Agricultural remote sensing image correction system based on data analysis

Abstract

The invention discloses an agricultural remote sensing image correction system based on data analysis, and relates to the technical field of remote sensing image processing. The operation state of the correction node can be accurately evaluated by monitoring the operation parameters of the correction node in real time, such as the processor load, the memory utilization rate, the network broadband and the queue length, and whether a correction adjustment instruction is generated or not is judged according to the operation state; the method comprises the following steps: analyzing spectral features of blocks of a remote sensing image, and classifying the blocks into label categories of high-density vegetation, low-density vegetation, water and non-vegetation; according to the method, processing values of all the blocks are obtained by analyzing the complexity and needed computing resources of all the blocks of the label categories, all the blocks are distributed to all the correction nodes according to operation state values of all the correction nodes to which the label categories belong, the blocks of different label categories are processed in parallel, and therefore the purpose that the blocks are distributed according to needs is achieved. According to the method, computing resources can be utilized to the maximum extent, resource waste is avoided, and meanwhile it is ensured that all the blocks can be properly processed.

Description

Technical Field

[0001] The invention relates to the technical field of remote sensing image processing, and in particular to an agricultural remote sensing image correction system based on data analysis. Background Art

[0002] Agricultural remote sensing images are ground images of agricultural-related areas acquired through remote sensing technology. Based on the different spectral characteristics in remote sensing images, different crop types can be distinguished, crop growth conditions can be analyzed, or diseases or pests can be monitored. In the process of acquiring remote sensing images, they are usually affected by many factors, such as the atmosphere, sensor performance, and ground object changes. These influences will cause deviations in the spectral and spatial information in the image, which in turn affects applications such as agricultural monitoring and crop growth assessment. Therefore, correction of remote sensing images is very necessary.

[0003] With the development of remote sensing technology, more and more high-resolution remote sensing images are used to monitor agriculture. However, high-resolution data are often accompanied by a larger amount of data. Usually, a large number of high-resolution images are processed and adjusted uniformly. This processing method has great limitations and cannot perform targeted processing on the resolution and features of the image, resulting in reduced image correction accuracy and efficiency. At the same time, in large-scale agricultural monitoring tasks, a single unified adjustment method will lead to a waste of computing resources and fail to efficiently utilize the advantages of hardware and cloud computing platforms, thus affecting the efficiency of the entire processing process. Summary of the invention

[0004] Based on this, it is necessary to provide an agricultural remote sensing image correction system based on data analysis to address the problems mentioned in the above background technology.

[0005] The object of the present invention can be achieved by the following technical solutions: an agricultural remote sensing image correction system based on data analysis, comprising a server, a data analysis module, a block classification module and a correction parallel module;

[0006] The server communicates with the remote sensing mapping equipment to obtain the remote sensing image, and cuts it into several blocks, and generates corresponding position coordinates for the position of the block in the remote sensing image, so that each block corresponds to a unique position coordinate;

[0007] The data analysis module determines the operating status of each correction node by performing real-time analysis on each correction node, and generates correction adjustment instructions accordingly; the correction adjustment instructions are sent to the block classification module and the correction parallel module;

[0008] The block classification module performs spectral feature analysis on the blocks of the remote sensing image based on the received correction adjustment instructions to classify them, specifically classifying each block into high-density vegetation labels, low-density vegetation labels, non-vegetation labels and water labels, and sends them to the correction parallel module;

[0009] The correction parallel module performs parallel correction based on the received correction adjustment instructions and the blocks of each label, and integrates the corrected blocks according to their corresponding position coordinates to form a complete remote sensing image.

[0010] In some embodiments, each correction node is analyzed in real time to determine the operation status of each correction node. The specific analysis and determination process is as follows:

[0011] 2-1: Assume that there are several correction nodes, obtain the operating parameters of the correction nodes at each collection time, the specific operating parameters include processor load, memory usage, network bandwidth and queue length, and record them as Fj, Cj, Kj and Mj respectively, where j = 1, 2, 3...J, J is a positive integer, J represents the total number of monitoring moments, and j represents the number of any monitoring moment; set each correction node to correspond to a processor standard parameter, the specific processor standard parameters include the maximum memory usage and the maximum network bandwidth, and record them as Cmax and Kmax respectively;

[0012] The processor load Fj, memory usage Cj, network bandwidth Kj, queue length Mj, maximum memory usage Cmax and maximum network bandwidth Kmax are normalized and their values ​​are taken. The values ​​are analyzed and calculated by formula to obtain the bottleneck value CFj at each acquisition time. The specific calculation formula is:

[0013]

[0014] Wherein λ1, λ2, λ3, and λ4 are respectively set proportional constants, and their values ​​are set by those skilled in the art;

[0015] 2-2: A two-dimensional rectangular coordinate system is constructed with time as the horizontal coordinate and the bottleneck value as the vertical coordinate. The operation value is input into the coordinate system according to its corresponding monitoring time, and the position of the bottleneck value in the coordinate system is recorded as the bottleneck point. The bottleneck points are connected in sequence with line segments to obtain a bottleneck value line graph; the bottleneck line graph is graphically analyzed to measure the operation status of the correction node and obtain the operation status value of the correction node;

[0016] 2-3: Thus, the operating status value of each correction node can be obtained, and it is compared and analyzed with the set state interval. When the operating status value is greater than the maximum value in the set state interval, the correction node is recorded as a high bottleneck node; when the operating status value is in the set state interval, the correction node is recorded as a moderate bottleneck node; when the operating status value is less than the minimum value in the set state interval, the correction node is recorded as a light bottleneck node; the number of high bottleneck nodes, moderate bottleneck nodes and light bottleneck nodes is counted respectively, and they are recorded as D1, D2 and D3 respectively; if D3≥D1+D2, no adjustment is required; otherwise, a correction adjustment instruction is generated.

[0017] In some embodiments, the bottleneck line graph is graphically parsed to measure the operating status of the correction node, specifically:

[0018] Assume that there is a bottleneck interval denoted as [P1, P2], and draw two straight lines parallel to the horizontal axis in the bottleneck value line graph, namely, straight line y=P1 and straight line y=P2; thus, the bottleneck value line graph can be divided into three parts by the two straight lines, one part is below the straight line y=P1, one part is between the straight line y=P1 and the straight line y=P2, and one part is above the straight line y=P2; calculate the shaded areas of the three parts respectively, and denoted them as SP1, SP1, P2 and SP2 respectively;

[0019] The shaded areas of the three parts SP1, SP1, P2 and SP2 are calculated and analyzed by formula to obtain the operating status value SP. The specific calculation formula is:

[0020]

[0021] Wherein μ1 and μ2 are respectively set proportional constants, and the specific values ​​are set by those skilled in the art.

[0022] In some embodiments, spectral feature analysis is performed on the image blocks to classify them. The specific analysis process is as follows:

[0023] The reflectance of the infrared band in the extracted image block is recorded as LNIR, the reflectance of the red band is recorded as LRED, the reflectance of the blue band is recorded as LBLUE, the short-wave infrared band is recorded as PSWIR, and the green band is recorded as PGREEN.

[0024] The vegetation index is obtained by formulating and analyzing the reflectance of the infrared band LNIR and the reflectance of the red light band LRED The specific calculation formula is:

[0025]

[0026] The reflectance of the infrared band LNIR, the reflectance of the red band LRED and the reflectance of the blue band are recorded as LBLUE and the enhanced vegetation index is obtained by formulating and analyzing. The specific calculation formula is:

[0027]

[0028] Where G is the set gain factor, c1 and c2 are the set coefficients, and D is the ground brightness;

[0029] The water index ZP is obtained by formulating and analyzing the short-wave infrared band PSWIR and the green band PGREEN. The specific calculation formula is:

[0030]

[0031] The vegetation index of each block can be obtained from this Enhanced Vegetation Index and water body index ZP, and classify the blocks accordingly.

[0032] In some embodiments, the specific classification process is:

[0033] Step 1: Extract vegetation index of the tile If the vegetation index Then execute step 2; otherwise, execute step 3;

[0034] Step 2: Extract the enhanced vegetation index of the tile If the enhanced vegetation index If the value is high, the image block is classified as high-density vegetation; otherwise, the image block is classified as low-density vegetation.

[0035] Step 3: Extract the water index of the image block. If the water index ZP>0, the image block is classified as a water body label; otherwise, the image block is classified as a non-vegetation label.

[0036] Step 4: Repeat steps 1 to 4 above to classify all tiles into corresponding labels, and send the classified tiles to the correction parallel module.

[0037] In some embodiments, the specific process of performing parallel correction on the image blocks under each label category is as follows:

[0038] 6-1: Assign each correction node to each label, thereby obtaining a number of correction nodes corresponding to each label category;

[0039] 6-2: Randomly select one of the target signatures, extract the corresponding correction node number under the target signature, integrate it into a re-monitoring instruction and send it to the data analysis module to obtain the latest operating status value of the correction node corresponding to the target signature, sort the correction nodes in descending order according to the operating status value, and select the one with the smallest operating status value as the target node;

[0040] 6-3: Quantitatively analyze the complexity of the correction processing of each block to obtain the processing value of each block;

[0041] 6-4: Sort the tiles under the label in descending order according to their corresponding processing values, select the tile with the largest processing value as the target tile, and assign the correction task of the target tile to the target node, thereby increasing the queue length of the target node's task list by one; whenever the correction node completes the correction of a tile, the queue length of the corresponding task list decreases by one;

[0042] 6-5: Repeat the above steps 6-1 to 6-4 until all tiles in the label are allocated;

[0043] 6-6: Repeat the above steps 6-1 to 6-5 until all blocks signed by various targets are calibrated, and merge the calibrated blocks according to their corresponding position coordinates to restore the complete remote sensing image.

[0044] In some embodiments, the specific process of quantitatively analyzing the complexity of the correction processing of the image block is as follows:

[0045] 7-1: Identify the pixels in the image block, extract the spatial resolution and spectral resolution of each pixel, and record them as Qn and Rn respectively, n = 1, 2, 3...N, N is a positive integer, N refers to the number of pixels in the image block, and n represents the number of any pixel; normalize the spatial resolution Qn and spectral resolution Rn of the pixel and take their values, and perform formula calculation and analysis on the values ​​to obtain the resolution value QRn of the pixel. The specific calculation formula is:

[0046] QRn=b1×Qn+b2×Rn

[0047] Wherein b1 and b2 are respectively set proportional coefficients, and their values ​​are set by personnel in this field according to actual needs;

[0048] 7-2: Compare and analyze the resolution value of each pixel with the set resolution interval to obtain the detail index of the block;

[0049] 7-3: Extract the area and pixel density of the block and record them as E and ρ respectively; normalize the area E, pixel density ρ and detail index Ad of the block and take their values, and perform formula calculation and analysis on the values ​​to obtain the processed value AE of the block. The specific calculation formula is:

[0050]

[0051] Among them, f1, f2, f3 are respectively set proportional constants, h1 and h2 are set coefficients; the specific values ​​are set by those skilled in the art.

[0052] In some embodiments, the calculation process of the detail index of a tile is:

[0053] If the resolution value of a pixel point is greater than the upper limit of the set resolution interval, the pixel point is recorded as a high resolution point; if the resolution value of a pixel point is within the set resolution interval, the pixel point is recorded as a medium resolution point; if the resolution value of a pixel point is less than the lower limit of the set resolution interval, the pixel point is recorded as a low resolution point; the number of high resolution points, medium resolution points and low resolution points in the block is counted respectively, and they are recorded as A1, A2, and A3 respectively;

[0054] The number of high-resolution points A1, the number of medium-resolution points A2, the number of low-resolution points A3 and the resolution value QRn of each pixel are normalized and their values ​​are taken. The values ​​are calculated and analyzed by formula to obtain the detail index Ad of the block. The specific calculation formula is:

[0055]

[0056] Wherein d1, d2, and d3 are respectively set proportional constants, and d1>d2>d3>1.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] 1. By real-time monitoring of the operating parameters of the correction node, such as processor load, memory usage, network bandwidth, and queue length, the operating status of the correction node can be accurately evaluated, and based on this, it can be determined whether to generate correction adjustment instructions;

[0059] 2. By analyzing the spectral characteristics of the blocks of remote sensing images, the blocks are classified into label categories of high-density vegetation, low-density vegetation, water bodies, and non-vegetation, so that the system can carry out targeted processing on different types of blocks, thereby improving the correction accuracy; providing support for more reasonable allocation of computing resources and improving processing efficiency;

[0060] 3. The processing value of each tile is obtained by analyzing the complexity of each tile of the label category and the required computing resources. Then, according to the running status value of each correction node to which the label category belongs, each tile is allocated to each correction node, and tiles of different label categories are processed in parallel, so as to further improve the correction efficiency and accuracy. The on-demand allocation method can maximize the use of computing resources, avoid resource waste, and ensure that all tiles can be properly processed.

[0061] In summary, the present invention monitors and optimizes correction nodes in real time, classifies blocks to achieve targeted processing, and uses parallel processing to improve efficiency. The system can more effectively process large-scale remote sensing image data, improve correction accuracy and processing speed, and thus better support agricultural monitoring and other related applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the conventional technology, the drawings required for use in the embodiments or the conventional technology descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0063] Figure 1 It is a principle block diagram of the present invention;

[0064] Figure 2 It is a graphical analysis schematic diagram of the bottleneck value line graph of the present invention. DETAILED DESCRIPTION

[0065] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention, so the present invention is not limited by the specific embodiments disclosed below.

[0066] like Figure 1 As shown, an agricultural remote sensing image correction system based on data analysis includes: a server, a data analysis module, a block classification module and a correction parallel module;

[0067] The server communicates with remote sensing mapping equipment (such as satellites, drones, and sensors on aviation platforms) to obtain remote sensing images, cuts large-scale remote sensing images into several blocks, and generates corresponding position coordinates for the positions of the blocks in the remote sensing images, so that each block corresponds to a unique position coordinate, which is convenient for data integration after correction; it should be noted that in order to avoid errors in subsequent splicing, a fixed overlapping area is set (usually those skilled in the art will set the overlapping area to 20%-30%) to facilitate the subsequent splicing and integration of blocks;

[0068] The data analysis module determines the operating status of each correction node by performing real-time analysis on each correction node, and generates correction adjustment instructions accordingly; specifically:

[0069] It is assumed that there are several correction nodes, which are used to perform image correction on each block; the operating parameters of the correction nodes at each acquisition time are obtained, and the specific operating parameters include processor load, memory usage, network bandwidth and queue length (queue length specifically refers to the number of tasks waiting to be processed), and they are recorded as Fj, Cj, Kj and Mj respectively, where j = 1, 2, 3...J, J is a positive integer, J represents the total number of monitoring moments, and j represents the number of any monitoring moment;

[0070] Each correction node is set to correspond to a processor standard parameter. The specific processor standard parameters include the maximum memory usage rate (the maximum memory usage rate refers to the predetermined memory upper limit of the correction node, indicating the maximum memory that the correction node can use during image processing) and the maximum network bandwidth (the maximum network bandwidth refers to the maximum bandwidth of the correction node's network, indicating the maximum amount of data that the node can transmit through the network. The transmission, processing and result return of remote sensing images all depend on the network bandwidth), and they are recorded as Cmax and Kmax respectively; it should be noted that when the processor of the correction node is running, when its memory usage rate is close to the maximum memory usage rate, it means that the node cannot carry more image processing tasks, indicating that the computing resources of the correction node are close to saturation and are in a computing resource bottleneck state; at the same time, when the network bandwidth of the processor is close to the maximum network bandwidth, the response speed of the correction node will decrease, resulting in increased latency and blocked tile correction processing;

[0071] The processor load Fj, memory usage Cj, network bandwidth Kj, queue length Mj, maximum memory usage Cmax and maximum network bandwidth Kmax are normalized and their values ​​are taken. The values ​​are analyzed and calculated by formula to obtain the bottleneck value CFj at each acquisition time. The specific calculation formula is:

[0072]

[0073] Among them, λ1, λ2, λ3, and λ4 are respectively set proportional constants, and their values ​​are set by personnel in this field. It can be seen from the formula that when the processor load is larger and the queue length is larger, it means that the load of the correction node is heavier, and the bottleneck value is larger; when the memory usage rate is closer to the maximum memory usage rate, it means that the computing resources of the correction node are close to saturation, and the bottleneck value is larger; when the network bandwidth is close to the maximum network bandwidth, the response speed of the correction node will decrease, and the bottleneck value is larger;

[0074] like Figure 2 As shown, a two-dimensional rectangular coordinate system is constructed with time as the horizontal coordinate and bottleneck value as the vertical coordinate, the operating value is input into the coordinate system according to its corresponding monitoring time, and the position of the bottleneck value in the coordinate system is recorded as the bottleneck point, and the bottleneck points are connected in sequence by line segments to obtain a bottleneck value line graph; it is assumed that there is a bottleneck interval recorded as [P1, P2], and two straight lines parallel to the horizontal axis are drawn in the bottleneck value line graph, which are straight line y=P1 and straight line y=P2; thus, the bottleneck value line graph can be divided into three parts by two straight lines, one part is below the straight line y=P1, one part is between the straight line y=P1 and the straight line y=P2, and one part is above the straight line y=P2; the shaded areas of the three parts are calculated respectively, and they are recorded as SP1, SP1,P2 and SP2 respectively; it should be noted that when SP1 is larger, it means that the bottleneck of the correction node is more serious; when SP1,P2 is larger, it means that the bottleneck of the correction node is at a normal level; when SP2 is larger, it means that the running state of the correction node is better and the bottleneck degree is lower;

[0075] The shaded areas of the three parts SP1, SP1, P2 and SP2 are calculated and analyzed by formula to obtain the operating status value SP. The specific calculation formula is:

[0076]

[0077] Wherein μ1 and μ2 are respectively set proportional constants, and the specific values ​​are set by the personnel in this field. It can be seen from the formula that when SP1 is larger, it means that the bottleneck of the correction node is more serious, and the operating state value is larger; when SP2 is larger, it means that the operating state of the correction node is better, and the lower the bottleneck degree, the smaller the operating state value; thus, the operating state value of each correction node can be obtained, and compared with the set state interval for analysis. When the operating state value is greater than the maximum value in the set state interval, it means that the bottleneck of the correction node is more serious, and it needs to be corrected and adjusted to improve its correction efficiency. The correction node is recorded as a high bottleneck node; when the operating state value is in the set state interval, the correction node is recorded as a moderate bottleneck node; when the operating state value is less than the minimum value in the set state interval, the correction node is recorded as a mild bottleneck node; the number of high bottleneck nodes, moderate bottleneck nodes and mild bottleneck nodes are counted respectively, and they are recorded as D1, D2 and D3 respectively; if D3≥D1+D2, it means that each correction node can accommodate the correction of the remote sensing image and no adjustment is required; otherwise, a correction adjustment instruction is generated and sent to the block classification module and the correction parallel module;

[0078] By real-time monitoring of the operating parameters of the correction nodes, such as processor load, memory usage, network bandwidth, and queue length, the operating status of the correction nodes can be accurately assessed, and based on this, it can be determined whether to generate correction adjustment instructions.

[0079] The tile classification module classifies remote sensing images by performing feature analysis, specifically:

[0080] The reflectance of different bands (such as red light, near infrared light, green light, etc.) of the image block is extracted using spectral features. Specifically, the reflectance of the infrared band is recorded as LNIR, the reflectance of the red light band is recorded as LRED, the reflectance of the blue light band is recorded as LBLUE, the short-wave infrared band is recorded as PSWIR, and the green band is recorded as PGREEN.

[0081] The vegetation index is obtained by formulating and analyzing the reflectance of the infrared band LNIR and the reflectance of the red light band LRED The specific calculation formula is:

[0082]

[0083] It should be noted that vegetation index is usually When it is greater than 0.3, it means that there is a vegetation area (such as farmland, grassland, etc.) in the block, otherwise it is a non-vegetation area (such as bare soil, etc.);

[0084] The reflectance of the infrared band LNIR, the reflectance of the red band LRED and the reflectance of the blue band are recorded as LBLUE and the enhanced vegetation index is obtained by formulating and analyzing. The specific calculation formula is:

[0085]

[0086] Where G is the set gain factor, c1 and c2 are the set coefficients, and D is the ground brightness. It should be noted that the enhanced vegetation index When it is greater than 0.2, it means that there is a high-density vegetation area in the block (such as farmland with high-density crops or dense forests. Since the specific application scenario of the present invention is plain farmland, the enhanced vegetation index is used in the judgment of the present invention. When it is greater than 0.2, the system defaults to farmland with high-density crops);

[0087] The water index ZP is obtained by formulating and analyzing the short-wave infrared band PSWIR and the green band PGREEN. The specific calculation formula is:

[0088]

[0089] It should be noted that when the water index ZP is greater than zero, it means that there is a water area in the block, otherwise, it means that there is no water area in the block;

[0090] The vegetation index of each block can be obtained from this Enhanced Vegetation Index and water body index ZP, and classify the blocks accordingly. The specific classification process is as follows:

[0091] Step 1: Extract vegetation index of the tile If the vegetation index Then execute step 2; otherwise, execute step 3;

[0092] Step 2: Extract the enhanced vegetation index of the tile If the enhanced vegetation index If the value is high, the image block is classified as high-density vegetation; otherwise, the image block is classified as low-density vegetation.

[0093] Step 3: Extract the water index of the image block. If the water index ZP>0, the image block is classified as a water body label; otherwise, the image block is classified as a non-vegetation label.

[0094] Step 4: Repeat steps 1 to 4 above to classify all tiles into corresponding labels, and send the classified tiles to the correction parallel module;

[0095] By analyzing the spectral characteristics of the blocks of remote sensing images, the blocks are classified into label categories of high-density vegetation, low-density vegetation, water bodies, and non-vegetation. This enables the system to perform targeted processing on different types of blocks, thereby improving correction accuracy and providing support for more reasonable allocation of computing resources and improving processing efficiency.

[0096] The correction parallel module performs parallel correction based on the received correction adjustment instructions and the blocks of each label to improve the correction accuracy and correction efficiency; specifically:

[0097] Step 1: Assign each correction node to each label, so that several correction nodes corresponding to each label category can be obtained; it should be noted that the specific correction node allocation is arranged by the technicians in this field according to actual needs. Usually, when arranging, the technicians will be responsible for more correction nodes for high-density vegetation labels than for other label categories. For example, if there are 30 correction nodes, the technicians can allocate them as follows: 12 correction nodes for high-density vegetation labels, 8 correction nodes for low-density vegetation labels, 5 correction nodes for non-vegetation labels, and 5 correction nodes for water body labels; this is because high-density farmland usually covers a larger area, and in remote sensing images, farmland areas usually have a higher detail density, many types of crops, complex distribution, and many monitoring indicators, so more computing resources are needed to process these blocks to ensure higher correction accuracy;

[0098] Step 2: Take any of the target tags and extract the corresponding correction node number under the target tag, integrate it into a re-monitoring instruction and send it to the data analysis module to obtain the latest operation status value SP of the correction node corresponding to the target tag, sort the correction nodes in descending order according to the operation status value, and select the one with the smallest operation status value as the target node;

[0099] Step 3: Identify the pixels in the image block, extract the spatial resolution and spectral resolution of each pixel, and record them as Qn and Rn respectively, n = 1, 2, 3...N, N is a positive integer, N refers to the number of pixels in the image block, and n represents the number of any pixel in it; the spatial resolution refers to the actual size of the ground represented by each pixel in the image. The higher the spatial resolution, the smaller the ground area that each pixel can describe; the spectral resolution refers to the performance of the reflectance value of each pixel in the remote sensing image in multiple bands. The higher the spectral resolution, the more spectral information the image can record; when the spatial resolution and spectral resolution of the image block are larger, it usually means that the overall resolution of the image block is higher, and the processing requirements and computational complexity of the image block increase;

[0100] The spatial resolution Qn and spectral resolution Rn of the pixel are normalized and their values ​​are taken. The values ​​are calculated and analyzed by formula to obtain the resolution value QRn of the pixel. The specific calculation formula is:

[0101] QRn=b1×Qn+b2×Rn

[0102] Wherein b1 and b2 are respectively set proportional coefficients, and their values ​​are set by the personnel in this field according to actual needs; the resolution value of each pixel point is compared and analyzed with the set resolution interval. If the resolution value of the pixel point is greater than the upper limit of the set resolution interval, the pixel point is recorded as a high resolution point; if the resolution value of the pixel point is in the set resolution interval, the pixel point is recorded as a medium resolution point; if the resolution value of the pixel point is less than the lower limit of the set resolution interval, the pixel point is recorded as a low resolution point; the number of high resolution points, medium resolution points and low resolution points in the block is counted respectively, and they are recorded as A1, A2, A3 respectively; the number of high resolution points A1, the number of medium resolution points A2, the number of low resolution points A3 and the resolution value QRn of each pixel point are normalized and their values ​​are taken, and the values ​​are calculated and analyzed by formula to obtain the detail index Ad of the block. The specific calculation formula is:

[0103]

[0104] Where d1, d2, and d3 are respectively set proportional constants, and d1>d2>d3>1;

[0105] The area and pixel density of the block are extracted and recorded as E and ρ respectively; the area E, pixel density ρ and detail index Ad of the block are normalized and their values ​​are taken, and the values ​​are calculated and analyzed by formula to obtain the processed value AE of the block. The specific calculation formula is:

[0106]

[0107] Among them, f1, f2, and f3 are respectively set proportional constants, and h1 and h2 are set coefficients; their specific values ​​are set by personnel in this field; from the formula, it can be seen that when the area of ​​the block is larger, the image processing requirement is higher, and the processing value is larger; when the pixel density is larger, the computational complexity during processing is also greater, and the processing value is larger; when the detail index is larger, the processing value is larger; among them, for blocks with high-density vegetation labels and low-density vegetation labels, their vegetation index Enhanced Vegetation Index The larger the value, the larger the corresponding processing value; therefore, the processing value of each block under this label;

[0108] Step 4: Sort the tiles under the label in descending order according to their corresponding processing values, select the tile with the largest processing value as the target tile, and assign the correction task of the target tile to the target node, thereby increasing the queue length of the target node's task list by one; whenever the correction node completes the correction of a tile, the queue length of the corresponding task list decreases by one;

[0109] Step 5: Repeat steps 1 to 4 above until all tiles in the label are assigned;

[0110] Step 6: Repeat steps 1 to 5 above until all blocks under various targets are calibrated, and merge the calibrated blocks according to their corresponding position coordinates to restore the complete remote sensing image; in the merging process, it is necessary to deal with edge effects and overlapping areas to ensure the smooth transition and accuracy of the final image; the final calibrated image can be output to a standard remote sensing image format (such as GeoTIFF) for subsequent analysis, visualization and decision support;

[0111] The processing value of each tile is obtained by analyzing the complexity of each tile in the label category and the required computing resources. Then, each tile is allocated to each correction node according to the running status value of each correction node to which the label category belongs. Tiles of different label categories are processed in parallel to further improve the correction efficiency and accuracy. The on-demand allocation method can maximize the use of computing resources and avoid resource waste, while ensuring that all tiles can be properly processed.

[0112] The above formulas are all obtained by collecting a large amount of data and performing software simulation, and a formula close to the actual value is selected. The coefficients in the formula are set by technicians in this field according to actual conditions.

[0113] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0114] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. An agricultural remote sensing image correction system based on data analysis, characterized in that: It includes a server, a data analysis module, a tile classification module and a correction parallel module; The server communicates with the remote sensing mapping equipment to obtain the remote sensing image, and cuts it into several blocks, generates corresponding position coordinates for the position of the block in the remote sensing image, and obtains a unique position coordinate corresponding to each block; The data analysis module determines the operating status of each correction node by performing real-time analysis on each correction node, and generates correction adjustment instructions accordingly; the correction adjustment instructions are sent to the block classification module and the correction parallel module; The block classification module performs spectral feature analysis on the blocks of the remote sensing image to classify them based on the received correction adjustment instructions, classifies each block into high-density vegetation labels, low-density vegetation labels, non-vegetation labels and water labels, and sends them to the correction parallel module; The correction parallel module performs parallel correction based on the received correction adjustment instructions and the blocks of each label, and integrates the corrected blocks according to their corresponding position coordinates to form a complete remote sensing image.

2. The agricultural remote sensing image correction system based on data analysis according to claim 1 is characterized in that: The operation status of each correction node is determined by real-time analysis of each correction node, including: 2-1: Assume that there are several correction nodes, obtain the operating parameters of the correction nodes at each collection time, the operating parameters include processor load, memory usage, network bandwidth and queue length; determine that each correction node corresponds to a processor standard parameter, the processor standard parameter includes maximum memory usage and maximum network bandwidth; normalize the processor load, memory usage, network bandwidth, queue length, maximum memory usage and maximum network bandwidth and take their values, perform formula calculation and analysis on the values ​​to obtain the bottleneck value at each collection time; 2-2: A two-dimensional rectangular coordinate system is constructed with time as the horizontal coordinate and the bottleneck value as the vertical coordinate. The operation value is input into the coordinate system according to its corresponding monitoring time, and the position of the bottleneck value in the coordinate system is recorded as the bottleneck point. The bottleneck points are connected in sequence with line segments to obtain a bottleneck value line graph; the bottleneck line graph is graphically analyzed to measure the operation status of the correction node and obtain the operation status value of the correction node; 2-3: Get the running status value of each correction node, and compare and analyze it with the set status interval. When the running status value is greater than the maximum value in the set status interval, the correction node is recorded as a high bottleneck node; when the running status value is in the set status interval, the correction node is recorded as a moderate bottleneck node; when the running status value is less than the minimum value in the set status interval, the correction node is recorded as a light bottleneck node; count the number of high bottleneck nodes, moderate bottleneck nodes and light bottleneck nodes respectively, and record them as D1, D2 and D3 respectively; if D3≥D1+D2, no adjustment is required; otherwise, generate a correction adjustment instruction.

3. The agricultural remote sensing image correction system based on data analysis according to claim 2 is characterized in that: Graphical analysis of the bottleneck line graph to measure the operating status of the correction node includes: Determine that there is a bottleneck interval denoted as [P1, P2], draw two straight lines parallel to the horizontal axis in the bottleneck value line graph, namely straight line y=P1 and straight line y=P2; the bottleneck value line graph is divided into three parts by the two straight lines, one part is below the straight line y=P1, one part is between the straight line y=P1 and the straight line y=P2, and one part is above the straight line y=P2; calculate the shaded areas of the three parts respectively, and denoted them as SP1, SP1, P2 and SP2 respectively; The shaded areas of the three parts SP1, SP1, P2 and SP2 are calculated and analyzed by formula to obtain the running status value SP, and the calculation formula is: Wherein μ1 and μ2 are respectively set proportional constants, and the specific values ​​are set by those skilled in the art.

4. The agricultural remote sensing image correction system based on data analysis according to claim 1, characterized in that: Analyzing the spectral characteristics of the tiles to classify them includes: The reflectance of the infrared band in the extracted image block is recorded as LNIR, the reflectance of the red band is recorded as LRED, the reflectance of the blue band is recorded as LBLUE, the short-wave infrared band is recorded as PSWIR, and the green band is recorded as PGREEN. The vegetation index is obtained by formulating and analyzing the reflectance of the infrared band LNIR and the reflectance of the red light band LRED The specific calculation formula is: The reflectance of the infrared band LNIR, the reflectance of the red band LRED and the reflectance of the blue band are recorded as LBLUE and the enhanced vegetation index is obtained by formulating and analyzing. The specific calculation formula is: Where G is the set gain factor, c1 and c2 are the set coefficients, and D is the ground brightness; The water index ZP is obtained by formulating and analyzing the short-wave infrared band PSWIR and the green band PGREEN. The specific calculation formula is: The vegetation index of each block can be obtained from this Enhanced Vegetation Index and water body index ZP, and classify the blocks accordingly.

5. The agricultural remote sensing image correction system based on data analysis according to claim 4 is characterized in that: The classification process is: Step 1: Extract vegetation index of the tile If the vegetation index Then execute step 2; Otherwise, go to step 3; Step 2: Extract the enhanced vegetation index of the tile If the enhanced vegetation index If the value is high, the image block is classified as high-density vegetation; otherwise, the image block is classified as low-density vegetation. Step 3: Extract the water index of the image block. If the water index ZP>0, the image block is classified as a water body label; otherwise, the image block is classified as a non-vegetation label. Step 4: Repeat steps 1 to 4 above to classify all tiles into corresponding labels, and send the classified tiles to the correction parallel module.

6. The agricultural remote sensing image correction system based on data analysis according to claim 1, characterized in that: Parallel correction of tiles under each label category includes: 6-1: Assign each correction node to each label to obtain a number of correction nodes corresponding to each label category; 6-2: Randomly select one of the target signatures, extract the corresponding correction node number under the target signature, integrate it into a re-monitoring instruction and send it to the data analysis module to obtain the latest operating status value of the correction node corresponding to the target signature, sort the correction nodes in descending order according to the operating status value, and select the one with the smallest operating status value as the target node; 6-3: Quantitatively analyze the complexity of the correction processing of each block to obtain the processing value of each block; 6-4: Sort the tiles under the label in descending order according to their corresponding processing values, select the tile with the largest processing value as the target tile, and assign the correction task of the target tile to the target node, thereby increasing the queue length of the target node's task list by one; whenever the correction node completes the correction of a tile, the queue length of the corresponding task list decreases by one; 6-5: Repeat the above steps 6-1 to 6-4 until all tiles in the label are allocated; 6-6: Repeat the above steps 6-1 to 6-5 until all blocks signed by various targets are calibrated, and merge the calibrated blocks according to their corresponding position coordinates to restore the complete remote sensing image.

7. The agricultural remote sensing image correction system based on data analysis according to claim 6, characterized in that: The quantitative analysis of the complexity of the correction process based on the image block includes: 7-1: Identify the pixels in the image block, extract the spatial resolution and spectral resolution of each pixel, normalize them and take their values, and perform formula calculation and analysis on the values ​​to obtain the resolution value of the pixel; 7-2: Compare and analyze the resolution value of each pixel with the set resolution interval to obtain the detail index of the block; 7-3: Extract the area and pixel density of the block, normalize them with the detail index and take its value, perform formulaic calculation and analysis on the value to obtain the processing value of the block; thus, the processing value of each block under this label is obtained.

8. The agricultural remote sensing image correction system based on data analysis according to claim 7, characterized in that: The calculation process of the detail index of a tile is: If the resolution value of a pixel point is greater than the upper limit of the set resolution interval, the pixel point is recorded as a high resolution point; if the resolution value of a pixel point is within the set resolution interval, the pixel point is recorded as a medium resolution point; if the resolution value of a pixel point is less than the lower limit of the set resolution interval, the pixel point is recorded as a low resolution point; the number of high resolution points, medium resolution points and low resolution points in the block is counted respectively; The number of high-resolution points, the number of medium-resolution points, the number of low-resolution points and the resolution value of each pixel are normalized and their values ​​are taken, and the values ​​are calculated and analyzed in a formula to obtain the detail index of the block.

Images