Multi-source fusion vegetation ecological monitoring method and system

Abstract

The present application belongs to the technical field of vegetation ecological monitoring, and particularly relates to a multi-source fusion vegetation ecological monitoring method and system, which comprises the following steps: synchronously collecting environmental data and visual data in a monitoring area; calculating fluctuation amplitude of the environmental data in a time sliding window, and constructing an environmental fluctuation potential energy index for representing instability of an external physical environment based on the fluctuation amplitude; calculating difference between a current frame image and a reference frame image according to the visual data, and constructing a visual saliency difference entropy in combination with image brightness information; performing weighted calculation on the visual saliency difference entropy by using the environmental fluctuation potential energy index, generating a self-adaptive return priority coefficient, and executing a corresponding data transmission strategy. The present application suppresses invalid visual changes by physical environmental fluctuation, effectively reduces the number of invalid communications, and realizes low-power consumption monitoring under normal conditions and reliable early warning under extreme disasters.

Description

Technical Field

[0001] This invention relates to the field of vegetation ecological monitoring technology. More specifically, this invention relates to a multi-source fusion method and system for vegetation ecological monitoring. Background Technology

[0002] When conducting ecological monitoring of vegetation in remote areas, eco-towns, or large forestry bases, the primary reliance is on monitoring systems composed of image sensor nodes deployed in the field. Since these sensor nodes are deployed outdoors, they mostly rely on wireless networks for data transmission, placing stringent requirements on power consumption and bandwidth.

[0003] Currently, existing ecological monitoring systems typically employ two data transmission modes: one is timed full transmission, which involves uploading high-definition images and sensor data at a fixed frequency without differentiation; the other is transmission triggered by simple visual motion detection.

[0004] However, existing monitoring systems have significant limitations in practical applications: Firstly, in the context of wireless networks in the wild, bandwidth resources are scarce and most devices are powered by solar energy. The timed full data transmissions contain a large amount of meaningless static images, and transmissions triggered by simple visual motion detection cannot effectively distinguish between phenomena such as leaves swaying in the wind and real biological invasions, which can easily lead to false alarms. Secondly, existing monitoring systems rely solely on visual information for judgment and lack a comprehensive perception of environmental dimensions. When encountering extreme weather such as typhoons or heavy rain, the system may stop working due to excessive environmental interference or fail to identify the true disaster characteristics, which can easily lead to the loss of critical disaster data and fail to meet the needs of all-weather ecological monitoring. Summary of the Invention

[0005] To address the problems of high energy consumption, frequent false alarms, and failure under extreme weather conditions in monitoring systems, this invention proposes a multi-source fusion vegetation ecological monitoring method and system. By introducing environmental physical quantities as a reference for visual judgment, an interference-resistant transmission evaluation system is constructed.

[0006] In a first aspect, the present invention provides a multi-source fusion vegetation ecological monitoring method, comprising: synchronously collecting environmental data and visual data within a monitoring area, and updating the environmental data and visual data in real time based on a preset time sliding window; calculating the fluctuation amplitude of the environmental data within the time sliding window, and constructing an environmental fluctuation potential energy index to characterize the instability of the external physical environment based on the fluctuation amplitude; calculating the difference between the current frame image and the reference frame image based on the visual data, and constructing a visual saliency difference entropy to characterize the drastic degree of image content change by combining image brightness information; weighting the visual saliency difference entropy using the environmental fluctuation potential energy index, introducing a disaster compensation mechanism, generating an adaptive backhaul priority coefficient, and executing a corresponding data transmission strategy based on the adaptive backhaul priority coefficient.

[0007] By adopting the above technical solution, and by simultaneously collecting environmental and visual data and introducing a time sliding window mechanism, the collaborative alignment of multi-source heterogeneous data in the time dimension is ensured. The instability of the physical environment is transformed into a dimensionless quantitative benchmark by utilizing the environmental fluctuation potential energy index. Combined with the visual saliency difference entropy with brightness adaptive characteristics, the impact of natural background disturbances and illumination changes on monitoring results is effectively overcome. The adaptive backhaul priority coefficient realizes the dynamic suppression and disaster compensation of visual judgment by environmental physical indicators. While significantly reducing the false alarm rate and communication energy consumption of low-power devices in the field, it ensures the forced backhaul of key data in extreme disaster scenarios, realizing all-weather and highly reliable intelligent monitoring of vegetation ecology.

[0008] Preferably, the environmental data includes vibration data of vegetation and wind speed data. The synchronous acquisition of environmental and visual data within the monitoring area includes: acquiring tree vibration data through accelerometers installed on the vegetation and acquiring wind speed data through an anemometer; acquiring real-time video streams through deployed low-power cameras, extracting the current frame image and converting it into a single-channel grayscale image as the visual data; establishing a time sliding window of a preset length in the system memory and updating the environmental data at a preset sampling rate.

[0009] By adopting the above technical solution, accelerometers and anemometers installed on vegetation are used to synchronously acquire tree vibration and wind speed data. Combined with real-time video streams collected by low-power cameras, a comprehensive perception of the physical environment and visual information is achieved. Single-channel grayscale conversion effectively eliminates color information interference, allowing the processing focus to be on brightness and contrast features to reduce data transmission volume. The time sliding window mechanism maintained in memory ensures accurate temporal alignment and real-time updates of multi-source heterogeneous data, providing an accurate and statistically significant data foundation for subsequent feature fusion analysis.

[0010] Preferably, the environmental fluctuation potential energy index satisfies the following relationship:

[0011]

[0012] in, This represents an indicator of the potential energy of environmental fluctuations. This indicates the total number of sensor types. Indicates the first Sensor-like devices at the current moment Real-time sampled values, Indicates the first The historical average value of the sensor within the time sliding window, Indicates the first Sensitivity weighting coefficients for sensor-like devices This represents the environmental gain coefficient.

[0013] By adopting the above technical solution, an environmental fluctuation potential energy index is constructed, which transforms the physical environmental fluctuations sensed by sensors into dimensionless values. This enables accurate assessment of the physical instability of the current environment. At the same time, the logarithmic function is used to handle huge numerical ranges, avoiding numerical explosion, and providing a unified evaluation benchmark for subsequent fusion calculations with visual data.

[0014] Preferably, the visual saliency difference entropy satisfies the following relationship:

[0015]

[0016] in, Entropy represents the visual saliency difference. Represents the gray level in the current frame image histogram. The probability of occurrence Represents the gray level in the histogram of the reference frame image. The probability of occurrence This represents the average brightness value of the current frame image. This represents the brightness overflow prevention constant. This represents the visual magnification factor.

[0017] By adopting the above technical solution, calculating the visual saliency difference entropy, using histogram difference instead of simple pixel subtraction, and introducing a brightness adaptive factor, the system can effectively cope with the changes in illumination caused by cloud cover and the low-light environment at night. This makes the system more sensitive to changes in low light and more resistant to noise in strong light, thus improving the robustness of the system.

[0018] Preferably, the adaptive backhaul priority coefficient satisfies the following relationship:

[0019]

[0020] in, This represents the adaptive backhaul priority coefficient. Entropy represents the visual saliency difference. This represents an indicator of the potential energy of environmental fluctuations. Indicates the visual nonlinearity index. Indicates the disaster compensation weight. This indicates the threshold for extreme environments.

[0021] By adopting the above technical solution, an adaptive feedback priority coefficient is generated. The entropy of visual significance difference is suppressed by division using the environmental fluctuation potential energy index. This effectively filters out invalid visual changes caused by natural wind. At the same time, by introducing a disaster compensation mechanism, the priority is forcibly increased when environmental fluctuations exceed extreme thresholds. This solves the problems of false alarms and energy consumption under normal conditions and prevents missed alarms under extreme disasters, achieving intelligent full-scenario coverage.

[0022] Preferably, the sensitivity weighting coefficient of the vibration data is greater than the sensitivity weighting coefficient of the wind speed data.

[0023] Preferably, the average brightness value of the current frame image is obtained by calculating the average of the grayscale values ​​of all pixels in the current frame image.

[0024] Preferably, the step of executing the corresponding data transmission strategy according to the adaptive backhaul priority coefficient includes: determining whether the adaptive backhaul priority coefficient is greater than a preset priority threshold; if the adaptive backhaul priority coefficient is greater than the preset priority threshold, it is determined to be a high-priority mode, and the current original image and environmental data are uploaded through the wireless transmission module; if the adaptive backhaul priority coefficient is less than or equal to the preset priority threshold, it is determined to be a low-priority mode, and a log is recorded locally and the system enters a sleep state.

[0025] Preferably, the scenarios corresponding to the high-priority mode include biological invasion events or severe natural disaster events, and the scenarios corresponding to the low-priority mode include natural background disturbances.

[0026] Secondly, the present invention provides a multi-source fusion vegetation ecological monitoring system, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned multi-source fusion vegetation ecological monitoring method is implemented.

[0027] By adopting the above technical solution, a computer program is generated from the above-mentioned multi-source fusion vegetation ecological monitoring method and stored in a memory so that it can be loaded and executed by a processor. A terminal device can then be made based on the memory and processor for convenient use.

[0028] The beneficial effects of this invention are as follows:

[0029] This invention ensures the coordinated alignment of multi-source heterogeneous data across time by simultaneously acquiring environmental and visual data and introducing a sliding window. It constructs an environmental fluctuation potential energy index to transform physical fluctuations into dimensionless values, utilizing a logarithmic function to handle large numerical ranges and avoid numerical explosion, providing a unified benchmark for fusion computing. Simultaneously, it calculates the visual saliency difference entropy and introduces a brightness adaptive factor to effectively address changes in illumination, making the system more sensitive in low light and more noise-resistant in strong light. Finally, it uses physical indices to suppress visual indices through division, effectively filtering out invalid changes caused by natural wind forces. Combined with a disaster compensation mechanism, it reduces false alarms and power consumption while ensuring mandatory alarms under extreme disasters, achieving intelligent full-scene coverage.

[0030] Furthermore, by acquiring tree vibration and wind speed data through accelerometers and anemometers installed on vegetation, and combining this with single-channel grayscale images captured by a low-power camera, the system focuses on brightness and contrast features while eliminating color interference, thus reducing signal transmission and memory usage. This invention, by setting differentiated sensitivity weights, visual nonlinearity exponents, and priority thresholds, enables the system to accurately identify high-priority events such as biological invasions or natural disasters and upload raw data, while remaining in a dormant state in low-priority scenarios, significantly improving the operational efficiency and endurance of the ecological monitoring system. Attached Figure Description

[0031] Figure 1 This is a flowchart of a multi-source fusion vegetation ecological monitoring method according to the present invention;

[0032] Figure 2 This is a schematic diagram of the real-time monitoring curve of multi-source feature data of the present invention;

[0033] Figure 3 This is a schematic diagram comparing the return method of the present invention with the return method of the prior art. Detailed Implementation

[0034] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0035] This invention discloses a multi-source fusion vegetation ecological monitoring method, referring to... Figure 1 This includes steps S1-S4:

[0036] S1. Synchronously collect environmental and visual data within the monitoring area, and update the environmental and visual data in real time based on a preset time sliding window.

[0037] In one optional embodiment, monitoring terminals are first deployed in remote areas, eco-towns, or large forestry bases, and data is collected by these terminals. In one specific embodiment, the monitoring terminal includes devices such as a three-axis accelerometer, a miniature anemometer, and a low-power camera.

[0038] In one specific implementation, two types of data need to be collected simultaneously through a hardware synchronous triggering mechanism: environmental data and visual data.

[0039] Specifically, when environmental data is needed, tree vibration data is monitored using deployed triaxial accelerometers, and wind speed data is monitored using deployed miniature anemometers. Both the triaxial accelerometers and the miniature anemometers collect data at a frequency of 10 Hz, and the data is recorded... Time of the first The instantaneous value of the sensor It should be noted that the dimensions of the data collected by the triaxial accelerometer and the miniature anemometer are inconsistent, so they cannot be directly fused. The present invention will address the fluctuations in the data collected by the triaxial accelerometer and the miniature anemometer.

[0040] When visual data is needed, real-time video streams are captured at a low frame rate of 1fps using deployed low-power cameras. It's important to note that after the low-power cameras acquire data, the system needs to extract the image from the current frame. It converts the current frame image to a single-channel grayscale image and scales the resolution to [resolution value missing]. .

[0041] Since mobile monitoring in vegetation ecology relies mainly on changes in light and shadow and contours rather than color, converting the current frame image into a single-channel grayscale image can eliminate interference from chromaticity information, allowing the processed information to focus on brightness and contrast features, while reducing the amount of data during signal transmission.

[0042] By scaling the resolution of the current frame image to This can reduce the total number of pixels in each frame of an image, thereby reducing memory usage when processing the current frame image later, while ensuring the statistical validity of the image data.

[0043] While collecting data through the monitoring terminal, in order to ensure the real-time performance and accuracy of the system's data processing, it is necessary to maintain a memory with a length of [missing information]. A time-sliding window. For example, setting a time-sliding window. The interval is 10 seconds, which means that the system memory always retains environmental and visual data from the last 10 seconds, thus providing a real-time statistical basis for subsequent calculations.

[0044] In this way, by using hardware synchronization triggering and sliding window mechanisms, the alignment of multi-source heterogeneous data in time can be ensured, providing an accurate data foundation for subsequent feature fusion analysis.

[0045] S2. Calculate the fluctuation amplitude of environmental data within a time sliding window, and construct an environmental fluctuation potential energy index based on the fluctuation amplitude to characterize the instability of the external physical environment.

[0046] In an optional embodiment, changes in the physical morphology of vegetation are mostly driven by physical environmental factors such as wind. Therefore, by calculating the fluctuation amplitude of environmental data, the instantaneous deviation of the current environment from a stable state can be extracted. Specifically, this embodiment characterizes the fluctuation amplitude of environmental data by calculating the fluctuation variance, and then maps the fluctuation variance of environmental data into a dimensionless environmental fluctuation potential energy index using a logarithmic function. This provides a unified standard for subsequent fusion calculations. The significance of using a logarithmic function to process the variance of environmental data lies in its ability to compress the variance into a linear evaluation interval, thus preventing some data points from exhibiting excessively high variance. The calculation method for the environmental fluctuation potential energy index is as follows:

[0047]

[0048] in, Indicators representing the potential energy of environmental fluctuations; In this invention, the total number of sensor types connected to the monitoring system is [number]. The number is 2, meaning that this invention only has two types of sensors for monitoring wind speed data and vibration data; Indicates the first Sensor-like devices at the current moment Real-time sampled values; Indicates the first Sensor-like sliding window Historical average within the period; It represents the degree of deviation of the environmental parameters at the current moment from the average state, i.e., instantaneous fluctuation; Indicates the first The sensitivity weighting coefficient for this type of sensor is set to weight the wind speed data because vibration values ​​are typically small. The value is 1, which is used to weight the vibration data. The value is 5; the value 1 serves as the baseline compensation for the logarithmic function. This represents the environmental gain coefficient. In one specific implementation, Take the value 1.

[0049] It is important to note that The method of obtaining it is:

[0050]

[0051] in, This represents the number of sampling points within the window.

[0052] To more clearly illustrate the role and calculation process of the environmental fluctuation potential energy index, the following example demonstrates its calculation:

[0053] First, let's assume at a certain moment... :

[0054] Data monitored by wind speed sensor Historical average Weight ,but ;

[0055] Data monitored by vibration sensors Historical average Weight ,but ;

[0056] Therefore, the total fluctuation and ;

[0057] Environmental fluctuation potential energy index ;

[0058] Assuming a strong wind occurs at this time, and the wind speed fluctuation term surges to 20, and the vibration fluctuation term surges to 5, then the environmental fluctuation potential energy index... .

[0059] Through the above comparison, even when encountering strong winds and a surge in speed and vibration fluctuation terms, the environmental fluctuation potential energy index is still compressed within a controllable range through logarithmic processing.

[0060] Thus, by constructing an environmental fluctuation potential energy index, environmental data with different physical dimensions can be uniformly mapped into a dimensionless index characterizing environmental instability, providing a standardized input for subsequent cross-modal fusion.

[0061] S3. Calculate the difference between the current frame image and the reference frame image based on visual data, and construct a visual saliency difference entropy to characterize the degree of change in image content by combining image brightness information.

[0062] In an optional embodiment, when cloud cover obscures the image, the brightness of all pixels in the current frame acquired by the system will be affected. Since traditional pixel subtraction is extremely sensitive to absolute brightness values, performing pixel subtraction on this basis will lead to data distortion and false alarms. Therefore, in this embodiment, difference entropy based on a grayscale histogram is used for calculation, and a brightness adaptive factor is introduced during the calculation process. This makes the system more sensitive to changes in low-light environments and more robust to noise in bright light environments. Visual saliency difference entropy The calculation formula is:

[0063]

[0064] in, Entropy represents visual saliency difference, which is used to characterize the degree of drastic change in image content; Represents the gray level in the current frame image histogram. The probability of occurrence is determined by statistics. medium grayscale value The number of pixels, divided by the total number of pixels; This represents the grayscale probability distribution of the previous reference frame; This represents the average brightness value of the current frame image; This represents the brightness overflow prevention constant. In one specific implementation, The value is 10, which is used to prevent the screen from being completely black. The calculation formula contains a denominator of 0; This represents the visual magnification factor, used to amplify small statistical differences for easier calculation. In one specific implementation, its value is set to 100.

[0065] The value range is from 0 to 255, and it is obtained as follows:

[0066]

[0067] To more clearly illustrate the role and calculation process of visual saliency difference entropy, the following example will demonstrate this:

[0068] Assuming the image resolution is minimal and the histogram difference summation is minimal... ;

[0069] When visual data is acquired during the daytime, if the average brightness at that time... The brightness overflow prevention constant is 190. The visual magnification factor is 10. It is 100;

[0070] but ;

[0071] When visual data is acquired at night, if the average brightness at that time... The value is 10, which is the brightness overflow prevention constant. The visual magnification factor is 10. The value is 100. Assuming that the movement of the object produces the same sum of squared differences in the histogram, that is... ;

[0072] but .

[0073] As can be seen from the comparative examples above, the value of the visual saliency difference entropy calculated under the same physical change is significantly higher at night than during the day, which means that the above algorithm increases the system's sensitivity at night.

[0074] Thus, by calculating the visual saliency difference entropy and combining it with the brightness adaptive mechanism, the impact of illumination changes on visual analysis can be effectively overcome, and the robustness of the system under different illumination conditions can be improved.

[0075] S4. Use the environmental fluctuation potential energy index to calculate the entropy of visual saliency difference, introduce a disaster compensation mechanism, generate an adaptive backhaul priority coefficient, and execute the corresponding data transmission strategy according to the adaptive backhaul priority coefficient.

[0076] In an optional embodiment, a fusion function is constructed using the environmental fluctuation potential energy index. Entropy of visual saliency differences The core logic of the suppression is as follows: when the environmental fluctuation potential energy index is large, it proves that the external environment is fluctuating greatly, which is likely due to weather such as wind. The final score is suppressed by the environmental fluctuation potential energy index, thereby ignoring the visual changes caused by the external environment fluctuation. When the environmental fluctuation potential energy index is so large that it exceeds the extreme threshold, it indicates that a natural disaster has occurred, triggering forced backhaul.

[0077] Adaptive backhaul priority coefficient in this embodiment of the invention The calculation method is as follows:

[0078]

[0079] in, This represents the adaptive backhaul priority coefficient, which the system uses to determine the transmission strategy; a value of 1 is a balance denominator constant, ensuring that the visual significance difference entropy is minimized in windless conditions. It can transmit data without loss, while preventing calculation errors caused by a denominator of 0; The visual nonlinearity index is set to 1.5 in this embodiment of the invention. It is used to nonlinearly amplify visual changes, so that significant changes score higher and small changes score lower. This represents the disaster compensation weight, which is set to 0.5 in this embodiment of the invention; This represents the extreme environment threshold, which is set to 3 in this embodiment of the invention. When the value exceeds 3, the corresponding actual physical wind speed fluctuation may have exceeded 20 m / s, at which point the system will automatically enter disaster mode.

[0080] In one specific implementation, the system automatically calculates the adaptive backhaul priority coefficient once per second and executes the following branch:

[0081] In high-priority mode, when the adaptive backhaul priority coefficient is greater than 1.5, it is determined to be an intrusion event. At this time, the system's wireless transmission module is immediately woken up and the current original high-definition image and environmental data of the previous and next 10 seconds are uploaded.

[0082] In low-priority mode, when the adaptive backhaul priority coefficient is no greater than 1.5, it is determined to be a natural background disturbance. At this time, only local logging is recorded and no wireless transmission is performed. The system automatically enters a sleep power-saving state.

[0083] To more clearly illustrate the role and calculation process of the adaptive backhaul priority coefficient, three specific exemplary scenarios will be used below:

[0084] Scenario 1: Strong wind interference:

[0085] Assuming a strong wind causes leaves to sway violently, the visual salience difference entropy... The highest value is 2, at which point the environmental fluctuation potential energy index is... If it is 2.5, then = ,at this time Values ​​not exceeding 1.5 are considered low priority and not reported back. False alarms were successfully suppressed.

[0086] Scenario 2: Real intrusion but no wind:

[0087] Assuming the environment is windless or has a light breeze, the environmental fluctuation potential energy index The visual significance difference entropy is 0.1. If it is 2, then ,at this time If the value is greater than 1.5, it is determined to be of high priority, and the current original high-definition image and environmental data of the previous and next 10 seconds are uploaded immediately;

[0088] Scenario 3: Extreme weather conditions:

[0089] Assuming a typhoon strikes, the environmental fluctuation potential energy index The maximum value is 5. At this point, the camera may be blurry due to rain, resulting in a significant difference in visual entropy. =1, ,but The value has exceeded 3, so the system automatically enters disaster mode and immediately uploads the current original high-definition image and environmental data for the previous and next 10 seconds.

[0090] In scenario three above, if the external wind force is stronger, for example, the environmental fluctuation potential energy index... If it reaches 6, then At this point, it is directly determined to be of high priority and the current original high-definition image and environmental data of the previous and next 10 seconds are uploaded. This ensures that even if the visual features are not obvious under extreme disaster conditions, the system can still force an alarm.

[0091] In summary, by calculating the adaptive backhaul priority coefficient and executing the strategy, intelligent fusion of multi-source data was achieved, minimizing false alarm rate and communication power consumption while ensuring no disaster monitoring is missed.

[0092] refer to Figure 2 The dashed line, representing the potential energy of environmental fluctuations, shows a significant peak in the 30-60 second range, corresponding to a scenario of strong wind interference in the wild. The value reaches a maximum of over 5.5, indicating that the external environment is extremely harsh. The solid line, representing the entropy of visual significance difference, also rises synchronously in the 30-60 second range due to swaying trees. In the 75-85 second range, it rises independently under stable environmental conditions, corresponding to an abnormal intrusion scenario.

[0093] Reference Figure 3 The dashed line, representing existing technology, has an extremely high detection value in the strong wind range of 30-60 seconds, leading to false alarms. The solid line, representing the technology of this invention, maintains its value consistently below the threshold due to the suppression of environmental fluctuation potential energy indicators, achieving zero false alarms. In the actual intrusion range of 75-85 seconds, the solid line representing the technology of this invention rapidly increases and exceeds the threshold, demonstrating that the system is extremely sensitive to capturing minute visual changes even in the absence of environmental interference.

[0094] This invention also discloses a multi-source fusion vegetation ecological monitoring system, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a multi-source fusion vegetation ecological monitoring method according to the present invention.

[0095] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0096] In the description of this specification, "multiple" or "several" means at least two, such as two, three or more, unless otherwise expressly and specifically defined.

Claims

1. A multi-source fusion method for vegetation ecological monitoring, characterized in that, include: Simultaneously collect environmental and visual data within the monitoring area, and update the environmental and visual data in real time based on a preset time sliding window; The fluctuation amplitude of environmental data within a time sliding window is calculated, and an environmental fluctuation potential energy index is constructed based on the fluctuation amplitude to characterize the instability of the external physical environment. The difference between the current frame image and the reference frame image is calculated based on visual data, and a visual saliency difference entropy is constructed by combining image brightness information to characterize the drastic degree of change in image content. The entropy of visual saliency difference is weighted by the environmental fluctuation potential energy index, and a disaster compensation mechanism is introduced to generate an adaptive backhaul priority coefficient. The corresponding data transmission strategy is then executed based on the adaptive backhaul priority coefficient. The environmental fluctuation potential energy index satisfies the following relationship: , This represents an indicator of the potential energy of environmental fluctuations. This indicates the total number of sensor types. Indicates the first Sensor-like devices at the current moment Real-time sampled values, Indicates the first The historical average value of the sensor within a sliding time window. Indicates the first Sensitivity weighting coefficients for sensor-like devices Indicates the environmental gain coefficient; The entropy of visual saliency differences satisfies the following relationship: , Entropy represents the visual saliency difference. Represents the gray level in the current frame image histogram. The probability of occurrence Represents the gray level in the histogram of the reference frame image. The probability of occurrence This represents the average brightness value of the current frame image. This represents the brightness overflow prevention constant. Indicates the visual magnification factor; The adaptive backhaul priority coefficient satisfies the following relationship: , This represents the adaptive backhaul priority coefficient. Indicates the visual nonlinearity index. Indicates the disaster compensation weight. This indicates the threshold for extreme environments.

2. The multi-source fusion vegetation ecological monitoring method according to claim 1, characterized in that, The environmental data includes vegetation vibration data and wind speed data. Simultaneously collected environmental and visual data within the monitoring area include: Vibration data of trees is collected by accelerometers installed on vegetation, and wind speed data is collected by an anemometer. Real-time video streams are captured by deploying low-power cameras, the current frame image is extracted and converted into a single-channel grayscale image as visual data; A time sliding window of a preset length is created in the system memory, and the environmental data is updated with a preset sampling rate.

3. The multi-source fusion vegetation ecological monitoring method according to claim 2, characterized in that, The sensitivity weighting coefficient for vibration data is greater than that for wind speed data.

4. The multi-source fusion vegetation ecological monitoring method according to claim 1, characterized in that, The average brightness value of the current frame image is obtained by calculating the average of the grayscale values ​​of all pixels in the current frame image.

5. The multi-source fusion vegetation ecological monitoring method according to claim 1, characterized in that, The step of executing the corresponding data transmission strategy based on the adaptive backhaul priority coefficient includes: Determine whether the adaptive backhaul priority coefficient is greater than the preset priority threshold; If the adaptive backhaul priority coefficient is greater than the preset priority threshold, it is determined to be in high priority mode, and the current original image and environmental data are uploaded through the wireless transmission module. If the adaptive return priority coefficient is less than or equal to the preset priority threshold, it is determined to be in low priority mode, logs are recorded locally, and the system enters a dormant state.

6. The multi-source fusion vegetation ecological monitoring method according to claim 5, characterized in that, The high-priority mode corresponds to scenarios including biological invasion events or severe natural disasters, while the low-priority mode corresponds to scenarios including natural background disturbances.

7. A multi-source integrated vegetation ecological monitoring system, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement a multi-source fusion vegetation ecological monitoring method according to any one of claims 1-6.

Images