A forestry resource collection method based on GIS technology

Abstract

The application discloses a forestry resource collection method based on GIS technology, relates to the technical field of data processing, and aims to solve the technical problem that existing catalysts generally have poor performance in a low-temperature flue gas environment. The method comprises the following steps: collecting remote sensing images of a target forest area through a remote sensing device; selecting a first target transmission strategy based on the remote sensing images, and transmitting the remote sensing images based on the first target transmission strategy; collecting sensor data through Internet of Things sensors arranged in the target forest area; selecting a second target transmission strategy based on the sensor data, and transmitting the sensor data based on the second target transmission strategy; performing dimension reduction processing and integration processing on the obtained remote sensing images and sensor data to obtain target forestry resource data; and the target forestry resource data comprises original collection data and prediction result information generated based on the original collection data.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a forestry resource collection method based on GIS technology. BACKGROUND

[0002] With the continuous development of modern science and technology, GIS (Geographic Information System) technology is increasingly widely used in various fields, and plays an important role in forestry resource management and monitoring. The collection, analysis and management of forestry resources have always been an important direction of forestry scientific research. Traditional forestry resource collection methods mostly rely on manual measurement and field investigation, which have problems such as low efficiency, poor precision and large time consumption. Therefore, how to efficiently and accurately obtain forestry resource information has become an important issue in current forestry management and research. SUMMARY

[0003] The main purpose of the present application is to provide a porous multi-element doped sludge carbon-based low-temperature denitration catalyst and a preparation method thereof, aiming to solve the technical problem that existing catalysts generally have poor performance in low-temperature flue gas environment.

[0004] To achieve the above-mentioned purpose, the technical solutions adopted by the embodiments of the present application are as follows:

[0005] In a first aspect, the embodiments of the present application provide a forestry resource collection method based on GIS technology, applied to a forestry resource collection system, the forestry resource collection system comprising a remote sensing image collection module, a sensor data collection module, a data transmission module and a terminal device; the remote sensing image collection module is used for image collection of target forest resource based on remote sensing collection equipment, and is transmitted to the terminal device based on the data transmission module; the sensor data collection module refers to a group of Internet of Things sensors arranged in the target forest area, used for real-time collection of sensor data and transmission to the terminal device based on the data transmission module; the data transmission module is used for transmitting data and dynamically adjusting the transmission strategy according to the data transmission environment; the terminal device comprises a high-dimensional data dimension reduction module and a data integration module;

[0006] The forestry resource collection method based on GIS technology comprises the following steps:

[0007] Remote sensing images of the target forest area are collected by a remote sensing device; based on the remote sensing images, a first target transmission strategy is selected, and the remote sensing images are transmitted based on the first target transmission strategy;

[0008] Sensor data are collected by the Internet of Things sensors arranged in the target forest area; the sensor data include real-time soil data, real-time water data and real-time temperature data; based on the sensor data, a second target transmission strategy is selected, and the sensor data are transmitted based on the second target transmission strategy;

[0009] The acquired remote sensing image and the sensor data are subjected to dimension reduction processing and integration processing to obtain target forestry resource data; the target forestry resource data includes original collection data and prediction result information generated based on the original collection data.

[0010] As some optional embodiments of the present application, the step of subjecting the acquired remote sensing image and the sensor data to dimension reduction processing and integration processing to obtain target forestry resource data comprises:

[0011] The acquired remote sensing image and the sensor data are subjected to dimension reduction processing and verification of whether target information is retained after the dimension reduction processing; if the verification is passed, low-dimensional data is obtained;

[0012] The acquired remote sensing image and the sensor data are subjected to integration processing to obtain integrated data;

[0013] After the low-dimensional data and the integrated data are subjected to decision-level data fusion processing, a target prediction model is inputted to obtain target forestry resource data and prediction result information;

[0014] The decision-level data fusion processing satisfies the following relationship: D fused =w low *D low +w full *D full ; wherein, D fused represents data after decision-level data fusion processing, w low and w full are weight values of the low-dimensional data and the integrated data, w low +w full＝ 1; D low represents the low-dimensional data, and D full represents the integrated data.

[0015] The target prediction model satisfies the following relationship:

[0016] wherein, represents prediction result information, D fused represents data after decision-level data fusion processing, θ represents a model parameter, and f(·) represents a prediction function.

[0017] The loss function of the target prediction model is:

[0018] wherein, represents a prediction value, y i represents a true value, n represents a sample number, represents a square error of each sample.

[0019] As some optional embodiments of the present application, the step of obtaining the low-dimensional data by dimension reduction processing the acquired remote sensing image and the sensor data comprises:

[0020] extracting features of the acquired remote sensing image based on a convolutional neural network to obtain remote sensing feature data; preprocessing the remote sensing feature data and the sensor data to obtain first multi-dimensional data; performing data centralization processing based on the first multi-dimensional data to obtain a centralized data matrix; calculating a covariance matrix of the centralized data based on the centralized data matrix; performing eigenvalue decomposition on the covariance matrix of the centralized data to obtain eigenvalues and eigenvectors; selecting a preset number of target eigenvectors to form a first target dimension reduction matrix; projecting the first multi-dimensional data to the first target dimension reduction matrix to obtain first low-dimensional data;

[0021] verifying whether the first low-dimensional data retains target information based on a three-dimensional scatter plot; if the verification is passed, the first low-dimensional data is taken as target low-dimensional data;

[0022] If the verification is not passed, the first multi-dimensional data is subjected to second dimension reduction processing based on a second dimension reduction strategy to obtain second low-dimensional data; the second dimension reduction processing is nonlinear dimension reduction processing;

[0023] The target information in the second low-dimensional data is also verified based on a three-dimensional scatter plot, and the second low-dimensional data that passes the verification is taken as target low-dimensional data.

[0024] As some optional embodiments of the present application, the step of obtaining the integrated data by integrating the acquired remote sensing image and the sensor data comprises:

[0025] After time synchronization processing of the acquired remote sensing image and the sensor data by interpolation processing, a plurality of image data pairs are obtained; the remote sensing image and the sensor data in the image data pair are consistent in time point;

[0026] The remote sensing image in each image data pair is converted into a feature vector based on a preset deep learning model, and the sensor data is converted into standard sensor data by Z-Score standardization processing to obtain a multi-dimensional data set;

[0027] The feature vector satisfies the following relationship formula: f I = f CNN (I); wherein, f CNN represents a preset deep learning model; f I represents a feature vector; I represents an input remote sensing image, and the dimension of I is (image height, image width, image channel number).

[0028] The standard sensor data satisfies the following relationship: Wherein, μ S And σ S Respectively represent the mean and standard deviation of the sensor data set, S i Indicates the i-th sensor data;

[0029] The multi-dimensional data set is subjected to feature splicing processing to obtain integrated data.

[0030] As some optional embodiments of the present application, the step of performing decision-level data fusion processing on the low-dimensional data and the integrated data to obtain target forestry resource data and prediction result information, comprising:

[0031] The low-dimensional data is input into a first preset prediction model to obtain coverage information of the target forest area; the first preset prediction model includes an input layer, a hidden layer and an output layer; the input layer is used to receive the low-dimensional data; the hidden layer includes a plurality of neurons, and a nonlinear activation function is used to learn the relationship between the low-dimensional data and the coverage information of the target forest area; the output layer is used to output the coverage information of the target forest area;

[0032] The integrated data is input into a second preset prediction model to obtain health status information, tree species distribution information and growth rate information of each tree species of the target forest area; the second preset prediction model includes an input layer, a hidden layer and an output layer; the input layer is used to receive the integrated data; the hidden layer includes three hidden sub-layers, and each hidden sub-layer extracts features of the integrated data through a combination of weighted sum and activation function; the output layer includes three output sub-layers, respectively used to output the health status information, the tree species distribution information and the growth rate information of each tree species of the target forest area.

[0033] As some optional embodiments of the present application, after the target forestry resource data and the prediction result information are obtained, the method further comprises:

[0034] Based on the coverage information, the health status information, the tree species distribution information and the growth rate information of each tree species of the target forest area, a forestry resource distribution map and a forestry resource prediction change map are generated to formulate a forestry management target strategy.

[0035] As some optional embodiments of the present application, based on the remote sensing image, a first target transmission strategy is selected, and the remote sensing image is transmitted based on the first target transmission strategy, comprising:

[0036] The remote sensing image is subjected to feature recognition processing to obtain feature information of the remote sensing image;

[0037] selecting a first restriction condition based on the size of the remote sensing image; the first restriction condition being a first target compression ratio, a first target bandwidth value, and a first target transmission times; selecting a second restriction condition based on the content of the remote sensing image; the second restriction condition being a first target resolution and whether to perform data compression; selecting a third restriction condition based on a transmission environment; the third restriction condition being a first target transmission path information; selecting a fourth restriction condition based on a transmission task of the remote sensing image; the fourth restriction condition being a first target transmission protocol;

[0038] obtaining a first target transmission strategy based on the first restriction condition, the second restriction condition, the third restriction condition, and the fourth restriction condition, and transmitting the remote sensing image based on the first target transmission strategy.

[0039] As some optional embodiments of the present application, the step of selecting a first restriction condition based on the size of the remote sensing image; the first restriction condition being a compression ratio, a bandwidth value, and a transmission times; selecting a second restriction condition based on the content of the remote sensing image; the second restriction condition being a resolution and whether to perform data compression; selecting a third restriction condition based on a transmission environment; the third restriction condition being a transmission path information; selecting a fourth restriction condition based on a transmission task of the remote sensing image; the fourth restriction condition being a transmission protocol, comprises:

[0040] if the remote sensing image is a low-resolution image or a single-band image, the first restriction condition is a low compression ratio and a low bandwidth value, and a single data transmission; if the remote sensing image is a high-resolution multi-spectral image or a hyperspectral image, the first restriction condition is a high compression ratio and a high bandwidth value, and a multi-time data transmission;

[0041] if there is an important geographical area or an abnormal area in the remote sensing image, the second restriction condition is a regional high-resolution transmission; if there are a large number of blank areas or repeated areas in the remote sensing image, data compression is performed using the redundant information in the image;

[0042] if the transmission path of the remote sensing image is long and the transmission network is congested, the third restriction condition is a multi-path transmission strategy; if the transmission path of the remote sensing image is short or the transmission network is not congested, the third restriction condition is a single-path transmission strategy;

[0043] if the transmission task of the remote sensing image is disaster monitoring, the transmission protocol is a user datagram protocol; if the transmission task of the remote sensing image is periodic and offline monitoring, the transmission protocol is a transmission control protocol.

[0044] As some optional embodiments of the present application, the step of selecting a second target transmission strategy based on the sensor data, and transmitting the sensor data based on the second target transmission strategy, comprises:

[0045] outputting the sensor data to a preset data grading model for data grading processing, and obtaining data grading values of the sensor data respectively; the data grading values comprise high priority values, medium priority values and low priority values;

[0046] selecting a second target transmission strategy based on the data grading values;

[0047] transmitting the sensor data based on the second target transmission strategy.

[0048] As some optional embodiments of the present application, the step of selecting a second target transmission strategy based on the data grading values, comprises:

[0049] for data with high priority values, using user datagram protocol for transmission;

[0050] for data with medium priority values, using transmission control protocol for transmission;

[0051] for data with low priority values, using message queue telemetry transmission for transmission.

[0052] The traditional forestry resource collection method mainly relies on manual investigation, ground measurement and other methods, which has low efficiency and low precision. The forestry resource collection method based on GIS technology proposed in the present application has obvious technical progress, which is embodied in the following aspects:

[0053] The traditional forestry resource collection method usually relies on a single data source, such as only using remote sensing images or only using ground measurement data. However, by comprehensively using the remote sensing image collection module and the Internet of Things sensor data collection module, the present application can realize all-round and multi-level resource collection from spatial information (remote sensing image) to environmental data (such as climate, soil, moisture, temperature, etc.). Remote sensing images provide spatial information of forest areas, while sensor data monitor the environmental status of forest areas in real time. The combination of the two greatly improves the accuracy and comprehensiveness of data collection.

[0054] In traditional forestry resource collection systems, data transmission is usually static and cannot adapt to different transmission environments. However, the data transmission module in this application has the ability to dynamically adjust the transmission strategy. According to the real-time situation of the collected data and the transmission environment, the system can intelligently select the optimal transmission strategy (such as selecting the appropriate compression method, transmission path, protocol, etc.), effectively avoiding the problems of data loss and low transmission efficiency caused by transmission delay, insufficient bandwidth or signal interference in traditional methods. This dynamic adjustment mechanism ensures the stability and real-time performance of data transmission in complex forest environments, greatly improving the overall performance of the system.

[0055] Remote sensing images and sensor-collected data usually have high dimensionality and large data volume, making data processing and analysis very complex and resource-intensive. In traditional forestry resource management, the technology for processing these high-dimensional data is usually simple and has problems such as slow analysis speed and inaccurate results. The solution in this application introduces a high-dimensional data dimension reduction module and a data integration module in the terminal device to perform dimension reduction and integration processing on remote sensing images and sensor data. This method can effectively reduce the dimensionality of data, remove redundant information, and retain key information in the data, significantly improving the efficiency of data processing. At the same time, the integrated data can provide more accurate and valuable basis for subsequent analysis and decision-making.

[0056] Due to the combination of sensors and remote sensing technology, this application can achieve real-time monitoring of target forest areas and timely acquisition of key environmental indicators such as climate, soil, moisture, and temperature. Compared with traditional manual survey methods, this application greatly improves the timeliness and accuracy of monitoring. This allows forestry resource managers to quickly respond to changes in forest resources and take timely measures to address potential ecological risks or resource losses. For example, in the event of a forest fire or insect infestation, the latest information about the forest area can be obtained through sensor data and remote sensing images in the first time, quickly locating the problem area, and taking appropriate emergency measures.

[0057] Existing traditional forestry resource collection methods often struggle when faced with large-scale forest areas and cannot quickly cover large areas of forest land. Unlike traditional methods, the system architecture of this application has good scalability and adaptability. As the demand for forestry resource management increases, the system can quickly expand the monitoring range and maintain efficient operation by adding remote sensing devices, sensor nodes, and other components. In addition, the data transmission module can adapt to various geographical conditions and climate environments by dynamically adjusting the transmission strategy according to changes in different environments, not only working efficiently in open areas, but also stably operating in complex and remote forest environments.

[0058] By deeply integrating and reducing the dimension of remote sensing images and sensor data, the application can provide intelligent decision support for forestry resource management. The system not only can monitor the changes of forest in real time, but also can automatically generate early warning information such as climate change, pest warning, resource anomaly, etc. through data analysis and pattern recognition. This enables forestry managers to identify potential problems in advance and take timely measures to effectively reduce the loss of forest resources and environmental risks.

[0059] In summary, the forestry resource collection method based on GIS technology proposed in the application has made significant technical progress in many aspects compared to the prior art. By integrating remote sensing image collection, Internet of Things sensor data collection, dynamic transmission strategy, high-dimensional data dimension reduction and integration processing, etc. advanced technologies, it realizes an efficient, accurate and real-time forestry resource collection and monitoring system. This scheme not only improves the efficiency and accuracy of forestry resource collection, but also provides intelligent support for real-time monitoring, data analysis and decision-making of forest resources, and has wide application prospects and practical significance. BRIEF DESCRIPTION OF DRAWINGS

[0060] Figure 1 The electronic device structure schematic diagram of the hardware running environment involved in the embodiment of the application;

[0061] Figure 2 The module schematic diagram of the forestry resource collection system provided by the embodiment of the application;

[0062] Figure 3 The flowchart of the forestry resource collection method based on GIS technology provided by the embodiment of the application;

[0063] Figure 4 The schematic diagram of a remote sensing image provided by the embodiment of the application;

[0064] Markings in the figure: 101-processor, 102-communication bus, 103-network interface, 104-user interface, 105-memory. DETAILED DESCRIPTION

[0065] It should be understood that the specific embodiments described herein are only used to explain the application and not to limit the application.

[0066] It should be understood that the specific embodiments described herein are only used to explain the application and not to limit the application.

[0067] Referring to the drawings Figure 1 , the drawings Figure 1For the hardware running environment of the embodiment of the present application, the electronic device structure schematic diagram can include: a processor 101, for example, a central processing unit (CPU), a communication bus 102, a user interface 104, a network interface 103, and a memory 105. The communication bus 102 is used to realize the connection and communication between the components. The user interface 104 can include a display, an input unit such as a keyboard, and an optional user interface 104 can also include a standard wired interface, a wireless interface. The network interface 103 can optionally include a standard wired interface, a wireless interface (such as a wireless fidelity (WIreless-FIdelity, WI-FI) interface). The memory 105 can be an independent storage device of the foregoing processor 101, and the memory 105 can be a high-speed random access memory (RAM) memory, or a stable non-volatile memory (NVM), for example, at least one disk memory; the processor 101 can be a general-purpose processor, including a central processing unit, a network processor, etc., and can also be a digital signal processor, an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component.

[0068] Those skilled in the art can understand that the structure shown in the foregoing Figure 1 does not constitute a limitation on the electronic device, and can include more or fewer components than the diagram, or combine certain components, or different component arrangements.

[0069] As shown in the foregoing Figure 1 , the memory 105 as a storage medium can include an operating system, a network communication module, a user interface module, and a forestry resource collection system.

[0070] In the foregoing Figure 1 electronic device, the network interface 103 is mainly used for data communication with a network server; the user interface 104 is mainly used for data interaction with a user; the processor 101 and the memory 105 in the present application can be arranged in the electronic device, and the electronic device calls the forestry resource collection system stored in the memory 105 through the processor 101, and executes the forestry resource collection method provided in the embodiment of the present application.

[0071] With the continuous development of modern technology, GIS (Geographic Information System) technology is increasingly widely used in various fields, especially in forestry resource management and monitoring. The collection, analysis and management of forestry resources have always been an important direction of forestry scientific research. Traditional forestry resource collection methods mostly rely on manual measurement and field investigation, which have problems such as low efficiency, poor accuracy, and large time consumption. Therefore, how to efficiently and accurately obtain forestry resource information has become an important issue in current forestry management and research.

[0072] Traditional forestry resource collection methods mainly rely on manual investigation and ground measurement. Although these methods are effective in a certain period, with the increasing demand for forest resource protection and the expansion of resource scale, traditional methods show many shortcomings. Manual measurement is limited by time, space and personnel, and cannot cover a wide range of forest areas. In addition, traditional collection methods also cannot monitor the changes of forest ecological environment in real time, and the data processing efficiency is low, which cannot meet the needs of modern forestry management.

[0073] Remote sensing technology, as an efficient resource monitoring method, has been widely used in the collection and monitoring of forestry resources. Through satellite remote sensing, unmanned aerial vehicle remote sensing and other equipment, large-scale, high-resolution forest image data can be obtained to monitor the changes of forests in real time, including forest coverage, tree species distribution, pest and disease invasion, etc. Remote sensing technology has the advantages of wide spatial coverage and strong timeliness, which can overcome the limitations of traditional collection methods. However, it also has problems such as large data volume, complex processing and inaccurate information, so it needs to be combined with other technical means for further processing and analysis of data.

[0074] The development of Internet of Things technology, especially the progress of sensor technology, makes it possible to monitor the real-time environment data of forest areas. By deploying a set of Internet of Things sensors in the target forest area, environmental data such as climate, soil, moisture and temperature can be collected in real time to provide timely dynamic data for the monitoring of forestry resources. These sensors can work continuously under different environmental conditions and provide high-frequency and high-precision data collection functions. Combined with the advantages of Internet of Things technology, changes in forest ecological environment can be captured in real time, and the accuracy and timeliness of forest resource management can be improved.

[0075] The collection of forestry resources often involves a large amount of data, especially remote sensing images and sensor data, which need to be transmitted to the central processing system for analysis and storage after collection. In remote areas such as forests, due to geographical location, environmental interference and other reasons, data transmission often faces problems such as insufficient bandwidth, large transmission delay, data loss, etc. Therefore, how to dynamically adjust the data transmission strategy according to different transmission environments to ensure efficient and secure data transmission has become a key technical problem of forestry resource collection systems.

[0076] In the process of forestry resource collection, remote sensing images and sensor data often have high-dimensional characteristics. If these data are used directly for analysis without processing, it can lead to waste of computing resources and low processing efficiency. Dimensionality reduction of high-dimensional data can effectively reduce the dimensionality of the data, remove redundant information, and improve the efficiency and accuracy of data processing. The reduced data will be more concise and useful, which can help decision-makers quickly extract valuable forestry resource information. In addition, data integration processing technology can unify different sources and formats of data to form comprehensive target forestry resource data, providing accurate information support for further analysis and decision-making.

[0077] GIS technology, as a powerful spatial data management and analysis tool, can combine geographic spatial information with forestry resource data, and provide support for forestry resource management and decision-making through spatial analysis and geographic information visualization. In the GIS-based forestry resource collection system, remote sensing images and sensor data can be integrated and analyzed through spatial information to accurately depict the spatial distribution of forest resources, thereby achieving accurate monitoring and dynamic management of forest ecological environment.

[0078] To overcome the shortcomings of the traditional method, modern forestry resource collection systems are gradually introducing the comprehensive application of remote sensing technology, Internet of Things technology, data transmission technology, and GIS technology, through precise environmental data collection, real-time data transmission, and efficient data processing, to achieve comprehensive, real-time, and efficient monitoring and management of forestry resources.

[0079] The technical solution of the present application proposes a forestry resource collection method based on GIS technology, which uses remote sensing image collection modules, sensor data collection modules, data transmission modules, and terminal devices, and combines remote sensing image data and Internet of Things sensor data, through dimensionality reduction and data integration, to provide accurate and real-time forestry resource data, improving the efficiency and accuracy of forestry resource management. Through this method, the ecological environment changes of the target forest area can be effectively monitored, and forest resource data can be obtained in a timely manner, providing scientific basis and decision support for forestry management departments.

[0080] In summary, with the advancement of remote sensing technology, Internet of Things technology, data transmission technology, and GIS technology, forestry resource collection methods based on these technologies can overcome the shortcomings of traditional methods and achieve more efficient and accurate forestry resource monitoring and management. Therefore, the technical solution of the present application not only improves the efficiency of resource collection, but also realizes efficient transmission and processing of data, providing a more reliable and intelligent tool for modern forestry management.

[0081] Specifically, the forestry resource collection method based on GIS technology described in the present application is applied to forestry resource collection systems, forestry resource management systems, forestry resource monitoring systems, and forestry resource analysis systems. Figure 2The forestry resource collection system shown comprises a remote sensing image collection module, a sensor data collection module, a data transmission module and a terminal device; the remote sensing image collection module is used for image collection of target forest resource based on a remote sensing collection device, and is transmitted to the terminal device based on the data transmission module; the sensor data collection module refers to a group of Internet of Things sensors arranged in the target forest area, and is used for real-time collection of sensor data and transmission to the terminal device based on the data transmission module; the data transmission module is used for transmission of data and dynamic adjustment of transmission strategies according to data transmission environments; and the terminal device comprises a high-dimensional data dimension reduction module and a data integration module.

[0082] The forestry resource collection method based on the GIS technology comprises the following steps as shown in the figure: Figure 3

[0083] Step S10: Remote sensing images of the target forest area are collected by a remote sensing device; based on the remote sensing images, a first target transmission strategy is selected, and the remote sensing images are transmitted based on the first target transmission strategy.

[0084] The remote sensing device can be a satellite remote sensor, an unmanned aerial vehicle (UAV) remote sensing device, an aerial remote sensing device or a ground remote sensing device; wherein the satellite remote sensor can comprise a Landsat series satellite, a Sentinel-1 and a Sentinel-2 or a MODIS. The unmanned aerial vehicle (UAV) remote sensing device can comprise an RGB (red, green and blue) camera, a multi-spectral camera, a laser radar (LiDAR) or an infrared thermal imager. The aerial remote sensing device can comprise a high-resolution digital camera, a laser radar (LiDAR) or an aerial photographic survey system. The ground remote sensing device can comprise a spectrometer, a global positioning system (GPS) device or a mobile sensor. The obtained remote sensing images are optical images, synthetic aperture radar (SAR) images, high-resolution images, laser radar (LiDAR) point cloud data, thermal infrared images, normalized vegetation index (NDVI) images, vegetation type and coverage images or change detection images. As shown in the figure, one of the remote sensing images obtained based on the remote sensing device is shown. Figure 4

[0085] As described in the present application, the remote sensing image is obtained by the following steps: selecting a Landsat8 satellite image, ensuring the timeliness (such as the latest image data) and spatial resolution (for example, 30-meter resolution) of the image. The obtained satellite image is preprocessed by radiation correction, geometric correction and the like to eliminate the influence of atmosphere, sensor and terrain and the like. The classification tool in ArcGIS is used to extract the vegetation type, classify the forest vegetation by using the NDVI and other vegetation indexes, and identify different vegetation types.

[0086] ​​In a further aspect, step S10 comprises:

[0087] Step S11, performing feature recognition processing on the remote sensing image to obtain feature information of the remote sensing image.

[0088] Before performing feature recognition processing on the remote sensing image, the remote sensing image can also be pre-processed, such as using a filtering algorithm (such as median filtering, Gaussian filtering, etc.) to remove noise in the image and improve image quality; or by adjusting the contrast, brightness, or using histogram equalization, etc. method, enhance the details of the remote sensing image, so that the features are more obvious.

[0089] The feature recognition extraction can include: using Canny, Sobel, Laplacian, etc. algorithm to detect the edges in the image, extract the contour and boundary information of the remote sensing image; or, through the gray level co-occurrence matrix (GLCM), local binary pattern (LBP), etc. method to extract the texture features in the image; or, shape recognition: using morphological operations (such as erosion, dilation, etc.) to extract the shape features of the target region; or, using machine learning or deep learning algorithms (such as support vector machine SVM, convolutional neural network CNN, etc.) to classify different features in the remote sensing image, and mark the targets (such as buildings, roads, forests, etc.) in the image.

[0090] Step S12, based on the size of the remote sensing image, selecting a first restriction condition; the first restriction condition is a first target compression ratio, a first target bandwidth value and a first target transmission number; based on the content of the remote sensing image, selecting a second restriction condition; the second restriction condition is a first target resolution and whether to perform data compression; based on the transmission environment, selecting a third restriction condition; the third restriction condition is a first target transmission path information; based on the transmission task of the remote sensing image, selecting a fourth restriction condition; the fourth restriction condition is a first target transmission protocol.

[0091] In order to facilitate those skilled in the art to understand step S12, the above four restriction conditions will be described in detail in combination with actual situations:

[0092] For the first restriction condition: for example, for large size images, a higher compression ratio may be needed to reduce the transmission file size. For example, if the image is large, a transmission path with higher bandwidth is selected to ensure that the transmission is completed within a specified time. For example, considering the size of the image and the bandwidth, it is decided whether multiple transmissions are needed to avoid retransmission caused by a failed transmission.

[0093] For the second restriction condition: for example, if the image content is relatively simple, a lower resolution can be selected; if high-precision analysis is required, a higher resolution can be selected. The resolution can be selected by setting a standard or predefined value, which is not specially limited here. For example, if the image content contains less details or redundancy, image compression can be selected to reduce data volume, but if the content is complex and details are crucial, compression can not be performed.

[0094] For the third restriction condition: for example, when the network condition is poor, a redundant path can be selected to ensure reliable transmission of data, or a stronger error correction algorithm can be used. For example, consider the load situation of the current transmission path to avoid selecting a path that is operating under high load.

[0095] For the fourth restriction condition: for example, in situations requiring high reliability and accuracy of transmission, TCP protocol can be selected; while in real-time transmission requiring low delay, UDP protocol can be selected. According to the image size and transmission delay requirements, appropriate protocols and strategies can be selected to balance speed and data loss risk.

[0096] That is, if the remote sensing image is a low-resolution image or a single-band image, the first restriction condition is low compression ratio and low bandwidth value and single data transmission; if the remote sensing image is a high-resolution multi-spectral image or a hyperspectral image, the first restriction condition is high compression ratio and high bandwidth value, and the data transmission is performed in multiple times;

[0097] If there is an important geographical area or an abnormal area in the remote sensing image, the second restriction condition is regional high-resolution transmission; if there are a large number of blank areas or repeated areas in the remote sensing image, data compression is performed using the redundancy information in the image;

[0098] If the transmission path of the remote sensing image is long and the transmission network is congested, the third restriction condition is a multi-path transmission strategy; if the transmission path of the remote sensing image is short or the transmission network is not congested, the third restriction condition is a single-path transmission strategy;

[0099] If the transmission task of the remote sensing image is disaster monitoring, the transmission protocol is user datagram protocol; if the transmission task of the remote sensing image is periodic and offline monitoring, the transmission protocol is transmission control protocol.

[0100] Step S13, based on the first restriction condition, the second restriction condition, the third restriction condition and the fourth restriction condition, a first target transmission strategy is obtained, and the remote sensing image is transmitted based on the first target transmission strategy.

[0101] Specifically, the selected constraints in step S12 are synthesized to analyze the mutual influence and optimal combination of various conditions (such as bandwidth, compression ratio, resolution, transmission path, transmission protocol, etc.). The optimal transmission scheme is determined by balancing between the conditions. For example, if the bandwidth is small, the compression ratio needs to be increased, and a lower resolution is selected; if the transmission environment is relatively stable, a high-resolution and non-compressed scheme can be selected.

[0102] After determining the first target transmission strategy, the transmission of the remote sensing image begins. During the transmission process, the image file is processed according to the predetermined strategy (such as compression, blocking, encryption, etc.), and is transmitted through the selected path and protocol. During the transmission process, the transmission state needs to be monitored to ensure data integrity and timely response to possible transmission problems (such as packet loss, timeout, etc.). After the transmission is completed, the received image is verified to ensure that its integrity, resolution, etc. meet the predetermined requirements. If transmission problems (such as image damage or loss) occur, retransmission or adjustment of the strategy is performed according to the feedback error information.

[0103] In step S20, sensor data is collected by Internet of Things sensors arranged in the target forest area; the sensor data includes real-time soil data, real-time water data, and real-time temperature data; based on the sensor data, a second target transmission strategy is selected, and the sensor data is transmitted based on the second target transmission strategy.

[0104] The arrangement position of the Internet of Things sensor will be described in detail below with examples:

[0105] Suppose there are three different areas in a certain forest area: dense forest area, forest edge transition area, and open area. Sensors can be arranged in these areas to ensure all-around monitoring:

[0106] In the dense forest area, the sensor type can be a soil moisture sensor and a temperature sensor; it can be arranged in the dense forest area (such as places close to trees), where the soil moisture is usually high and the temperature change is relatively stable, and the sensor can be arranged at the root of several different types of trees.

[0107] In the forest edge transition area, the sensor type can be a soil moisture sensor and a real-time water data sensor. It can be arranged near the edge of the forest, where the soil moisture changes greatly and is more significantly affected by external climate. Sensors can be arranged here to monitor the change of soil moisture and the influence of external temperature.

[0108] In the open area, the sensor type can be a temperature sensor and a soil moisture sensor. When arranging, the open area without shelter can be selected, such as grassland or shrub area. The climate conditions in these areas are usually extreme, so temperature and humidity need to be strictly monitored.

[0109] The soil real-time data includes soil humidity value and soil temperature value, the soil humidity value refers to the percentage of soil water content, reflecting the dry and wet conditions of the soil; usually the sensor obtains humidity data through resistance or capacitance change. The soil temperature value refers to the temperature of the soil, which affects the growth of plants, microbial activity, etc. The sensor obtains temperature data through thermistor or thermocouple.

[0110] The soil real-time data such as soil temperature, humidity, pH value, etc. needs to consider geographical distribution in transmission, which may cover agricultural or environmental monitoring scenarios. The water real-time data usually refers to water source, groundwater or soil moisture, which involves dynamic response to environmental changes. The temperature real-time data as an important environmental parameter requires real-time transmission with significant timeliness. As can be seen, due to the difference of the collected data, the priority in transmission is also different, therefore, based on the sensor data, the second target transmission strategy is selected, and the sensor data is transmitted based on the second target transmission strategy, including:

[0111] The sensor data is output to a preset data classification model for data classification processing, and the data classification value of the sensor data is obtained respectively; the data classification value includes high priority value, medium priority value and low priority value; based on the data classification value, a second target transmission strategy is selected; and the sensor data is transmitted based on the second target transmission strategy.

[0112] Specifically, the step of selecting a second target transmission strategy based on the data classification value includes:

[0113] For data with high priority value, user datagram protocol is used for transmission; for example, temperature and humidity in climate data, water data, which need to ensure real-time transmission;

[0114] For data with medium priority value, transmission control protocol is used for transmission; for example, soil data, certain temperature data, which can be slightly delayed, and are suitable for lower bandwidth and stable connection;

[0115] For data with low priority value, message queue telemetry transmission is used for transmission; for example, redundant auxiliary data or background information, which can be transmitted through low priority queue.

[0116] In step S30, the obtained remote sensing image and sensor data are processed by dimension reduction and integration, and target forestry resource data is obtained; the target forestry resource data includes original collected data and prediction result information generated based on the original collected data.

[0117] To ensure data security, the target forestry resource data are respectively stored in the local and the cloud, and the local and the cloud are synchronized in time to ensure the consistency of the data.

[0118] In a further aspect, step S30 comprises:

[0119] Step S31, the acquired remote sensing image and the sensor data are processed by dimension reduction, and whether the target information is retained after the dimension reduction processing is verified, if the verification is passed, low-dimensional data is obtained;

[0120] Specifically, this step comprises:

[0121] The acquired remote sensing image is extracted based on a convolutional neural network to obtain remote sensing feature data; the remote sensing feature data and the sensor data are preprocessed to obtain first multi-dimensional data; the first multi-dimensional data are processed by data centralization to obtain a centralized data matrix;

[0122] The centralized data matrix satisfies the following relationship:

[0123] X centerd =D multi -μ

[0124] In the formula, μ is the mean value of each column, D multi is the first multi-dimensional data, and X centerd is the centralized data matrix;

[0125] Based on the centralized data matrix, a covariance matrix of the centralized data is calculated and obtained; the covariance matrix of the centralized data is subjected to eigenvalue decomposition to obtain eigenvalues and eigenvectors; a preset number of target eigenvectors are selected to form a first target dimension reduction matrix; the first multi-dimensional data are projected to the first target dimension reduction matrix to obtain first low-dimensional data;

[0126] Whether the first low-dimensional data retains the target information is verified based on a three-dimensional scatter plot; if the verification is passed, the first low-dimensional data is taken as target low-dimensional data;

[0127] If the verification is not passed, the first multi-dimensional data are subjected to second dimension reduction processing based on a second dimension reduction strategy to obtain second low-dimensional data; the second dimension reduction processing is nonlinear dimension reduction processing, such as t-SNE;

[0128] The target information in the second low-dimensional data is also verified based on a three-dimensional scatter plot, and the second low-dimensional data that passes the verification is taken as target low-dimensional data.

[0129] Specifically, the verification of whether the first low-dimensional data retains the target information based on the three-dimensional scatter plot is to compare the data distribution before and after dimension reduction. Generally, the retention of target information means that the clustering structure and classification structure of the data should be as unchanged as possible. For the verification stage, the similarity of the data in space after dimension reduction and the original data can be calculated. Common similarity measures include cosine similarity and Euclidean distance.

[0130] The cosine similarity is used to measure the similarity of two vectors, with a value range of [-1, 1]. The greater the value, the higher the similarity. The Euclidean distance can calculate the direct distance between data points and is suitable for verifying the change in the position of data in space.

[0131] Step S32, integrating the obtained remote sensing image and sensor data to obtain integrated data;

[0132] Step S32 includes:

[0133] After time synchronization processing of the obtained remote sensing image and sensor data using interpolation processing, a plurality of image data pairs are obtained. The remote sensing image and sensor data in the image data pair are time consistent.

[0134] The remote sensing image in each image data pair is converted into a feature vector based on a preset deep learning model, and the sensor data is converted into standard sensor data through Z-Score standardization processing to obtain a multidimensional data set.

[0135] The feature vector satisfies the following relationship: I f CNN (I); wherein f CNN represents a preset deep learning model; f I represents a feature vector; I represents an input remote sensing image, and the dimension of I is (image height, image width, image channel number).

[0136] The standard sensor data satisfies the following relationship: wherein μ S and σ S respectively represent the mean and standard deviation of the sensor data set, represents the i-th standard sensor data, S i represents the i-th sensor data.

[0137] The multidimensional data set is subjected to feature splicing processing to obtain integrated data.

[0138] Step S33, after decision-level data fusion processing of the low-dimensional data and the integrated data, inputting a target prediction model to obtain target forestry resource data and prediction result information.

[0139] The decision-level data fusion processing satisfies the following relationship: D fused =w low *D low +w full *D full ; wherein, D fused represents the data after decision-level data fusion processing, w low and w full are weight values of low-dimensional data and integrated data respectively, w low +w full＝ 1; D low represents low-dimensional data, D full represents integrated data;

[0140] The target prediction model satisfies the following relationship:

[0141] wherein, represents the prediction result information, D fused represents the data after decision-level data fusion processing, and θ represents the model parameter, and f(·) represents the prediction function.

[0142] The loss function of the target prediction model is:

[0143] wherein, represents the predicted value, y i represents the true value, n represents the number of samples, represents the squared error of each sample.

[0144] The step of performing decision-level data fusion processing on the low-dimensional data and the integrated data to obtain target forestry resource data and prediction result information comprises:

[0145] inputting the low-dimensional data into a first preset prediction model to obtain coverage rate information of a target forest area; the first preset prediction model comprises an input layer, a hidden layer and an output layer; the input layer is used to receive the low-dimensional data; the hidden layer comprises a plurality of neurons and learns the relationship between the low-dimensional data and the coverage rate information of the target forest area by using a nonlinear activation function; and the output layer is used to output the coverage rate information of the target forest area.

[0146] input the integrated data into a second preset prediction model to obtain health status information, tree species distribution information, and growth rate information of each tree species of the target forest area; the second preset prediction model comprises an input layer, a hidden layer, and an output layer; the input layer is used to receive the integrated data; the hidden layer comprises three hidden sub-layers, and each hidden sub-layer extracts features of the integrated data through a combination of weighting and an activation function; the output layer comprises three output sub-layers, and is used to output the health status information, the tree species distribution information, and the growth rate information of each tree species of the target forest area, respectively.

[0147] After the target forestry resource data and the prediction result information are obtained, the method further comprises:

[0148] Based on the coverage information, the health status information, the tree species distribution information, and the growth rate information of each tree species of the target forest area, a forestry resource distribution map and a forestry resource prediction change map are generated to formulate a forestry management target strategy.

[0149] The training processes of the models in the technical solution are based on conventional methods and are not the focus of the technical solution of the present application, and thus will not be described in detail here.

[0150] In summary, the forestry resource collection method based on GIS technology has obvious technical progress, which is embodied in the following aspects:

[0151] Traditional forestry resource collection methods usually rely on a single data source, such as using only remote sensing images or only using ground measurement data. However, the present application can realize all-around and multi-level resource collection from spatial information (remote sensing images) to environmental data (such as climate, soil, moisture, temperature, etc.) by comprehensively using a remote sensing image collection module and an Internet of Things sensor data collection module. Remote sensing images provide spatial information of the forest area, and sensor data monitor the environmental state of the forest area in real time. The combination of the two greatly improves the accuracy and comprehensiveness of data collection.

[0152] In traditional forestry resource collection systems, data transmission is usually static and cannot adapt to different transmission environments. However, the data transmission module in the present application has the ability to dynamically adjust the transmission strategy. According to the real-time situation of the collected data and the transmission environment, the system can intelligently select the optimal transmission strategy (such as selecting appropriate compression methods, transmission paths, protocols, etc.), effectively avoiding the problems of data loss and low transmission efficiency caused by transmission delay, insufficient bandwidth, or signal interference in traditional methods. This dynamic adjustment mechanism ensures the stability and real-time performance of data transmission in complex forest environments, greatly improving the overall performance of the system.

[0153] Remote sensing images and sensor-collected data often have high dimensions and large data volumes, making data processing and analysis very complex and resource-intensive. In traditional forestry resource management, the techniques for processing these high-dimensional data are often rudimentary, with slow analysis speeds and inaccurate results. The present application introduces a high-dimensional data dimension reduction module and a data integration module into the terminal device to reduce and integrate remote sensing images and sensor data. This method effectively reduces the dimensionality of the data, removes redundant information, retains key information in the data, and significantly improves the efficiency of data processing. At the same time, the integrated data can provide more accurate and valuable basis for subsequent analysis and decision-making.

[0154] GIS technology, as a powerful spatial data analysis tool, can effectively integrate and analyze multi-source data such as remote sensing images and sensor data with geographic spatial information. In traditional forestry resource collection systems, geographic spatial information is often not fully utilized, and the combination of resource data and spatial data is weak. The present application integrates processing based on GIS, enabling data from different sources to be correlated and analyzed based on geographic space, enabling accurate positioning and dynamic monitoring of forest resources. Through GIS technology, the resource status and environmental changes in different forest areas can be comprehensively evaluated, providing real-time and accurate support for forestry resource management for decision-makers.

[0155] Due to the combination of sensors and remote sensing technology, the present application can achieve real-time monitoring of target forest areas and timely acquisition of key environmental indicators such as climate, soil, moisture, and temperature. Compared to traditional manual survey methods, the present application greatly improves the timeliness and accuracy of monitoring. This enables forestry resource managers to quickly respond to changes in forest resources and take timely measures to address potential ecological risks or resource losses. For example, in the event of a forest fire or insect infestation, the latest information about the forest area can be obtained through sensor data and remote sensing images, quickly locating the problem area and taking appropriate emergency measures.

[0156] Existing traditional forestry resource collection methods often struggle to cover large-scale forest areas quickly. Unlike traditional methods, the system architecture of the present application has good scalability and adaptability. As the demand for forestry resource management increases, the system can quickly expand the monitoring range by adding remote sensing devices, sensor nodes, and other components while maintaining efficient operation. In addition, the data transmission module can adapt to various geographic conditions and climate environments by dynamically adjusting the transmission strategy according to changes in different environments, allowing it to work efficiently in open areas and stably in complex and remote forest environments.

[0157] By deeply integrating and dimensionality reduction processing of remote sensing images and sensor data, the application can provide intelligent decision support for forestry resource management. The system not only can monitor the changes of the forest in real time, but also can automatically generate early warning information such as climate change, pest and disease warning, resource anomaly, etc. through data analysis and pattern recognition. This enables forestry managers to identify potential problems in advance and take timely measures, thereby effectively reducing the loss of forest resources and environmental risks.

[0158] In summary, the forestry resource collection method based on GIS technology proposed in the application has made significant technical progress in many aspects compared to the prior art. By integrating remote sensing image collection, Internet of Things sensor data collection, dynamic transmission strategy, high-dimensional data dimensionality reduction and integration processing, etc., it realizes an efficient, accurate and real-time forestry resource collection and monitoring system. This scheme not only improves the efficiency and accuracy of forestry resource collection, but also provides intelligent support for real-time monitoring, data analysis and decision-making of forest resources, and has wide application prospects and practical significance.

[0159] Based on the same inventive concept as in the foregoing embodiments, the embodiments of the application also provide a computer-readable storage medium storing a computer program, which, when loaded and executed by a processor, implements the forestry resource collection method based on GIS technology provided by the embodiments of the application.

[0160] Based on the same inventive concept as in the foregoing embodiments, the embodiments of the application also provide an electronic device comprising a processor and a memory, wherein,

[0161] The memory is configured to store a computer program;

[0162] The processor is configured to load and execute the computer program to enable the electronic device to perform the forestry resource collection method based on GIS technology provided by the embodiments of the application.

[0163] In some embodiments, the computer-readable storage medium can be FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface memory, optical disc, or CD-ROM, etc. memory; or various devices including one or any combination of the above memories. The computer can be various computing devices including smart terminals and servers.

[0164] In some embodiments, the executable instructions can be in the form of programs, software, software modules, scripts or codes, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and can be deployed in any form, including as standalone programs or as modules, components, subroutines or other units suitable for use in computing environments.

[0165] As an example, executable instructions can correspond to a file in a file system, but in many cases will reside in a portion of the main memory (e.g., random access memory) during execution; in some cases, the executable instructions can also reside on a secondary storage device (e.g., a disk) from which they are loaded into the main memory during execution. In some cases, the file system can be a local file system; in other cases, the file system can be accessed remotely (e.g., via a network).

[0166] As an example, executable instructions can be deployed on one computing device, or on multiple computing devices located at one site, or distributed across multiple sites and interconnected by a communication network.

[0167] It should be noted that, as used in this document, the terms "includes" and / or "containing" or variations thereof, mean that the specified element is included in the process, method, article, or system that includes that element, without the exclusion of any additional elements that the process, method, article, or system specified in the statement of subject matter can include. Limiting the scope of the subject matter recited in the statement of subject matter to that which is recited in the statement of subject matter, the element specified by the phrase "comprising a" does not exclude the presence of additional identical elements in the process, method, article, or system that includes that element.

[0168] The above-mentioned sequence numbers of the embodiments of the present application are only for description, and do not represent advantages or disadvantages of the embodiments.

[0169] From the above description of the embodiments, one skilled in the art can clearly understand that the above-mentioned embodiment methods can be realized by means of software and necessary general hardware platforms, of course, can also be realized by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory / random access memory, a magnetic disk, an optical disk), and includes a plurality of instructions for causing a multimedia terminal device (which can be a mobile phone, a computer, a television receiver, or a network device, etc.) to execute the methods described in the various embodiments of the present application.

[0170] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A GIS technology-based forestry resource collection method, characterized in that, The application is applied to a forestry resource collection system, and the collection method comprises the following steps: The remote sensing image of the target forest area is collected by a remote sensing device and subjected to feature recognition processing to obtain feature information of the remote sensing image; if the remote sensing image is a low-resolution image or a single-band image, a first restriction condition is determined as a low compression ratio and a low bandwidth value and single data transmission; if the remote sensing image is a high-resolution multi-spectral image or a hyperspectral image, the first restriction condition is determined as a high compression ratio and a high bandwidth value, and the data transmission is performed in multiple times; if there is an important geographical area or an abnormal area in the remote sensing image, a second restriction condition is determined as regional high-resolution transmission; if there are a large number of blank areas or repeated areas in the remote sensing image, data compression is performed by using the redundant information in the image; if the transmission path of the remote sensing image is long and the transmission network is congested, a third restriction condition is a multi-path transmission strategy; if the transmission path of the remote sensing image is short or the transmission network is not congested, the third restriction condition is a single-path transmission strategy; if the transmission task of the remote sensing image is disaster monitoring, a transmission protocol in a fourth restriction condition is a user datagram protocol; if the transmission task of the remote sensing image is periodic and offline monitoring, the transmission protocol is a transmission control protocol; based on the first restriction condition, the second restriction condition, the third restriction condition and the fourth restriction condition, a first target transmission strategy is obtained, and the remote sensing image is transmitted based on the first target transmission strategy; Sensor data is collected based on an Internet of Things sensor arranged in the target forest area; the sensor data comprises real-time soil data, real-time water data and real-time temperature data; based on the sensor data, a second target transmission strategy is selected, and the sensor data is transmitted based on the second target transmission strategy; The remote sensing feature data and the sensor data are preprocessed to obtain first multi-dimensional data; the first multi-dimensional data is subjected to data centralization processing to obtain a centralized data matrix; based on the centralized data matrix, a covariance matrix of the centralized data is calculated; the covariance matrix of the centralized data is subjected to eigenvalue decomposition to obtain eigenvalues and eigenvectors; a preset number of target eigenvectors are selected to form a first target dimension reduction matrix; the first multi-dimensional data is projected to the first target dimension reduction matrix to obtain first low-dimensional data; whether the first low-dimensional data retains target information is verified based on a three-dimensional scatter plot; if the verification is passed, the first low-dimensional data is taken as target low-dimensional data; if the verification is not passed, the first multi-dimensional data is subjected to second dimension reduction processing based on a second dimension reduction strategy to obtain second low-dimensional data; the second dimension reduction processing is nonlinear dimension reduction processing; target information in the second low-dimensional data is also verified based on a three-dimensional scatter plot, and the second low-dimensional data that passes the verification is taken as target low-dimensional data. Integrate the obtained remote sensing image and the sensor data to obtain integrated data; after decision-level data fusion processing of the low-dimensional data and the integrated data, input a target prediction model to obtain target forestry resource data and prediction result information; the target forestry resource data includes original collection data and prediction result information generated based on the original collection data; the prediction result information includes coverage information of a target forest area, health condition information of the target forest area, tree species distribution information and growth rate information of each tree species.

2. The forestry resource collection method based on the GIS technology according to claim 1, characterized in that, The decision level data fusion process satisfies the following relation: D fused = w low * D low + w full * D full ; wherein D fused represents the data after the decision level data fusion processing, w low and w full are weight values of the low-dimensional data and the integrated data, respectively, w low +w full= 1; D low represents the low-dimensional data, D full represents the integrated data; The target prediction model satisfies the following relationship: ; wherein, represents prediction result information, D fused represents data after decision level data fusion processing, represents model parameters, and f(·) represents a prediction function. The loss function of the target prediction model is: ; wherein, represents a predicted value, represents a true value, n represents a sample number, represents a square error of each sample.

3. The forestry resource collecting method based on GIS technology according to claim 2, characterized in that, The step of integrating the obtained remote sensing image and the sensor data to obtain integrated data comprises: After time synchronization processing of the obtained remote sensing image and the sensor data by interpolation processing, a plurality of image data pairs are obtained; the remote sensing image and the sensor data in the image data pair are consistent in time point; The remote sensing image in each image data pair is converted into a feature vector based on a preset deep learning model, and the sensor data is converted into standard sensor data through Z-Score standardization processing to obtain a multidimensional data set; The feature vector satisfies the following relationship: f I = f CNN (I); wherein, f CNN represents a preset deep learning model; f I represents a feature vector; I represents an input remote sensing image, and the dimension of I is (image height, image width, image channel number). The standard sensor data satisfy the following relation: ; wherein, and denote the mean and the standard deviation of the set of sensor data, respectively, denotes the i-th standard sensor data, denotes the i-th sensor data; The multidimensional data set is subjected to feature splicing processing to obtain integrated data.

4. The forestry resource collecting method based on the GIS technology according to claim 2, characterized in that, The step of obtaining target forestry resource data and prediction result information by decision-level data fusion processing of the low-dimensional data and the integrated data comprises: The low-dimensional data is input into a first preset prediction model to obtain coverage information of a target forest area; the first preset prediction model comprises an input layer, a hidden layer and an output layer; the input layer is used to receive the low-dimensional data; the hidden layer comprises a plurality of neurons, and a nonlinear activation function is used to learn the relationship between the low-dimensional data and the coverage information of the target forest area; the output layer is used to output the coverage information of the target forest area; The integrated data is input into a second preset prediction model to obtain health condition information of a target forest area, tree species distribution information and growth rate information of each tree species; the second preset prediction model comprises an input layer, a hidden layer and an output layer; the input layer is used to receive the integrated data; the hidden layer comprises three hidden sub-layers, and each hidden sub-layer extracts features of the integrated data through a combination of weighting and an activation function; the output layer comprises three output sub-layers, which are respectively used to output the health condition information of the target forest area, the tree species distribution information and the growth rate information of each tree species.

5. The forestry resource collecting method based on GIS technology according to claim 2, characterized in that, After the target forestry resource data and the prediction result information are obtained, the method further comprises: Based on the coverage information of the target forest area, the health condition information, the tree species distribution information and the growth rate information of each tree species, generate a forestry resource distribution map and a forestry resource prediction change map to formulate a forestry management target strategy.

6. The forestry resource collecting method based on GIS technology according to claim 1, characterized in that, The step of selecting a second target transmission strategy based on the sensor data and transmitting the sensor data based on the second target transmission strategy comprises: The sensor data is output to a preset data grading model for data grading processing, and data grading values of the sensor data are obtained respectively; the data grading values include high priority values, medium priority values and low priority values; A second target transmission strategy is selected based on the data grading values; The sensor data is transmitted based on the second target transmission strategy.

7. The forestry resource collecting method based on the GIS technology according to claim 6, characterized in that, The step of selecting the second target transmission strategy based on the data grading values includes: For data with a high priority value, user datagram protocol is used for transmission; For data with a medium priority value, transmission control protocol is used for transmission; For data with a low priority value, message queue telemetry transmission is used for transmission.

Images