Unmanned aerial vehicle intelligent ecological checking system with multi-source data fusion

Abstract

The invention discloses an unmanned aerial vehicle intelligent ecological checking system with multi-source data fusion, and relates to the technical field of ecological environment protection and monitoring. The system comprises an unmanned aerial vehicle platform, a multi-source data acquisition module, a data processing and analysis module, an intelligent decision module, a communication and data transmission module and a ground control and management platform. According to the invention, by integrating the high-resolution optical camera, the multispectral camera and the thermal imager multi-source data acquisition module, multi-angle and all-directional data acquisition is realized, cooperative work of the devices on the unmanned aerial vehicle platform is combined with an advanced data processing and analysis algorithm, an earth surface change area can be accurately detected, and the change degree and property can be evaluated, so that the unmanned aerial vehicle can be widely applied. Especially, a deep learning algorithm is adopted for feature extraction and classification, and a dynamic threshold change detection algorithm is adopted, so that the accuracy and efficiency of ecological checking are greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of ecological environment protection and monitoring technology, and specifically to an unmanned aerial vehicle intelligent ecological verification system with multi-source data fusion. Background Art

[0002] With the increasing awareness of global ecological and environmental protection, the demand for monitoring and management of nature reserves is becoming increasingly urgent. As an important habitat for biodiversity, real-time monitoring and assessment of the ecological conditions of nature reserves are of great significance for maintaining ecological balance and protecting biodiversity.

[0003] Traditional drone ecological verification technology has the following main shortcomings: First, the data collection method is single, usually relying only on optical cameras, which makes it difficult to obtain multi-dimensional surface information, such as spectral characteristics and thermal radiation characteristics, resulting in incomplete and inaccurate detection of surface changes; second, data processing and analysis capabilities are limited. Traditional methods often lack advanced algorithm support, making it difficult to effectively extract surface features and accurately determine the causes and nature of surface changes; third, the level of intelligence is low. Traditional systems cannot automatically analyze data and generate early warning information, requiring manual intervention, which reduces verification efficiency and response speed; finally, data transmission reliability is poor. Traditional systems are susceptible to interference in complex environments, resulting in data loss or transmission delays, affecting the timeliness and accuracy of verification results.

[0004] To sum up, traditional UAV ecological verification technology can no longer meet the needs of modern ecological environmental protection. In order to overcome these shortcomings, it is particularly important to develop a UAV intelligent ecological verification system with multi-source data fusion. Summary of the Invention

[0005] The purpose of this invention is to make up for the shortcomings of the existing technology and provide a drone intelligent ecological verification system with multi-source data fusion. It can realize multi-angle and all-round data collection by integrating multi-source data acquisition modules such as high-resolution optical cameras, multispectral cameras and thermal imagers. Combined with advanced data processing and analysis algorithms, it can accurately detect surface change areas and evaluate the degree and nature of changes; the built-in intelligent decision-making module can automatically analyze data processing and analysis results, judge the legality of human activities, and automatically generate early warning information.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: an intelligent ecological verification system for drones with multi-source data fusion, the system comprising the following components: a drone platform, a multi-source data acquisition module, a data processing and analysis module, an intelligent decision-making module, a communication and data transmission module, and a ground control and management platform; The drone platform: selects a multi-rotor or fixed-wing drone with long flight time, high stability and high maneuverability, equipped with high-precision GPS and IMU to ensure accurate flight, and installed with obstacle avoidance sensors. Through selection and layout, the drone's flight safety in complex environments is improved; The multi-source data acquisition module integrates a high-resolution optical camera, a multispectral camera, and a thermal imager, and integrates and optimizes the configuration of these devices so that they can work together on the UAV platform to achieve multi-angle and all-round data acquisition, thereby improving the efficiency and quality of data acquisition; The data processing and analysis module performs denoising, enhancement, and geometric correction on optical and multispectral images, uses deep learning algorithms and combines them with the characteristics of human activity patches in nature reserves for targeted training and optimization, implements feature extraction and classification, performs change detection using difference and ratio algorithms, compares current and historical data, detects areas of surface change, and assesses the extent and nature of the change; The intelligent decision-making module establishes a decision-making model based on rules and machine learning. The machine learning part is trained and optimized through a large amount of labeled human activity data. The legality of human activities is judged based on the results of data processing and analysis. If illegal activities are found, early warning information is automatically generated and detailed location and type information is provided, realizing intelligent and automated judgment and early warning. The communication and data transmission module uses 4G, 5G or satellite communication wireless communication technology to achieve real-time data transmission between the UAV and the ground control station. The ground control station can remotely control the UAV flight mission and data acquisition parameters, and receive data for storage and further processing; The ground control and management platform: operators use this platform to plan, monitor and manage drone missions. The platform has data visualization capabilities and optimizes the human-computer interaction interface to make it more in line with the operating habits and needs of nature reserve verification work, thereby improving the work efficiency of operators.

[0007] Furthermore, the obstacle avoidance sensor is a laser radar or a visual sensor, and the flight safety of the UAV in a complex environment is guaranteed by selection and layout. The layout algorithm is: Assume that the coordinate system of the UAV body is The laser radar is installed in front of the UAV at a distance from the center of the fuselage. meters, in The offset on the axis is rice, The offset on the axis is Meters, the visual sensor is installed at the bottom of the drone, horizontal distance from the center of the fuselage m, with The angle between the axis and The axis angle is By collecting and analyzing a large amount of complex terrain environment data, we can determine 、 、 、 、 , and the optimal values ​​of β to achieve all-round monitoring of surrounding obstacles and precise obstacle avoidance, prevent the UAV from colliding with mountains and trees in the nature reserve during flight, and ensure the smooth progress of the data collection task.

[0008] Furthermore, in the multi-source data acquisition module, the high-resolution optical camera, multispectral camera and thermal imager work in coordination using a time synchronization and space calibration algorithm. The time synchronization algorithm is: assuming that the camera 、 、 They are high-resolution optical camera, multispectral camera and thermal imager, with the high-precision clock inside the drone as the reference clock. ,camera The collection time is ,camera The collection time is ,camera The collection time is , by calculating , , using the adjustment factor 、 To the camera and The acquisition time is adjusted by the following formula: ,in and By analyzing the errors of multiple sets of collected data under different lighting conditions and flight speeds, the data collected by the three cameras are accurately synchronized in time to ensure the consistency and relevance of the collected data. The spatial calibration algorithm is to establish a unified geographic coordinate system. , based on the GPS positioning point of the drone, the camera is calculated through trigonometric function relationship 、 、 The geographic coordinate conversion parameters of the collected images ensure that the images collected by different cameras are accurately aligned in geographic space, which facilitates subsequent data fusion and analysis.

[0009] Furthermore, the feature extraction algorithm of the data processing and analysis module is an improved adaptive feature extraction algorithm. , first perform multi-scale decomposition to obtain images at different scales , , then, calculate the gradient magnitude of the image at each scale and gradient direction The formula is: , Then, based on the prior knowledge of human activity patterns in nature reserves, the weight vectors of different features are determined. ,in Indicates the The weight of the feature, is the total number of features, weight By performing principal component analysis and information gain analysis on a large number of labeled nature reserve image samples, the weight vector is determined to highlight the key features related to human activities. Finally, the feature vector is obtained by weighted fusion of gradient features at different scales. , the formula is ,in It is a function that encodes features according to the gradient direction. The algorithm can adaptively extract the features of human activity patches in nature reserves and improve the accuracy and pertinence of feature extraction.

[0010] Furthermore, the change detection algorithm of the data processing and analysis module is a dynamic threshold change detection algorithm. Assume that the current captured image is , the historical reference image is , first calculate the image difference , calculate the dynamic threshold based on the local statistical information of the image , the formula is: in is the image difference In The mean of the neighborhood centered at is the standard deviation, It is a coefficient that is dynamically adjusted according to different nature reserve types and seasonal changes. The determination method is to collect multiple sets of image data of different nature reserves in different seasons, analyze the characteristics and distribution patterns of the changing areas, and use cluster analysis and regression analysis methods to establish The mathematical model of nature reserve type and seasonal factors is When the pixel point is determined to be a change point, the setting of this dynamic threshold can effectively adapt to the changes in the nature reserve environment, accurately detect the real change area of ​​the surface, and reduce misjudgment.

[0011] Furthermore, the decision model based on machine learning of the intelligent decision module is a deep integrated decision forest model, which consists of multiple decision trees. Each decision tree is trained based on different sample subsets and feature subsets. , decision tree De Han Chuwei , which means that in the feature vector Belong to category The final output probability of the model is During the training process, the sample subset is obtained from the training data set by random sampling method, and the feature subset is determined by feature selection algorithm based on information gain and correlation analysis. The information gain calculation method is: Set features The dataset Divided into Part , information gain in It is a dataset The information entropy of is a subset The information entropy of the model is used, and the Pearson correlation coefficient is used for correlation analysis to calculate the correlation between features and target categories, remove features with low correlation, improve the training efficiency and decision-making accuracy of the model, and achieve more accurate judgment on the legality of human activities in nature reserves.

[0012] Furthermore, the communication and data transmission module adopts adaptive modulation and coding technology combined with multipath diversity transmission algorithm. The adaptive modulation and coding technology dynamically adjusts the modulation mode and coding rate according to the channel quality index. Assume that the channel quality index is , the modulation mode set is , the coding rate set is , by looking for pre-established Mapping table with modulation mode and coding rate to determine the optimal modulation mode under current channel conditions and coding rate , the multipath diversity transmission algorithm is: Divide data into multiple sub-data streams , transmitted through different paths, the receiving end uses the maximum ratio combining algorithm to combine the data received by multiple paths, and assumes The signal received by each path is , its gain is , the combined signal is ,In this way, the reliability and anti-interference ability of data transmission are improved,,ensuring that the data collected by the UAV can be transmitted to the ground control station stably and quickly.

[0013] Furthermore, the human-computer interaction interface optimization of the ground control and management platform adopts a personalized interface design method based on user behavior analysis. By collecting the operator's operating behavior data during the use of the platform, including operation frequency, operation path, and stay time, the operators are divided into different user types using a cluster analysis algorithm. For each user type, their operating habits and needs are analyzed, and personalized interface layout and functional module arrangement rules are established. For user types who frequently plan tasks, the task planning module is placed in a prominent position on the interface, and their operating process is optimized. For user types who pay more attention to the analysis of verification results, the data visualization analysis module is highlighted. Through this personalized design, the operator's work efficiency is improved, operational errors are reduced, and the platform is more in line with the usage habits of different users.

[0014] Furthermore, the system also includes a data storage and backup module, which adopts distributed storage and redundant backup strategies. Distributed storage stores the collected data in multiple storage nodes. Suppose the storage node set is According to the type and time attributes of the data, the consistent hashing algorithm is used to distribute the data to different storage nodes. The consistent hashing algorithm ensures the balanced distribution of data among the storage nodes by mapping the data and storage nodes to a hash ring. The redundant backup strategy is: for important data, back up on multiple storage nodes, and the number of backups is dynamically adjusted according to the importance of the data and the system resource situation. In this way, the security and reliability of the data are improved, data loss is prevented, and the data storage needs of long-term monitoring of nature reserves are met.

[0015] Compared with existing technologies, this drone intelligent ecological verification system with multi-source data fusion has the following beneficial effects: First, the system achieves multi-angle and all-round data collection by integrating high-resolution optical cameras, multispectral cameras and thermal imagers. The collaborative work of these devices on the UAV platform, combined with advanced data processing and analysis algorithms, can accurately detect areas of surface changes and assess the extent and nature of changes. In particular, the use of deep learning algorithms for feature extraction and classification, as well as dynamic threshold change detection algorithms, greatly improves the accuracy and efficiency of ecological verification.

[0016] 2. The system uses a built-in intelligent decision-making module to establish a decision-making model based on rules and machine learning, which can automatically analyze data processing and analysis results and judge the legality of human activities. Once illegal activities are detected, the system will automatically generate early warning information and provide detailed location and type information, realizing intelligent and automated judgment and early warning. In addition, the system also uses adaptive modulation and coding technology and multipath diversity transmission algorithm to ensure the reliability and anti-interference ability of data transmission, so that the data collected by the drone can be transmitted to the ground control station stably and quickly, further improving the overall intelligence and automation level of the system.

[0017] Other advantages, objects and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be learned from the practice of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0019] Figure 1 Provide a functional flow chart for a UAV intelligent ecological verification system with multi-source data fusion; Figure 2 This is a flowchart of the overall architecture of a drone intelligent ecological verification system with multi-source data fusion. DETAILED DESCRIPTION

[0020] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0021] Example 1 In a certain mountain nature reserve, its rich mineral resources have attracted the covetousness of some lawless elements, and illegal mining activities occur from time to time, which not only damages the local ecological environment, but also poses a serious threat to the survival of rare animals and plants in the reserve. In order to effectively curb illegal mining activities and protect the ecological balance of the nature reserve, the relevant management departments have activated the drone intelligent ecological verification system for all-round and real-time monitoring.

[0022] Over the mountains of the reserve, the drone platform, equipped with a multi-source data acquisition module, slowly ascended from a temporary take-off and landing point at the edge of the reserve. The drone's GPS and IMU recorded its position and attitude information in real time, and the obstacle avoidance sensor (a laser radar was used, installed in front of the drone at a distance from the center of the fuselage) meters, in The offset on the axis is rice, The offset on the axis is m) to constantly scan the surrounding environment to ensure the drone flies safely in complex mountainous environments and avoid towering mountains and dense trees.

[0023] High-resolution optical cameras, multispectral cameras, and thermal imagers begin working together, using the drone's internal high-precision clock as a reference clock. , assuming that a high-resolution optical camera is the camera 、Multispectral camera is camera 、Thermal imager is camera , by calculating , using the adjustment factor 、 To the camera and Adjust the acquisition time ( ) to achieve data time synchronization, using the GPS positioning point of the drone as a benchmark, and calculating the geographic coordinate conversion parameters through trigonometric function relationships to complete spatial calibration. The cameras quickly shoot at set time intervals to clearly record the surface conditions of the protected area, leaving no corner untouched.

[0024] The collected data is quickly transmitted to the data processing and analysis module. First, the optical images and multispectral images are denoised to remove interference signals caused by light changes and sensor noise factors. Then, an enhancement operation is performed to highlight the key information in the image so that even subtle changes can be clearly presented. Then, geometric correction is performed to ensure the geographic location in the image is accurate.

[0025] Using the improved adaptive feature extraction algorithm, for the input image Perform multi-scale decomposition to obtain images at different scales ( ), calculate the gradient magnitude of the image at each scale , calculate the dynamic threshold based on the local statistical information of the image ( is the image difference in The mean of the neighborhood centered at is the standard deviation, Dynamic adjustments are made based on different nature reserve types and seasonal variations), comparing current and historical data to detect areas of surface change and assessing the extent and nature of the changes.

[0026] The deep integration decision forest model of the intelligent decision module quickly analyzes the received feature vectors. The model consists of multiple decision trees. Each decision tree is trained based on different sample subsets and feature subsets. , decision tree The output is , which means that in the feature vector Belong to category The probability of the model's final output is ,If illegal mining activities are determined, early warning information will be automatically generated and detailed location and type information will be provided.

[0027] The communication and data transmission module adopts satellite communication technology to ensure stable data transmission even in mountainous areas with weak signals. Adaptive modulation and coding technology is used according to channel quality indicators. , from the modulation mode set and coding rate set In the process, the optimal modulation mode under the current channel conditions is determined by searching the mapping table. and coding rate , the multipath diversity transmission algorithm divides the data into multiple sub-data streams , transmitted through different paths, the receiving end uses the maximum ratio combining algorithm (No. The signal received by each path is , its gain is Merge the data.

[0028] At the reserve's management center, operators closely monitor the monitoring situation in front of the ground control and management platform. A personalized interface design method based on user behavior analysis has customized a simple and efficient operation interface for operators who focus on monitoring illegal activities. The alarm information and map positioning modules are placed in a prominent position. When the platform receives an early warning message, a prompt window will quickly pop up, displaying the specific location and related information of the illegal mining. The operator will immediately notify nearby law enforcement officers to go to the scene to deal with it, stop the illegal mining behavior in time, and protect the ecological environment of the nature reserve.

[0029] Example 2 A certain wetland ecosystem is the habitat of many rare birds and a breeding ground for a variety of fish and aquatic plants. Its ecological status is extremely important. With the economic development of the surrounding areas, the impact of human activities on wetlands has gradually increased. In order to timely grasp the changes in the wetland ecological environment and assess the impact of human activities on wetlands, relevant departments use drone intelligent ecological verification systems to conduct regular comprehensive monitoring.

[0030] The drone platform, carrying a multi-source data acquisition module, took off from a dedicated take-off and landing point near the wetland. Before takeoff, the operator carefully checked all the equipment on the drone to ensure its normal operation. After the drone took off, it flew in the wetland airspace according to a pre-set route. During flight, the obstacle avoidance sensor constantly monitored the environment below and around it to prevent the drone from colliding with obstacles such as trees and buildings in the wetland. The high-resolution optical camera, multispectral camera and thermal imager in the multi-source data acquisition module worked together. The high-resolution optical camera, with its high definition, clearly captured the overall topography, vegetation distribution and bird habitats of the wetland. The multispectral camera focused on obtaining spectral information of different vegetation and water bodies in the wetland. By analyzing spectral differences, it determined the health of the vegetation and the degree of pollution of the water body. The thermal imager uses the principle of thermal radiation to detect whether there are abnormal temperature areas in the wetland, such as local overheating or overcooling areas that may be caused by human activities. They use time synchronization and spatial calibration algorithms to ensure that the collected image data is consistent in time and space, providing a reliable basis for subsequent precise analysis. The collected data is transmitted to the data processing and analysis module in real time. First, the image is denoised to remove noise caused by light reflection, atmospheric interference and other factors to make the image clearer. Then, enhancement operations are performed to highlight key features in the image, such as enhancing the boundaries between vegetation and water bodies to facilitate subsequent analysis. Then, geometric correction is performed to accurately match the geographical location in the image with the actual geographical coordinates. The improved adaptive feature extraction algorithm is used to extract features related to changes in the wetland ecological environment, such as extracting the texture features of vegetation and the color features of water bodies. The dynamic threshold change detection algorithm is used to compare current and historical image data to detect changed areas of the wetland, such as changes in water area and vegetation coverage, and to evaluate the degree and nature of the changes. If it is found that the water area in a certain area has suddenly shrunk or the color of the vegetation has changed abnormally, the system will conduct further in-depth analysis to determine whether the changes are caused by natural factors or human activities. Based on the results of data processing and analysis, the intelligent decision-making module uses a deep integrated decision forest model to determine whether wetland changes are affected by illegal human activities. The model has been trained and optimized with a large amount of labeled wetland human activity data and can accurately identify various activity patterns. If there are abnormal changes and it is determined to be caused by illegal human activities, such as illegal reclamation of wetlands, the model automatically generates early warning information, including detailed location and activity type information. For example, when a new cofferdam is detected in a wetland and the surrounding vegetation is damaged, the system quickly determines that this is illegal reclamation and issues an early warning.

[0031] The communication and data transmission module adopts 5G communication technology, and with its advantages of high speed and low latency, it can quickly transmit the collected data back to the ground control and management platform. At the same time, combined with adaptive modulation and coding technology and multipath diversity transmission algorithm, it dynamically adjusts the modulation mode and coding rate according to the channel quality, and transmits data through multiple paths to ensure the reliability and real-time performance of data transmission and avoid data loss or delay.

[0032] At the wetland management station, operators operate through the personalized interface of the ground control and management platform. If the operator is more concerned about the analysis of verification results, the platform highlights the data visualization analysis module, which displays the changes in the wetland ecological environment in the form of intuitive charts and maps. For example, different colored areas are used to represent the increase or decrease in water area, and line graphs are used to show the changing trend of vegetation coverage. Operators can use this visual information to promptly identify problems, carry out task planning, monitoring and management, and provide decision-making support for wetland ecological protection. The data storage and backup module adopts distributed storage and redundant backup strategies, and stores the collected data in multiple storage nodes. It uses the consistent hashing algorithm to allocate storage nodes according to data type and time attributes, and performs multi-node backup of important data. The number of backups is dynamically adjusted according to the importance of the data and system resources to ensure data security and recoverability.

[0033] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as above in terms of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can, without departing from the scope of the technical solution of the present invention, make some changes or modifications to equivalent embodiments using the technical contents disclosed above. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. An intelligent UAV ecological verification system with multi-source data fusion, characterized by: The system includes the following components: UAV platform, multi-source data acquisition module, data processing and analysis module and intelligent decision-making module as well as communication and data transmission module and ground control and management platform; The UAV platform: A UAV is selected, equipped with GPS and IMU, and installed with obstacle avoidance sensors; The multi-source data acquisition module integrates a high-resolution optical camera, a multispectral camera, and a thermal imager, and integrates and optimizes the configuration of these devices so that they can work together on the UAV platform; The data processing and analysis module performs denoising, enhancement, and geometric correction on optical and multispectral images, uses deep learning algorithms and combines them with the characteristics of human activity patches in nature reserves for targeted training and optimization, performs change detection using difference and ratio algorithms, compares current and historical data, detects areas of surface change, and assesses the extent and nature of the change; The intelligent decision-making module establishes a decision-making model based on rules and machine learning. The machine learning part is trained and optimized through a large amount of labeled human activity data. The legality of human activities is judged based on the results of data processing and analysis. If illegal activities are found, early warning information is automatically generated and detailed location and type information is provided. The communication and data transmission module uses 4G, 5G or satellite communication wireless communication technology. The ground control station can remotely control the UAV flight mission and data collection parameters, and receive data for storage and further processing; The ground control and management platform: operators use this platform to plan, monitor and manage UAV missions. The platform has data visualization capabilities and optimizes the design of the human-computer interaction interface.

2. The UAV intelligent ecological verification system with multi-source data fusion according to claim 1 is characterized in that: The obstacle avoidance sensor is a laser radar or a visual sensor. The flight safety of the UAV in a complex environment is guaranteed by selection and layout. The layout algorithm is as follows: Assume that the coordinate system of the UAV body is The laser radar is installed in front of the UAV at a distance from the center of the fuselage. meters, in The offset on the axis is rice, The offset on the axis is Meters, the visual sensor is installed at the bottom of the drone, horizontal distance from the center of the fuselage m, with The angle between the axis and The axis angle is To prevent the drone from colliding with mountains, trees and other obstacles in the nature reserve during flight.

3. The UAV intelligent ecological verification system with multi-source data fusion according to claim 1 is characterized in that: In the multi-source data acquisition module, the high-resolution optical camera, multispectral camera and thermal imager work together using a time synchronization and space calibration algorithm. The time synchronization algorithm is: 、 、 They are high-resolution optical camera, multispectral camera and thermal imager, with the high-precision clock inside the drone as the reference clock. ,camera The collection time is ,camera The collection time is ,camera The collection time is , by calculating , , using the adjustment factor 、 To the camera and The acquisition time is adjusted by the following formula: , so that the data collected by the three cameras are synchronized in time, the spatial calibration algorithm is: establish a unified geographic coordinate system , based on the GPS positioning point of the drone, the camera is calculated through trigonometric function relationship 、 、 The geographic coordinate transformation parameters of the acquired image.

4. The UAV intelligent ecological verification system with multi-source data fusion according to claim 1 is characterized in that: The feature extraction algorithm of the data processing and analysis module is an improved adaptive feature extraction algorithm. , first perform multi-scale decomposition to obtain images at different scales , , then, calculate the gradient magnitude of the image at each scale and gradient direction The formula is: ， Then, based on the prior knowledge of human activity patterns in nature reserves, the weight vectors of different features are determined. ,in Indicates the The weight of the feature, is the total number of features. Finally, the feature vector is obtained by weighted fusion of gradient features at different scales. , the formula is ,in It is a function that encodes features according to the gradient direction.

5. The UAV intelligent ecological verification system with multi-source data fusion according to claim 1 is characterized in that: The change detection algorithm of the data processing and analysis module is a dynamic threshold change detection algorithm. Suppose the current captured image is , the historical reference image is , first calculate the image difference , calculate the dynamic threshold based on the local statistical information of the image , the formula is: in is the image difference In The mean of the neighborhood centered at is the standard deviation, It is a coefficient that is dynamically adjusted according to different nature reserve types and seasonal changes.

6. The UAV intelligent ecological verification system with multi-source data fusion according to claim 1 is characterized in that: The decision model based on machine learning of the intelligent decision module is a deep integrated decision forest model, which consists of multiple decision trees. Each decision tree is trained based on different sample subsets and feature subsets. , decision tree De Han Chuwei , which means that in the feature vector Belong to category The final output probability of the model is During the training process, the sample subset is obtained from the training data set by random sampling method, and the information gain is calculated as follows: Set features The dataset Divided into Part , information gain in It is a dataset The information entropy of is a subset The information entropy of the feature is used, and the Pearson correlation coefficient is used for correlation analysis to calculate the correlation between the feature and the target category, and remove the features with low correlation.

7. The UAV intelligent ecological verification system with multi-source data fusion according to claim 1 is characterized in that: The communication and data transmission module adopts adaptive modulation and coding technology combined with multipath diversity transmission algorithm. Adaptive modulation and coding technology dynamically adjusts the modulation mode and coding rate according to the channel quality index. Suppose the channel quality index is , the modulation mode set is , the coding rate set is , by looking for pre-established Mapping table with modulation mode and coding rate to determine the optimal modulation mode under current channel conditions and coding rate , the multipath diversity transmission algorithm is: Divide data into multiple sub-data streams , transmitted through different paths, the receiving end uses the maximum ratio combining algorithm to combine the data received by multiple paths, and assumes The signal received by each path is , its gain is , the combined signal is .

8. The UAV intelligent ecological verification system with multi-source data fusion according to claim 1 is characterized in that: The human-computer interaction interface optimization of the ground control and management platform adopts a personalized interface design method based on user behavior analysis. By collecting the operator's operation behavior data during the use of the platform, including operation frequency, operation path, and residence time, the operators are divided into different user types using a cluster analysis algorithm. For each user type, analyze their operating habits and needs, establish personalized interface layout and functional module arrangement rules, for user types who frequently plan tasks, place the task planning module in a prominent position on the interface, and optimize their operating procedures, for user types who pay more attention to verification result analysis, highlight the data visualization analysis module.

9. The UAV intelligent ecological verification system with multi-source data fusion according to claim 1 is characterized in that: The system also includes a data storage and backup module, which adopts distributed storage and redundant backup strategies. Distributed storage stores the collected data in multiple storage nodes. Suppose the storage node set is ,According to the type and time attributes of the data, the consistent hashing algorithm is used to distribute the data to different storage nodes.,The consistent hashing algorithm maps the data and storage nodes to a hash ring.,The redundant backup strategy is: for important data, it is backed up on multiple storage nodes, and the,number of backups is dynamically adjusted according to the importance of the data and the,system resource situation.

Images