An image acquisition distribution method based on a monitoring array

Abstract

The scheme provides an image acquisition distribution method based on a monitoring array, relates to the technical field of agricultural analysis, and comprises the following steps: establishing basic data information, connecting a geographic information system by a back-end computing platform to obtain satellite photos and DEM data, establishing a topographic map of a to-be-measured region, obtaining NDVI data of the to-be-measured region by satellite remote sensing, and storing the topographic map and the NDVI data into a platform memory of the back-end computing platform, system preposition setting, array preposition setting, monitoring execution, result processing, and result output. The system has the characteristics of being capable of being temporarily built and having low cost, can guarantee calculation accuracy and reliability, and can meet the needs of temporary building in remote areas in consideration of economy.

Description

[0001] This application is a divisional application of application number CN202211580904.7, filed on December 9, 2022, entitled "An Agricultural Sensing and Monitoring System and Its Application Method". Technical Field

[0002] This invention relates to the field of agricultural analysis technology, specifically to an image acquisition and allocation method based on a monitoring array. Background Technology

[0003] Smart agriculture applies Internet of Things (IoT) technology to traditional agriculture, using sensors and software to control agricultural production via mobile or computer platforms, making traditional agriculture more "intelligent." This includes precise sensing, control, and decision-making management.

[0004] The application of smart agriculture management in large areas of farmland often requires the participation of satellite remote sensing technology. However, the accuracy of satellite remote sensing is often poor, and it cannot achieve precise control and analysis compared with on-site images. Some regions have established farmland monitoring stations to collect, analyze and guide farmland production. However, the cost of farmland monitoring stations required for complete smart agriculture is high, which is difficult for some economically disadvantaged areas to afford. On the other hand, establishing simple monitoring stations will reduce the accuracy of data collection, affect the analysis results, and make it difficult to achieve accurate and high-definition results. Summary of the Invention

[0005] This invention proposes an agricultural sensing and monitoring system, an image acquisition and allocation method based on a monitoring array, a UAV state adjustment method, and a multi-view monitoring method based on a monitoring array. The system can be temporarily set up, resulting in low setup costs. The image acquisition and allocation method based on the monitoring array can quickly and stably acquire images of the farmland under test and process them in batches. The UAV state adjustment method can increase sampling stability through control methods. The multi-view monitoring method based on the monitoring array can improve the accuracy of image acquisition. When the area under test is in a remote region with weak GPS signals, the UAV can self-reset when it deviates. This application can be easily set up temporarily, solving the problem of high hardware support costs. It can ensure system accuracy and computational efficiency at a low cost, reducing hardware requirements.

[0006] The technical solution of the present invention is as follows: An agricultural sensing and monitoring system includes a monitoring array, which is regularly arranged in a monitoring area. The monitoring array is communicatively connected to a back-end computing platform. The monitoring array includes multiple sampling nodes. The sampling nodes adjust the distribution state of the area to be tested according to the actual conditions of the area to be tested. Each sampling node includes a power supply, a node processing module, a camera module, a node memory and a node storage and a node signal transceiver module. The camera module, the node storage and the node storage are connected to the node processing module. The power supply is connected to each module in the sampling node. The back-end computing platform includes a terminal processor, which is communicatively connected to a back-end storage unit. The back-end computing platform is used to analyze the data collected by the monitoring array. It also includes a field station, which comprises a main processing module, a ground memory, and a ground storage device. The main processing module is connected to the ground storage device, and the main processing module is connected to the ground memory. The field station is located near the area to be measured. The field station includes a field signal transceiver module. The node signal transceiver modules of the monitoring array are connected to the field signal transceiver module of the field station via a local area network. The field station is communicatively connected to the back-end computing platform. Each sampling node includes a dynamic task allocation module and an image task primary processing module. The dynamic task allocation module controls and connects to the camera module, and the image task primary processing module communicatively connects to both the dynamic task allocation module and the camera module. The camera module transmits the acquired raw image data to the image task primary processing module, and the image task primary processing module transmits the processed image information to the dynamic task allocation module. The image task primary processing module connects to other sampling nodes, site stations, and the back-end computing platform in the monitoring array via the association docking unit, the local area network unit, and the internet unit, respectively. The dynamic task allocation module connects to other sampling nodes and site stations in the monitoring array via the association docking unit and the local area network unit, respectively. The site station includes a macroscopic status task allocation module and an image task secondary processing module. The macroscopic status task allocation module is used to adjust the overall field of view of the monitoring array. The macroscopic status task allocation module and the image task secondary processing module are connected to the sampling node through the site signal transceiver module.

[0007] An image acquisition and allocation method based on a monitoring array, which is based on an agricultural sensing and monitoring system. This includes establishing basic data information, connecting the backend computing platform with the geographic information system to obtain satellite photos and DEM data to create a topographic map of the area to be measured, and using satellite remote sensing to obtain NDVI data of the area to be measured, storing the topographic map and NDVI data in the platform memory of the backend computing platform; The system pre-configures a unique code for each sampling node and sets a pre-command identification code based on the code of each sampling node. The code and pre-command identification code of each sampling node are then stored in the platform memory of the back-end computing platform. The main processing module divides the ground storage into a site information database, a ground key database, an array information database, a raw image database, and a secondary image database. The site signal transceiver module of the site station transmits topographic maps and NDVI data to the site information database, and transmits the code and pre-command identification code of each sampling node to the ground key database of the site station. The node processing module divides the node storage into a command library, a node local information library, and a raw visual library. Through an external device, the node processing module in each sampling node inputs the code and pre-command identification code corresponding to the sampling node, as well as the code and pre-command identification code corresponding to the nearest neighboring sampling node, into the node's local information database. The array pre-setting involves calling the macroscopic status task allocation module to set the macroscopic matrix distribution of the monitoring array based on the topographic map and NDVI data, setting the arrangement information of the sampling nodes, storing the arrangement information in the array information database of the site station and the platform memory of the back-end computing platform, and enabling the monitoring array based on the arrangement information. The monitoring is performed by a fixed monitoring array that takes pictures of a selected area. The adjacent sampling nodes have overlapping pictures at the boundary of the pictures, which are used as reference blocks for image stitching. The monitoring array sends the image data to the site station through the local area network unit. Each image is numbered according to the code of the sampling node. The site station stores the image data in the original image database. In the result processing, the site station retrieves image information from the original image database, applies image stitching technology to obtain image data of the entire area to be tested, and stores the integrated image data into a secondary image database. The results are output, the integrated image data is identified and analyzed, the analysis results are output through the human-computer interaction unit, and the original image database and the secondary image database are called. The original image data directly sent by the monitoring array and the secondary image data obtained by stitching are transmitted to the back-end computing platform through the site signal transceiver module for storage and further image analysis.

[0008] A method for adjusting the state of a drone, based on an agricultural sensing and monitoring system, involves determining and placing drone monitoring nodes using GPS. The method is characterized by a wireless rangefinder confirming the distance to adjacent drone monitoring nodes, a node processing module sending the ranging result to a physical motion analysis module, and the physical motion analysis module adjusting the drone's flight state according to the ranging command from the drone's power controller.

[0009] A multi-view monitoring method based on a monitoring array, which is based on an agricultural sensing and monitoring system. Step 1: Each sampling node receives image information collected by neighboring sampling nodes via a temporary matrix network and performs overlap analysis. The node processing module places the obtained image information and the image information sent by neighboring sampling nodes into the node memory, and performs a first-level overlap analysis on each. Overlapping images are extracted, counted, and placed in the node memory. Overlapping images that have undergone one count are analyzed pairwise to extract mutually overlapping image information and added with a technical mark, and stored in the node memory. Overlapping images that have undergone two counts are analyzed pairwise, and the image information that overlaps again is added with a count mark. This process is repeated μ times until there is no overlapping image information. The node processing module then crops the images collected locally and obtains individual image information for each region according to the number of overlaps. Step 2: Treat the above-mentioned interactive sampling nodes as a whole first-level unit. Using this first-level unit as the center, receive the second-level images sent by the sampling nodes of the neighboring first-level units through a temporary array network. Perform overlap analysis on the image fragments after counting in Step 1 and the second-level images pairwise. Extract the images that overlap, add a count mark, and place them in the memory of the node. Repeat the operation of Step 1. Step 3: Take the sampling node that interacted with the first-level unit in Step 2 as the second-level unit, and use the temporary array network to receive the three-layer images sent by the sampling nodes of the neighboring second-level units. Perform overlap analysis on the image fragments obtained in Step 2 and the three-layer images pairwise, and repeat the operation of Step 2. Step four: Repeat steps one to three ξ times until there is no more overlapping image information. The node processing module records the overlapping information of each region of the image acquired by the node and stores it in the original visual library of the local machine. At the same time, it calls the local area network unit to send it to the site station. Step 5: The node processing module calls the dynamic task allocation module and the temporary array network to connect with the sampling nodes that have interaction, and performs corresponding multi-view camera analysis on image fragments with different degrees of overlap based on the number of counts.

[0010] The working principle and beneficial effects of this invention are as follows: The monitoring array in this application can be temporarily set up. In conjunction with the site station and the Internet, it can complete the collection and preliminary processing of information. The monitoring array consists of multiple sampling nodes, which are generally drones. When it is necessary to analyze farmland, the drones are placed in the airspace above the farmland in a reasonable manner, so that the area to be monitored can be photographed and sampled continuously.

[0011] The image acquisition and allocation method of this application system can issue commands to multiple sampling nodes through a single site station, forming a temporary network among the sampling nodes. This can increase the correlation and cooperation among the sampling nodes and enable stable acquisition of information about the farmland under test by relying on the monitoring array.

[0012] When the drone monitoring node shakes in the drone state adjustment method of this application, the position can be corrected by sensing the position of surrounding drones. When the test site is remote, the position correction of the drone can be achieved without relying on GPS and the Internet.

[0013] The multi-view monitoring method based on monitoring arrays involves multiple monocular cameras working together through a temporary array network to form a whole, conducting multi-view surveys of overlapping areas and increasing the accuracy of observation results. Attached Figure Description

[0014] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0015] Figure 1 This is a structural block diagram of the agricultural sensing and monitoring system of the present invention; Figure 2 This is a block diagram of the monitoring array in an agricultural sensing and monitoring system. Figure 3 This is a flowchart of the image acquisition and allocation method based on a monitoring array in this invention; Figure 4 This is a flowchart of the array pre-setting in the image acquisition and allocation method based on the monitoring array of the present invention; Figure 5 This is a flowchart illustrating the detection execution process in the image acquisition and allocation method based on a monitoring array according to the present invention. Figure 6 This is a flowchart illustrating the result analysis in the image acquisition and allocation method based on a monitoring array according to the present invention. Detailed Implementation

[0016] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0017] An agricultural sensing and monitoring system includes a monitoring array, which is regularly arranged in the area to be monitored. The monitoring array is communicatively connected to a back-end computing platform. The monitoring array includes multiple sampling nodes. The sampling nodes adjust the distribution state of the area to be measured according to the actual conditions of the area to be measured. Each sampling node includes a power supply, a node processing module, a camera module, a node memory and a node signal transceiver module. The camera module, node memory and node storage are connected to the node processing module. The power supply forms a power supply connection with each module in the sampling node. The back-end computing platform includes a terminal processor, which is communicatively connected to a back-end storage unit. The back-end computing platform is used to analyze the data collected by the monitoring array. It also includes a field station, which includes a main processing module, ground memory, and ground storage. The main processing module is connected to the ground storage and the ground memory. The field station is set up near the area to be measured. The field station includes a field signal transceiver module. The node signal transceiver modules of the monitoring array are connected to the field signal transceiver module of the field station via a local area network. The field station is connected to the back-end computing platform.

[0018] Traditional agricultural image acquisition methods involve piloting drones to fly over the area to be measured and taking pictures of the site while in motion. This method has a large error and the images are difficult to process. In this application, each sampling node takes a fixed image of a single area. The monitoring array is stationary relative to the farmland, and the position of each sampling node is fixed, resulting in more stable images. The site station plays a macro-management role over the monitoring array. Since the computing speed of the monitoring array is limited, the site station can stably store and further analyze the images acquired by the monitoring array. The site station can connect to the back-end computing platform via a network to provide data and computing support. Generally, a wired network is used, but a cellular network can also be used.

[0019] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. The specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the subsequent method embodiments.

[0020] In the embodiments provided by this invention, it should be understood that the disclosed apparatus / terminal devices and methods can be implemented in other ways. For example, the apparatus / terminal device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the shown or discussed mutual couplings or direct couplings or communication connections may be through some interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms. Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, i.e., they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0021] The node signal transceiver module includes an Internet unit, a local area network unit, and an association docking unit. The sampling node communicates with the site station through the local area network unit. The Internet unit is used to connect to the Internet and communicate with the back-end computing platform. The association docking unit is used to form a communication connection with the association docking units of other sampling nodes in the monitoring array. Each sampling node forms a temporary array network through the association docking unit. The temporary array network is used for information exchange within the monitoring array.

[0022] This system can provide multiple communication methods. When the cellular network is in good condition, the sampling nodes can directly connect to the back-end computing platform. The local area network unit is the communication method between the site station and each sampling node. Information exchange between each sampling node can also be realized through the association docking unit. The association docking unit can use devices such as wireless bridges. In this application, each sampling node can form a temporary array network with the help of the association docking units of each sampling node, so that each node in the monitoring array can communicate with each other, connect each sampling node together, perform joint calculations, and treat the monitoring array as a whole to facilitate information exchange.

[0023] Each sampling node includes a UAV monitoring node, which also includes a perception aggregation module and a physical motion analysis module. The perception aggregation module is signal-connected to the signal input terminal of the physical motion analysis module. The perception aggregation module is used to collect the status information of the UAV monitoring node. The physical motion analysis module is signal-connected to the UAV power controller. The UAV power controller controls the actuator motor of the corresponding UAV monitoring node. The physical motion analysis module also communicates with the site station and other sampling nodes through the node signal transceiver module.

[0024] The sensing aggregation module also includes a gyroscope and GPS positioning device for the sampling node's location needs, as well as temperature and humidity sensors, barometric pressure sensors, etc. This module can collect real-time atmospheric environmental conditions and transmit them to the site station. It can gather environmental information and, through the data from the sensing module, determine weather changes, adjusting the drone's detection node status in real time when problems occur. The physical motion analysis module allows the drone to automatically adjust its flight status based on changes in the sensing aggregation module's data. This reduces reliance on communication equipment and increases the level of intelligence.

[0025] The perception aggregation module includes a ranging unit set up at the UAV monitoring node. The ranging unit includes a wireless signal rangefinder evenly distributed at the center of the UAV monitoring node. The output of the wireless signal rangefinder is connected to the signal input of the physical motion analysis module. The wireless signal rangefinder establishes an identification channel with the ranging units of adjacent UAV monitoring nodes.

[0026] Each sampling node's wireless rangefinder connects to the nearest surrounding node. For example, in a checkerboard-like rectangular matrix, each node is in the center of a rectangular grid. Each drone's wireless rangefinder establishes an information channel with the wireless rangefinders of its four surrounding nodes. In a hexagonal matrix, each node is at the center of a hexagon, and each drone's wireless rangefinder establishes an information channel with the wireless rangefinders of its six surrounding nodes. Each node only needs to calculate its own distance, exchange data through a temporary array network, and make further adjustments. This method reduces the computational burden on the wireless rangefinder and node hardware; calculations are performed based on the data, eliminating the need for excessive analysis and judgment.

[0027] Each sampling node includes a dynamic task allocation module and an image task primary processing module. The dynamic task allocation module controls the connection to the camera module, and the image task primary processing module communicates with the dynamic task allocation module and the camera module. The camera module transmits the acquired raw image data to the image task primary processing module, and the image task primary processing module transmits the processed image information to the dynamic task allocation module. The image task primary processing module connects to other sampling nodes, site stations, and the back-end computing platform in the monitoring array via an association docking unit, a local area network unit, and an Internet unit, respectively. The dynamic task allocation module connects to other sampling nodes and site stations in the monitoring array via an association docking unit and a local area network unit, respectively. The site station includes a macroscopic status task allocation module and an image task secondary processing module. The macroscopic status task allocation module is used to adjust the overall layout of the monitoring array. The macroscopic status task allocation module and the image task secondary processing module connect to the sampling nodes via a site signal transceiver module.

[0028] The image task primary processing module can be used for image package analysis. The dynamic task allocation module can adjust the state of the camera module of the sampling node according to the image information. In this scheme, image processing and mechanical control calculations are performed separately from the node processing module, which can increase the calculation efficiency. Since the sampling node needs to analyze and judge the image information and intelligently adjust the shooting mode, the image task primary processing module is directly connected to the camera module to receive the image data collected by the camera module, which can reduce the calculation process. The original image information is compressed and the original image and the compressed image are transferred to the node memory. The compressed image facilitates information interaction between the sampling nodes. The physical motion analysis module can adjust the shooting in time according to the data of the image task primary processing module. Block processing can reduce the calculation pressure to meet the preliminary image analysis and reduce the hardware pressure of the monitoring array's cooperative calculation. The macroscopic status task allocation module in the site station is used to adjust the arrangement of the monitoring array and is connected to a manned-machine interface device. The operator can see the image information captured by each sampling node and make suggestions on the field of view and angle of the camera module based on the image clarity and overlap. The macroscopic status task allocation module will call the pre-command identification code of the corresponding sampling node and send a command to the dynamic task allocation module of the corresponding node to make corresponding adjustments.

[0029] In the embodiments of this invention, the functional units can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory, random access memory, electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content contained in computer-readable media may be appropriately added to or subtracted from the requirements of legislation and patent practice in a jurisdiction. For example, in some jurisdictions, computer-readable media may not include electrical carrier signals and telecommunication signals, in accordance with legislation and patent practice.

[0030] An image acquisition and allocation method based on a monitoring array, which is based on an agricultural sensing and monitoring system. This includes establishing basic data information, connecting the backend computing platform with the geographic information system to obtain satellite photos and DEM data to create a topographic map of the area to be measured, and using satellite remote sensing to obtain NDVI data of the area to be measured, storing the topographic map and NDVI data in the platform memory of the backend computing platform; The backend computing platform connects to the geographic information system. After authorization, the data can be directly downloaded for reference, and the data from the geographic information system can be used as the basis for subsequent models.

[0031] The system pre-configuration involves uniquely encoding each sampling node and setting a pre-command identification code based on the code of each sampling node. The code and pre-command identification code of each sampling node are stored in the platform memory of the back-end computing platform. The main processing module divides the ground memory into a site information database, a ground key database, an array information database, a raw image database, and a secondary image database. It receives topographic maps and NDVI data using the site signal transceiver module of the site station and saves them to the site information database. It also receives the code and pre-command identification code of each sampling node using the site signal transceiver module of the site station and saves them to the ground key database of the site station. Each sampling node is uniquely named using an encoding, and a pre-command identification code is assigned to each code. This code serves as a switch for the sampling node to confirm receipt of commands, i.e., a start command. At the same time, the pre-command identification code also serves as a recognition mark for other nodes in the site station or monitoring array to identify a specific node. The machine's unique code can be used as the pre-command identification code, or the unique code can be encrypted according to rules.

[0032] The site information database stores geographic information data, topographic maps, NDVI data, historical climate conditions, crop types and yields, and disaster information for the farmland to be monitored by the module. The ground key database stores the encoding information and pre-command identification code of each sampling node, and can also store the unique code of each machine and the conversion rules for the pre-command identification code. The array information data stores the machine number corresponding to each node in the arranged monitoring array and the arrangement method of the monitoring array. The raw image database stores the raw image data directly sent by the monitoring array and the compressed image data. The secondary image database stores the processed overall farmland image information.

[0033] The node processing module divides the node memory into a command library, a node local information library, and a raw visual library. Input can be achieved through external devices, such as a human-machine interface module, either manually or by scanning a code. Each sampling node's processing module inputs the corresponding code and preceding command identification code of that sampling node, as well as the codes and preceding command identification codes of its nearest neighboring sampling nodes, into the node local information library. The command library stores instruction programs that control the actions of the sampling nodes, including adaptive adjustments to the camera and sampling node actions based on the acquired images and data from the sensor collection module, and is used to analyze the actions sent by the site station.

[0034] Prepare for data collection and analysis by contacting the backend computing platform and field stations, and establishing connections between the backend computing platform, field stations, and monitoring arrays.

[0035] The array pre-setting calls the macro-state task allocation module to set the macro-matrix distribution of the monitoring array based on the topographic map and NDVI data, set the arrangement information of the sampling nodes, and store the arrangement information in the array information database of the site station and the platform memory of the back-end computing platform, and activate the monitoring array based on the arrangement information. Depending on the layout of the monitoring array, sampling nodes can be arranged horizontally or at equal height. The density of sampling nodes is proportional to the NDVI value. Alternatively, operators can manually set the layout based on experience and the actual farmland conditions, and then set up the sampling nodes according to the layout information. Common arrangement methods are checkerboard or hexagonal grid distributions. Checkerboard grids are the simplest and easiest to adjust, with the lowest requirement for overlap between adjacent images, reducing the number of sampling nodes needed and making them suitable for large plains areas, increasing the field of view. Hexagonal grid distributions offer denser sampling and higher overlap, making them more suitable for areas with many roads and rivers. Triangular distributions can also be used, offering high coverage, allowing for more multi-camera acquisition, avoiding errors over a wider area, and pursuing higher precision; they are suitable for areas with high vegetation coverage.

[0036] The monitoring array takes pictures of a selected area. There is overlap in the pictures taken by adjacent sampling nodes at the boundary of the pictures, which is used as a reference block for image stitching. The monitoring array sends the image data to the site station through the local area network unit. Each image is numbered according to the code of the sampling node. The site station stores the image data in the original image database. During monitoring, each sampling node analyzes the selected agricultural area, uses the overlapping position as a reference area and sets up a beacon. The size of the compressed image is adjusted according to the actual horizontal height of the sampling node. The overlap of the sampling areas of other sampling nodes is compared and multi-view analysis is performed with the help of a temporary array network. The sampled image information is sent to the field station, where the image is further processed.

[0037] In the result processing, the site station retrieves image information from the original image database, applies image stitching technology to obtain image data of the entire area to be measured, and stores the integrated image data into the secondary image database. The site station combines images according to the arrangement of sampling nodes, which can effectively reduce the dependence on the overlap rate. Traditional image stitching recognition requires an overlap rate of more than 70% to obtain the relationship between images based on the overlap. However, the application of this scheme only requires an overlap rate of more than 40%. By using a dependency method, the corresponding images can be filled in the appropriate position, which can replace the recognition of overlap, reduce the iteration in the calculation process, and simplify the calculation process.

[0038] The results are output, and the integrated image data is identified and analyzed. The analysis results are output through the human-computer interaction unit, and the original image database and the secondary image database are called. The original image data directly sent by the monitoring array and the secondary image data obtained by stitching are transmitted to the back-end computing platform through the site signal transceiver module for storage and further image analysis.

[0039] The algorithm of multi-view camera is used to identify the stitched image. Except for the edge of the farmland, most areas are covered by overlapping images. Therefore, multi-view operation can be performed on the obtained secondary images. The collected secondary image information is then transmitted to the back-end computing platform. The back-end computing platform calls the information data on pests and diseases, climate regulation, etc. in the back-end database to identify and analyze the collected images.

[0040] The array pre-configuration steps include: S3-1, the back-end computing platform, sets the macro-matrix distribution mode of the monitoring array based on the topographic map and NDVI data, stores the macro-matrix distribution mode in the back-end memory, and sends it to the site station via the Internet. The main processing module stores the macro-matrix distribution mode in the array information database. S3-2, the main processing module calls topographic map data from the site information database, extracts the encoding information of each node from the ground key database, and calls the macroscopic matrix distribution method of the monitoring array from the array information database. The main processing module places the above information in the ground memory. The main processing module uses the topographic map data as the underlying layer of the macroscopic distribution method of the monitoring array. Based on the topographic map data, it samples nodes in the matrix distribution map to obtain the geographic coordinates of each node. It assigns a geographic coordinate to each node with its corresponding code, thereby obtaining the distribution information data of the code of the sampled node on the topographic map. The separate information data of the code of the sampled node on the topographic map is stored in the array information database. S3-3, the main processing module extracts the pre-command identification code information from the ground key database to the ground memory, assigns the pre-command identification code to the coordinate information corresponding to the sampling node, so that the two correspond one-to-one. The main processing module calls the site signal transceiver module to send the coordinate data with the pre-command identification code to the monitoring array. S3-4, the node processing module in each sampling node calls the pre-command identification code in the node's local information database to memory. The node processing module calls the local area network unit to receive the coordinate data sent by S3-3 to memory, matches the coordinate data with the pre-command identification code, and obtains the geographic coordinate information corresponding to the node. The node processing module stores the obtained geographic coordinate information in the node's local information database, thereby obtaining the spatial distribution location of each sampling node in the monitoring array.

[0041] The monitoring execution steps include, S4-1, the node processing module in each sampling node calls the lens control module to adjust the wide-angle of the camera module so that the overlap of adjacent images is greater than the set value. The lens control module sends a shooting frequency command to the camera module. S4-2, the node processing module copies the image information acquired by the camera module into two copies, one of which is stored in the original visual library and the other is placed directly in the node memory. The node processing module calls the image compression algorithm of the first-level image task processing module to the memory, and compresses the original image data in the node memory according to the image compression algorithm. The node processing module calls the pre-command identification code of the local and neighboring sampling nodes in the node's local information library and places it in the node memory. In this scheme, the steps of image task and processing module to extract images can be eliminated, simplifying the calculation process. S4-3, the node processing module arranges the pre-command identification codes of nearby sampling nodes in an OR relationship and uses them as the pre-commands of the image information to be sent. The node processing module then uses a temporary array network to send the compressed image information to the N nearest sampling nodes. S4-4, the node processing module adds the corresponding pre-command identification code of the sampling node to the original image information and sends it to the site station through the local area network unit. The main processing module calls the site signal transceiver module to receive the image information with the pre-command identification code and stores the image information with the pre-command identification code into the original image database. S4-5, the node processing module compares the compressed image data sent by the nearest N sampling nodes received by the docking unit with the image on the local machine to check the overlap. Then, it calls the dynamic task allocation module to adjust the wide-angle range of the lens according to the overlap comparison result.

[0042] S4-5 includes S4-5-1, where the node processing module calls the location information of neighboring sampling nodes and the preceding command identification code from the node's local information database and stores them in the node's memory; and the node processing module calls the overlap judgment algorithm information from the image task's first-level processing module and stores it in the node's memory. S4-5-2, The image task first-level processing module selects a received compressed image, proportionally changes the size of the local image based on the horizontal height difference between the local machine and the corresponding sampling node of the compressed image, obtains the position of the sampling node corresponding to the image based on the pre-command identification code of the compressed image, and translates the image acquired by the local machine towards the direction of the corresponding sampling node of the comparison image according to the image scale. When the reference beacon coincides, the proportion of the area of ​​the overlapping region to the area of ​​the local machine's acquired image is calculated, and it is determined whether the overlapping region is within the overlap reference threshold. S4-5-3: If yes, record the wide-angle information in that direction as normal data; otherwise, if the overlap is greater than the overlap reference threshold, record that the wide-angle in that direction is too large; if the overlap is less than the overlap reference threshold, record that the wide-angle in that direction is too small. Then, process the next sampling node to send the compressed image and repeat step S4-5-2. If everything is normal, skip step S4-5-4. In section S4-5-4, the dynamic task allocation module weights the overlap of each sampling node based on the recorded results and overlap ratio in section S4-5-3. The weighted result yields the adjustment degree of the camera module's wide-angle, which is then commanded to be adjusted. Neighboring sampling nodes are connected using the node's north (n) and west (w) as two dimensions for calculating overlap, determining the number of sampling nodes (x). Each sampling node has a weight of 1 / x. The weighted value is the cosine of the node connection and the calculated dimension. This weighted value represents the weight of a neighboring sampling node relative to the selected sampling node in the selected dimension. Multiplying this weighted value by the overlap of the two images gives the overlap weight of that neighboring sampling node. The absolute values ​​of the overlap weights of all sampling nodes are summed to obtain the calculated overlap of the selected sampling node's field of view. This result is compared with the overlap threshold to adjust the camera module's wide-angle angle. If the difference is significant, the dynamic task allocation module of the sampling node can be manually controlled by the macroscopic status task allocation module of the site station to send command information to adjust the camera mode.

[0043] When calculating overlap, the image size needs to be adjusted so that each image is in the same coordinate system. For example, images selected by sampling nodes with a wide angle or higher horizontal height need to be enlarged by the image task and processing module so that the reference object or beacon coincides with the size of the center sampling node. Similarly, if the horizontal height is low, the image needs to be scaled down for comparison and the scaled image needs to be stored in the original visual library.

[0044] Result processing includes, S5-1, the main processing module calls the matrix filling algorithm of the image secondary processing module and places it in the ground memory. The main processing module calls the arrangement information of the monitoring array in the array information database and places it in the ground memory. The matrix filling algorithm establishes a matrix frame that matches the arrangement rules of the monitoring array. There are areas to be filled at the nodes of this matrix frame. The main processing module calls the pre-command identification code of each sampling node in the ground key database and places it in the memory. The image secondary processing module sets the corresponding pre-command identification code in the corresponding area to be filled according to the matrix frame established in S5-1. S5-2, the image secondary processing module matches the pre-command identification code of the area to be filled with the pre-command identification code of each image information, thereby filling the image into the matrix frame; S5-3, the main processing module calls the overlap processing algorithm of the image secondary processing module to mark the overlap layer number of the overlapping areas of adjacent image information, obtain complete site information, and store it in the secondary image database.

[0045] This method can be used to first establish an image framework, with each position that needs to be filled corresponding to a sampling node. Based on the pre-command identification code, the image of each sampling node is scaled and placed in the corresponding position in the framework. The overlap of each region is calculated to obtain the overlap of each part in the secondary image information processed by the site station, and the sampling nodes involved in each overlap are recorded to facilitate subsequent multi-view analysis.

[0046] Between the array pre-setting and monitoring execution steps, a beacon is also established to ensure that the overlap and delay of the sampling at each sampling node in the monitoring array are appropriate. Based on the macroscopic matrix distribution of the monitoring array, image information that can be released at a fixed frequency periodically varies is placed at the intersection of every two sampling nodes.

[0047] The beacon emits different light signals at a fixed frequency. The simplest way is to flash two or more different colors or numbers. Alternatively, the beacon's changing rules can be pre-written into the local information database of the site station and sampling nodes by emitting gradually changing light. The sampling nodes capture the beacon and compare it with other sampling nodes through a temporary array network. The image task's first-level processing module compares the acquired beacon information with the beacon information in the images of other sampling nodes and adjusts the shooting frequency accordingly.

[0048] A method for adjusting the state of a drone is disclosed. This method is based on an agricultural sensing and monitoring system. The location of the drone monitoring node is determined and placed according to GPS. A wireless signal rangefinder confirms the distance to the neighboring drone monitoring nodes. The node processing module sends the ranging result to the physical motion analysis module. The physical motion analysis module adjusts the drone's flight state according to the ranging command from the drone's power controller.

[0049] In general, drones rely on GPS for positioning. However, in mountainous areas, GPS signals are weak and unreliable. Therefore, a temporary array network is used to correct node positions in a timely manner. A ranging unit within the sensing system, typically a wireless rangefinder but also potentially an infrared or ultrasonic rangefinder, determines changes in the sampling node's position based on collected distance information, allowing for adaptive adjustments. Distance change data can also be transmitted to the field station via a local area network unit, enabling operators to send real-time commands to the sampling nodes based on the distance information.

[0050] A multi-view monitoring method based on a monitoring array, which is based on an agricultural sensing and monitoring system. Step 1: Each sampling node receives image information collected by neighboring sampling nodes via a temporary matrix network and performs overlap analysis. The node processing module places the obtained image information and the image information sent by the surrounding sampling nodes into the node memory, and performs a first-level overlap analysis on each of them. Overlapping images are extracted, counted, and placed in the node memory. Overlapping images that have been counted once are analyzed pairwise to extract mutually overlapping image information and add a technical mark, which is then stored in the node memory. Overlapping images that have been counted twice are analyzed pairwise, and the image information that overlaps again is marked again. This process is repeated μ times until there is no overlapping image information. The node processing module then crops the images collected by the node and obtains individual image information for each region according to the number of overlaps. Step 2: Treat the above-mentioned interactive sampling nodes as a whole first-level unit. Using this first-level unit as the center, receive the second-level images sent by the sampling nodes of the neighboring first-level units through a temporary array network. Perform overlap analysis on the image fragments obtained in Step 1 and the second-level images pairwise. Extract the images that overlap, add a count mark, and place them in the node memory. Repeat Step 1. Step 3: Take the sampling node that interacted with the first-level unit in Step 2 as the second-level unit, and use the temporary array network to receive the three-layer images sent by the sampling nodes of the neighboring second-level units. Perform overlap analysis on the image fragments obtained in Step 2 and the three-layer images pairwise, and repeat the operation of Step 2. Step 4: Repeat steps 1 to 3 ξ times until there is no more overlapping image information. The node processing module records the overlapping information of each region of the image acquired by the node and stores it in the original visual library of the local machine. At the same time, it calls the local area network unit to send it to the site station. Step 5: The node processing module calls the dynamic task allocation module and the temporary array network to connect with the sampling nodes that have interaction. For image fragments with different degrees of overlap, the module performs corresponding multi-view camera analysis based on the number of counts.

[0051] The number of overlaps in overlapping areas is marked, and the encoding of the sampling nodes involved is recorded each time the technique is performed, which facilitates the selection of nodes during multi-view imaging.

[0052] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An image acquisition and allocation method based on a monitoring array, characterized in that, This method is based on an agricultural sensing and monitoring system, which consists of a monitoring array, a back-end computing platform, and a field station. The monitoring array comprises multiple sampling nodes and is used to perform sensing and monitoring in the area to be monitored. The back-end computing platform is communicatively connected to the monitoring array and is used to analyze the data collected by the monitoring array. The field station is communicatively connected to both the monitoring array and the back-end computing platform. The field station is used to control the status of the monitoring array on-site, analyze the collected data on-site, analyze the data transmitted by the back-end computing platform, and adapt the information transmitted by the back-end computing platform to the information transmitted by the monitoring array. The image acquisition and allocation method based on the monitoring array includes: The equipment and system operation preparation includes the back-end computing platform acquiring satellite remote sensing data and sending it to the site station, encoding each sampling node in the monitoring array so that the site station can control each sampling node through encoding in the future, thereby realizing the overall control of the monitoring array, and the site station arranging the monitoring array according to the satellite remote sensing data; The monitoring is performed by a fixed monitoring array that takes pictures of a selected area. The adjacent sampling nodes have overlapping pictures at the boundary of the pictures, which are used as reference blocks for image stitching. The monitoring array sends the image data to the site station through the local area network unit. Each image is numbered according to the code of the sampling node. The site station stores the image data in the original image database. In the result processing, the site station retrieves image information from the original image database, applies image stitching technology to obtain image data of the entire area to be tested, and stores the integrated image data into a secondary image database. The operational preparation of the equipment and system includes, The system pre-configures a unique code for each sampling node and sets a pre-command identification code based on the code of each sampling node. The code and pre-command identification code of each sampling node are then stored in the platform memory of the back-end computing platform. The main processing module divides the ground storage into a site information database, a ground key database, an array information database, a raw image database, and a secondary image database. The site station receives topographic maps and NDVI data using its site signal transceiver module and saves them to the site information database. It also receives the code and pre-command identification code of each sampling node using the same module and saves them to the site station's ground key database. The node processing module divides the node storage into a command library, a node native information library, and a raw visual library. Through an external device, the node processing module in each sampling node inputs the code and pre-command identification code corresponding to the sampling node, as well as the code and pre-command identification code corresponding to the nearest neighboring sampling node, into the node's local information database.

2. The image acquisition and allocation method based on a monitoring array according to claim 1, characterized in that, The preparation for the operation of the equipment and system includes establishing basic data information, connecting the back-end computing platform with the geographic information system to obtain satellite photos and DEM data to create a topographic map of the area to be measured, and using satellite remote sensing to obtain NDVI data of the area to be measured, and storing the topographic map and NDVI data in the platform memory of the back-end computing platform.

3. The image acquisition and allocation method based on a monitoring array according to claim 1, characterized in that, The preparation for the operation of the equipment and system includes array pre-setting, calling the macro-state task allocation module, setting the macro-matrix distribution mode of the monitoring array according to the topographic map and NDVI data, setting the arrangement information of the sampling nodes, storing the arrangement information in the array information database of the site station and the platform memory of the back-end computing platform, and activating the monitoring array according to the arrangement information.

4. The image acquisition and allocation method based on a monitoring array according to claim 3, characterized in that, The array pre-configuration steps include: S3-1, the back-end computing platform sets the macro-matrix distribution mode of the monitoring array according to the topographic map and NDVI data, stores the macro-matrix distribution mode in the back-end memory, and sends it to the site station via the Internet. The main processing module stores the macro-matrix distribution mode in the array information database. S3-2, the main processing module calls topographic map data from the site information database, extracts the encoding information of each node from the ground key database, calls the macroscopic matrix distribution method of the monitoring array from the array information database, and places the above information in the ground memory. The main processing module uses the topographic map data as the underlying layer of the macroscopic distribution method of the monitoring array, samples nodes in the matrix distribution map based on the topographic map data, obtains the geographic coordinates of each node, assigns a geographic coordinate to each node with its corresponding encoding, thereby obtaining the distribution information data of the encoding of the sampled nodes on the topographic map, and stores the information data of the encoding of the sampled nodes on the topographic map into the array information database. S3-3, the main processing module extracts the pre-command identification code information from the ground key database and stores it in the ground memory. It then assigns the pre-command identification code to the coordinate information corresponding to the sampling node, so that the two correspond one-to-one. The main processing module calls the site signal transceiver module to send the coordinate data with the pre-command identification code to the monitoring array. In step S3-4, the node processing module in each sampling node calls the pre-command identification code in the node's local information database to memory. The node processing module receives the coordinate data sent by the site station in step S3-3 from the local network unit and sends it to the node's memory. The coordinate data is matched with the pre-command identification code to obtain the geographic coordinate information corresponding to the node. The node processing module stores the obtained geographic coordinate information in the node's local information database, thereby obtaining the spatial distribution location of each sampling node in the monitoring array.

5. The image acquisition and allocation method based on a monitoring array according to claim 1, characterized in that, The monitoring execution steps include, S4-1, the node processing module in each sampling node calls the lens control module to adjust the wide-angle of the camera module so that the overlap of adjacent images is greater than a set value, and the lens control module issues a shooting frequency command to the camera module. S4-2, the node processing module copies the image information acquired by the camera module into two copies, one copy is stored in the original visual library, and the other copy is placed directly in the node memory. The node processing module calls the image compression algorithm of the image task first-level processing module to the memory, and compresses the original image data in the node memory according to the image compression algorithm. The node processing module calls the pre-command identification code of the local and neighboring sampling nodes in the node local information library and places it in the node memory. S4-3, the node processing module arranges the pre-command identification codes of nearby sampling nodes in an OR relationship and uses them as the pre-commands of the image information to be sent. The node processing module then uses a temporary array network to send the compressed image information to the N nearest sampling nodes. S4-4, the node processing module attaches the corresponding pre-command identification code of the sampling node to the original image information and sends it to the site station through the local area network unit. The main processing module calls the site signal transceiver module to receive the image information with the pre-command identification code and stores the image information with the pre-command identification code into the original image database. S4-5, the node processing module compares the compressed image data sent by the nearest N sampling nodes received by the docking unit with the image on the local machine to determine the degree of overlap, and calls the dynamic task allocation module to adjust the wide-angle range of the lens according to the degree of overlap comparison results.

6. The image acquisition and allocation method based on a monitoring array according to claim 1, characterized in that, The result processing includes, S5-1, the main processing module calls the matrix filling algorithm of the image secondary processing module and places it in the ground memory. The main processing module calls the arrangement information of the monitoring array in the array information database and places it in the ground memory. The matrix filling algorithm establishes a matrix frame that matches the arrangement rules of the monitoring array. There are areas to be filled at the nodes of the matrix frame. The main processing module calls the pre-command identification code of each sampling node in the ground key database and places it in the memory. The image secondary processing module sets the corresponding pre-command identification code in the corresponding area to be filled according to the matrix frame established in S5-1. S5-2, the image secondary processing module matches the pre-command identification code of the area to be filled with the pre-command identification code of each image information, thereby filling the image into the matrix frame; S5-3, the main processing module calls the overlap processing algorithm of the image secondary processing module to mark the overlap layer number of the overlapping areas of adjacent image information, obtain complete site information, and store it in the secondary image database.

7. The image acquisition and allocation method based on a monitoring array according to claim 6, characterized in that, In S4-5, S4-5-1, the node processing module retrieves the location information of neighboring sampling nodes and the preceding command identification code from the node's local information database and stores them in the node's memory. The node processing module also retrieves the overlap determination algorithm information from the image task's first-level processing module and stores it in the node's memory. S4-5-2, the image task first-level processing module selects a received compressed image, proportionally changes the size of the local image based on the horizontal height difference between the local machine and the corresponding sampling node of the compressed image, obtains the position of the sampling node corresponding to the image based on the pre-command identification code of the compressed image, and translates the image acquired by the local machine towards the direction of the corresponding sampling node of the comparison image according to the image scale. When the reference beacon coincides, the proportion of the area of ​​the overlapping region to the image acquired by the local machine is calculated, and it is determined whether the overlapping region is within the overlap reference threshold. S4-5-3, If yes, record the wide-angle information in that direction as normal data; otherwise, if the overlap is greater than the overlap reference threshold, record that the wide-angle in that direction is too large; if the overlap is less than the overlap reference threshold, record that the wide-angle in that direction is too small. Then, execute the processing of sending compressed images to the next sampling node and repeat step S4-5-2. If everything is normal, skip step S4-5-4; S4-5-4 The dynamic task allocation module weights the overlap of each sampling node based on the recorded results and overlap ratio in S4-5-3, obtains the adjustment degree of the wide-angle of the camera module based on the weighting result, and commands the camera module to make the adjustment.

8. The image acquisition and allocation method based on a monitoring array according to claim 6, characterized in that, Between the array setup and monitoring execution steps, a beacon is established to ensure appropriate overlap and delay in sampling from each sampling node in the monitoring array. Based on the macroscopic matrix distribution of the monitoring array, image information that can be released at a fixed frequency periodically is placed at the junction of every two sampling nodes.

9. The image acquisition and allocation method based on a monitoring array according to claim 1, characterized in that, It also includes result output, identification and analysis of the integrated image data, output of the analysis results through the human-computer interaction unit, and calling the original image database and the secondary image database. The original image data directly sent by the monitoring array and the secondary image data obtained by stitching are transmitted to the back-end computing platform through the site signal transceiver module for storage and further image analysis.

Images