operation log.

[0177] Reading process: The central server cluster generates a monitoring report based on plant status trends, plant distribution trends, and image deviations, and stores it in the file system; the trend data is compared with a preset alarm threshold, and if the threshold is exceeded, an alarm message is generated; a data acquisition frequency adjustment instruction is sent to the edge computing terminal through a message queue; after receiving the instruction, the edge computing terminal adjusts the data acquisition frequency of the image acquisition device.

[0178] The main technical contribution of this embodiment is to realize closed-loop intelligent management of environmental and water conservation monitoring: by generating environmental and water conservation monitoring reports, the monitoring results are presented to the management personnel in a visual manner; by setting alarm thresholds, the abnormal situation is automatically identified and alarmed; by sending data acquisition frequency adjustment instructions to the edge computing terminal, the monitoring resources are dynamically allocated on demand, forming a complete intelligent monitoring closed loop of "monitoring, analysis, alarm, and response".

[0179] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for monitoring environmental water conservation in power transmission and transformation lines based on image fusion, characterized in that, The system is applied to the environmental and water conservation monitoring system for power transmission and transformation lines, and includes: The image acquisition device is configured to connect to the edge computing terminal via a wireless data transmission link, and is used to acquire multispectral image data of the monitoring area during the monitoring period. The tower sensor array is configured to interconnect with the edge computing terminal via an Internet of Things communication protocol to collect sensing data based on the tower coordinate system; The edge computing terminal is configured to connect to the central server cluster via a wired or wireless network, and is used to perform denoising and coordinate correction on the image data; The central server cluster is configured to be interconnected with the storage array via a high-speed data bus for performing cross-time period image fusion and trend determination. The method includes the following steps: S1: Acquire image data of the monitoring area within the monitoring period and acquire sensing data based on the tower coordinate system; wherein, the monitoring period includes the first monitoring period and the second monitoring period; the tower coordinate system is composed of the above-ground vegetation distribution network and the groundwater variation network, and the above-ground vegetation distribution network and the groundwater variation network are coupled. S2: Based on the first monitoring time period, determine the first index value corresponding to at least one environmental and water conservation index associated with vegetation status, and extract the first region image of the first index region to which the first index value belongs; wherein, the image data is used to determine the plant status characteristics corresponding to the first index value and to determine the plant distribution characteristics corresponding to the first index region, and the sensing data is the hydrological and soil characteristics that cause changes in plant status characteristics and plant distribution characteristics; at the same time, based on the second monitoring time period, extract the second region image corresponding to the environmental and water conservation index; S3: Perform spatiotemporal alignment before image fusion, and set spatial reference parameters using the positioning data of the tower coordinate system; based on the spatial reference parameters, compare the second region image with the first region image to determine the pixel deviation; S4: Based on the pixel deviation, the second region image corresponding to the environmental and water conservation index within the second monitoring time period is fused with the first region image to generate a fusion difference map. The image deviation is determined based on the fusion difference map, and the plant status trend and plant distribution trend are determined based on the image deviation.

2. The method for monitoring environmental protection and water conservation of power transmission and transformation lines based on image fusion according to claim 1, characterized in that, The image data includes satellite remote sensing image data and UAV aerial image data; the satellite remote sensing image data is acquired by the satellite ground receiving station and received through the central server cluster. The monitoring method employs a three-tiered, progressive regulatory framework to oversee the monitored area: Level 1: Conduct a comprehensive survey of the monitoring area using satellite remote sensing to obtain satellite remote sensing images of the entire area and identify areas with suspected environmental and water conservation problems; Level 2: Based on the suspected environmental and water conservation problem areas, conduct detailed investigations of key areas using drones to obtain drone aerial images of key areas and confirm whether environmental and water conservation problems exist in the suspected problem areas. Level 3: Based on the areas identified as having environmental and water conservation issues in Level 2, supplementary verification of these areas will be conducted through on-site inspections. The scope of the detailed investigation of key areas at the second level is dynamically triggered and determined by the results of the overall survey at the first level, while the scope of the on-site verification at the third level is dynamically triggered and determined by the results of the detailed investigation at the second level.

3. The method for monitoring environmental water conservation in power transmission and transformation lines based on image fusion according to claim 1, characterized in that, Acquire image data and perception data, including: During the monitoring period, a reference frame is acquired. The reference frame is a target frame image of the monitoring area that has no missing features during the first monitoring period. The target frame image is configured as the starting frame within the monitoring time period and is used as an input frame for subsequent processing. When the target frame image is used as an input frame, the quantitative data of the starting frame is used as a reference, and incremental data is superimposed to obtain the forward monitoring frame image of the monitoring area. The incremental data is the measured change of environmental and water conservation indicators between two adjacent frames. The first monitoring time period includes a starting frame and at least one forward monitoring frame image representing the real-time value of at least one environmental and water conservation index. The forward monitoring frame images are arranged in chronological order, and the last forward monitoring frame image in the first monitoring time period is the initial frame image of the environmental and water conservation index corresponding to the second monitoring time period.

4. The method for monitoring environmental water conservation in power transmission and transformation lines based on image fusion according to claim 3, characterized in that, During image data acquisition, the calibrated reflectivity is determined based on the target frame image; When a forward monitoring frame image is received, it is determined whether the real-time reflectance of the forward monitoring frame image conforms to the calibrated reflectance. Among them, when the difference between the real-time reflectance and the calibrated reflectance is within the preset threshold range, it is determined to meet the calibrated reflectance, and the forward monitoring frame images are arranged in sequence. When the difference between the real-time reflectance and the calibrated reflectance exceeds the preset threshold range, the CMOS gain and aperture value of the image acquisition device for the forward monitoring frame image are adjusted according to the RGB mean value of the target frame image, and the image is re-acquired until the difference between the brightness value of the re-acquired forward monitoring frame image and the brightness value of the target frame image is within the preset brightness tolerance range. Then, the forward monitoring frame images are arranged sequentially.

5. The method for monitoring environmental protection and water conservation of power transmission lines based on image fusion according to claim 1, characterized in that, The first indicator area is defined, including: When the edge computing terminal receives the image data, it divides the image into 200×200 pixel grid units, and each grid unit is superimposed with the sensing data at the corresponding location to generate a feature matrix. Based on a preset vegetation coverage threshold, vegetation areas are determined in each grid cell; When the sensing data of consecutive adjacent grid cells can all be used to calculate the first index value, the corresponding grid cell is classified as the first index region; wherein, the first index value is the value of the environmental and water conservation index associated with the vegetation status.

6. The method for monitoring environmental water conservation in power transmission and transformation lines based on image fusion according to claim 1, characterized in that, The first region image is enhanced by filtering with the first filtering band to form a first bandpass filtered image, and then filtered with the second filtering band to form a second bandpass filtered image. Based on the response difference between the first bandpass filtered image and the second bandpass filtered image, the visible adjacency boundary between the vegetated area and the non-vegetated area is extracted.

7. The method for monitoring environmental water conservation in power transmission and transformation lines based on image fusion according to claim 1, characterized in that, The specific steps involved in performing spatiotemporal alignment and determining pixel offsets in S3 are as follows: Spatial reference parameters are set using the positioning data from the tower coordinate system; Based on spatial reference parameters, the vegetation features of the second region image and the first region image are compared pixel by pixel to determine the pixel deviation. Based on pixel deviation, determine the trend of plant status and plant distribution: for the same pixel location, determine the trend of plant status improvement or deterioration based on the positive or negative value of the vegetation index difference; determine the trend of plant distribution expansion or contraction based on the increase or decrease of the number of pixels in the vegetation coverage area.

8. The method for monitoring environmental protection and water conservation of power transmission lines based on image fusion according to claim 1, characterized in that, Before image fusion, the following steps are also included: Binarized vegetation distribution maps of the first and second region images are determined respectively, and pixel determination criteria for vegetation regions are set. Based on the pixel determination criteria, morphological processing is used to extend the vegetation outline on the binary vegetation distribution map, and predefined structural elements are used to enhance the outline. The enhanced outlines are overlaid onto the georeferencing base map to analyze the area characteristics and morphological properties of the regions with outline differences. The vegetation change category is determined based on the differences in morphological characteristics, and the fusion anomaly area is determined based on the vegetation change category.

9. The method for monitoring environmental water conservation in power transmission and transformation lines based on image fusion according to claim 1, characterized in that, The second region image is fused with the first region image, including: Using the tower coordinates as the reference point, a geometrically registered vector layer of the vegetation restoration area is constructed. Based on the vector layers, calculation channels corresponding to different image deviation indices are constructed.

10. The method for monitoring environmental water conservation in power transmission and transformation lines based on image fusion according to claim 1, characterized in that, The image acquisition equipment is a multispectral drone or a fixed-point long-term image monitoring device, and the tower sensor group includes a soil moisture sensor, a runoff velocity sensor, and a rainfall sensor. Edge computing terminals are installed in substations or inspection stations along the line, while the central server cluster is deployed in the power grid company's data center. The method further includes: Based on plant status trends, plant distribution trends, and image deviations, an environmental and water conservation monitoring report is generated. When the deviation of the plant state determined according to the plant state trend exceeds the preset alarm threshold, or when the proportion of pixels with the attenuation distribution trend in the plant distribution trend exceeds the preset alarm threshold, a data acquisition frequency adjustment command is sent to the edge computing terminal.

Images