A locust disaster prediction method and system based on collected data

Abstract

The application provides a locust disaster prediction method and system based on collected data, and relates to the technical field of disaster prediction. The method obtains a gray-scale map of a target map based on a locust-related data set, uses a superpixel algorithm to perform image segmentation on the gray-scale map of the target map to obtain a first sub-image list, further obtains a second sub-image group, obtains a key aggregation probability list according to the locust aggregation probability corresponding to the second sub-image group, and determines whether the first sub-image needs to be merged and updated or directly obtains a target sub-image according to the key aggregation probability and the gray-scale value variance. The aggregation behavior of locusts is considered. When the target sub-image is obtained, the target sub-image is marked using the color corresponding to the target event average occurrence probability corresponding to the target sub-image. The data amount is small, the target sub-image can be quickly determined and color marked, and locust disaster prediction can be comprehensively and accurately realized. The target event is a locust disaster event occurring at a target time point.

Description

Technical Field

[0001] This invention relates to the field of disaster prediction technology, and in particular to a method and system for predicting locust disasters based on collected data. Background Technology

[0002] Locust plagues refer to the widespread damage caused to agriculture, ecosystems, and human society by swarms of locusts. These plagues not only affect crop yields but may also threaten food security and social stability. Therefore, locust plague forecasting is of paramount importance. Based on accurate locust plague forecasts, relevant departments can formulate and deploy effective prevention and control measures in advance, thereby significantly mitigating the negative impacts of locust plagues and ensuring food security and social stability.

[0003] In the prior art, patent (202111589159.8) provides a method for identifying potential high-risk areas for locust plagues. In this method, long-term multi-temporal remote sensing image data of the study area are used to extract and analyze the water bodies and suitable habitats for locusts in the study area to obtain the water body change index and suitable habitat change index of the study area. When the water body change index and suitable habitat change index meet the first preset condition, it is confirmed that there is a potential high-risk area for locust plagues in the study area. Then, the habitat suitability index calculated by multi-temporal remote sensing image data and the distance between each region and the water bodies in the study area are used to confirm the potential high-risk areas for locust plagues in the study area.

[0004] However, the above method also has the following technical problems: The above methods rely heavily on long-term multi-temporal remote sensing image data. If the quality of the remote sensing data is poor (such as cloud cover, insufficient resolution, etc.) or the data acquisition is discontinuous, it may affect the accuracy and reliability of the analysis results. In addition, the amount of long-term multi-temporal remote sensing image data is huge, and processing this data requires a lot of time, which makes it slow to identify potential high-risk areas for locust plagues. Furthermore, the process of identifying potential high-risk areas for locust plagues does not take into account the aggregation behavior of locusts, which may lead to the identification of potential high-risk areas for locust plagues being inaccurate or incomplete. Summary of the Invention

[0005] To address the aforementioned technical problems, the technical solution adopted by this invention is as follows: According to a first aspect of the present invention, a method for predicting locust plagues based on collected data is provided, the method comprising the following steps: S1. Obtain the grayscale image of the target map based on the locust-related dataset corresponding to each initial grid area.

[0006] S2. Use the superpixel algorithm to perform image segmentation on the grayscale image of the target map to obtain a list of first sub-images LB, which includes several first sub-images.

[0007] S3. Two adjacent first sub-images in the grayscale image of the target map are treated as two second sub-images in a second sub-image group to obtain the second sub-image group set B = {B1, B2, ..., B...}. j B n}, B j Let j be the j-th second sub-image group, where j ranges from 1 to n, and n is the number of second sub-image groups.

[0008] S4, Obtain B j The corresponding locust aggregation probability C j And sort C1, C2, ..., C in descending order. j C n Sort the data to obtain the key cluster probability list D.

[0009] S5. Let a = 1, where a is a preset count value.

[0010] S6, if D a ≥D 0 And E a ≤E 0 Then LB and D 1 a The two first sub-images corresponding to the two second sub-images in the image are merged into one first sub-image so that LB is updated and proceeding to step S3; otherwise, proceeding to step S7, D a Let D be the a-th key cluster probability. 0 To preset the aggregation probability, D 1 a D a The corresponding second sub-image group, E a D 1 a The corresponding variance of grayscale values, E 0 This is the preset grayscale value variance.

[0011] S7. If a≠n, let a=a+1 and proceed to step S6. If a=n, take the first sub-image in LB as the target sub-image and mark the target sub-image with the color corresponding to the average occurrence probability of the target event.

[0012] According to a second aspect of the present invention, a locust plague prediction system based on collected data is provided, the system comprising: The grayscale image acquisition module is used to acquire grayscale images of the target map based on the locust-related dataset corresponding to each initial grid area.

[0013] The first sub-image acquisition module is used to perform image segmentation on the grayscale image of the target map using a superpixel algorithm to obtain a first sub-image list LB that includes several first sub-images.

[0014] The second sub-image group acquisition module is used to obtain a second sub-image group set B={B1, B2, ..., B...} by taking two adjacent first sub-images in the grayscale image of the target map as two second sub-images in a second sub-image group. j B n}, B j Let j be the j-th second sub-image group, where j ranges from 1 to n, and n is the number of second sub-image groups.

[0015] The key clustering probability list acquisition module is used to obtain B. j The corresponding locust aggregation probability C j And sort C1, C2, ..., C in descending order. j C n Sort the data to obtain the key cluster probability list D.

[0016] The count value initialization module is used to set a=1, where a is a preset count value.

[0017] The first sub-image update module is used if D a ≥D 0 And E a ≤E 0 Then LB and D 1 a The two first sub-images corresponding to the two second sub-images in the algorithm are merged into one first sub-image so that the LB is updated and the algorithm enters the second sub-image group acquisition module; otherwise, the algorithm enters the target sub-image acquisition module. a Let D be the a-th key cluster probability. 0 To preset the aggregation probability, D 1 a D a The corresponding second sub-image group, E a D 1 a The corresponding variance of grayscale values, E 0 This is the preset grayscale value variance.

[0018] The target sub-image acquisition module is used to set a=a+1 and enter the first sub-image update module if a≠n, and if a=n, the first sub-image in LB is used as the target sub-image, and the target sub-image is marked with the color corresponding to the average occurrence probability of the target event corresponding to the target sub-image.

[0019] The present invention has at least the following beneficial effects: This invention provides a method and system for predicting locust plagues based on collected data. The method acquires a grayscale image of a target map from a locust-related dataset, uses a superpixel algorithm to segment the grayscale image to obtain a first sub-image list, further acquires a second sub-image set, obtains a key aggregation probability list based on the locust aggregation probabilities corresponding to the second sub-image sets, and determines whether to merge and update the first sub-images or directly use the first sub-images as target sub-images based on the variance of the grayscale values ​​corresponding to the key aggregation probabilities and the second sub-image sets corresponding to the key aggregation probabilities. When a target sub-image is acquired, it is marked with the color corresponding to the average occurrence probability of the target event corresponding to the target sub-image, where the target event is a locust plague event occurring at a target time point. This invention considers the aggregation behavior of locusts and mainly relies on a locust-related dataset to determine the target sub-images. It uses the color corresponding to the average occurrence probability of the target event corresponding to the target sub-images to mark them, thus achieving locust plague prediction. The data volume is relatively small, allowing for rapid determination and color marking of target sub-images, enabling comprehensive and accurate locust plague prediction. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A flowchart illustrating a locust plague prediction method based on collected data, provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a locust disaster prediction system based on collected data, provided in an embodiment of the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar tasks and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that comprises a series of steps or units is not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0024] Embodiments of the present invention provide a locust plague prediction method based on collected data, such as... Figure 1 As shown, the method includes the following steps: S1. Obtain the grayscale image of the target map based on the locust-related dataset corresponding to each initial grid region, wherein the initial grid region is a square region obtained by dividing the target map according to a preset map distance.

[0025] Specifically, the side length of each initial grid area is the same as the preset map distance.

[0026] Specifically, the preset map distance is a distance pre-set by those skilled in the art based on the preset geographical distance and the map scale of the target map. For example, if the map scale of the target map is 1:100,000, that is, 1 centimeter on the target map is equivalent to 100,000 centimeters (i.e., 1 kilometer) on the ground, and the preset geographical distance is 1 kilometer, then the preset map distance is 1 centimeter. The preset geographical distance is set by those skilled in the art according to actual needs, such as 1 kilometer, 2 kilometers, 3 kilometers, which will not be elaborated here.

[0027] Furthermore, any two initial grid regions do not overlap.

[0028] Specifically, the target map is a map of the geographical area where locust plague prediction is to be conducted.

[0029] Specifically, the locust-related dataset includes a list of locust parameters, a list of environmental feature parameters, and a list of climate feature parameters. The locust parameter list includes several locust parameters collected from the actual geographic region corresponding to the initial grid area during the current time period. These parameters include locust type, the number of eggs for each locust type, the breeding cycle for each locust type, the distribution density of eggs, and the distribution characteristics of eggs. The environmental feature parameter list includes several environmental feature parameters collected from the actual geographic region corresponding to the initial grid area during the current time period. These parameters include soil type, vegetation cover, soil moisture, vegetation type, and water conditions. The climate feature parameter list includes several climate feature parameters collected from the actual geographic region corresponding to the initial grid area during the current time period. These parameters include temperature, humidity, and wind speed.

[0030] Specifically, the end time of the current time period is the current time point, and the length of the current time period is set by those skilled in the art according to actual needs, such as 10 days, 20 days, 30 days, which will not be elaborated here.

[0031] Specifically, the dimensions of the locust parameters in the locust parameter lists of the locust-related datasets corresponding to different initial grid regions remain consistent; this can be understood as: different locust parameter lists contain the same number and type of locust parameters.

[0032] Specifically, the dimensions of the environmental feature parameters in the environmental feature parameter lists of locust-related datasets corresponding to different initial grid regions remain consistent; this can be understood as: different environmental feature parameter lists contain the same number and type of environmental feature parameters.

[0033] Specifically, the dimensionality of the climate feature parameters in the climate feature parameter lists of the locust-related datasets corresponding to different initial grid regions remains consistent; this can be understood as: different climate feature parameter lists contain the same number and type of climate feature parameters.

[0034] S2. Use a superpixel algorithm to perform image segmentation on the grayscale image of the target map to obtain a first sub-image list LB including several first sub-images. The first sub-image is an image obtained by image segmentation of the grayscale image of the target map using a superpixel algorithm. Those skilled in the art know that any superpixel algorithm in the prior art is within the protection scope of this invention, such as SLIC, SEEDS and LSC, which will not be described in detail here.

[0035] S3. Two adjacent first sub-images in the grayscale image of the target map are treated as two second sub-images in a second sub-image group to obtain the second sub-image group set B = {B1, B2, ..., B...}. j B n}, Bj Let B be the j-th second sub-image group, where j ranges from 1 to n, and n is the number of second sub-image groups; this can be understood as: B j1 and B j2 Both belong to the first sub-image list and B j1 and B j2 In the grayscale image of the target map, B is adjacent. j1 For B j The first second sub-image in the middle, B j2 For B j The second sub-image in the middle.

[0036] S4, Obtain B j The corresponding locust aggregation probability C j And sort C1, C2, ..., C in descending order. j C n Sort the data to obtain the key cluster probability list D.

[0037] Specifically, D includes n key clustering probabilities, which are the sorted locust clustering probabilities.

[0038] S5. Let a = 1, where a is a preset count value.

[0039] S6, if D a ≥D 0 And E a ≤E 0 Then LB and D 1 a The two first sub-images corresponding to the two second sub-images in the image are merged into one first sub-image so that LB is updated and proceeding to step S3; otherwise, proceeding to step S7, D a Let D be the a-th key cluster probability. 0 To preset the aggregation probability, D 1 a D a The corresponding second sub-image group, E a D 1 a The corresponding variance of grayscale values, E 0 The preset grayscale value variance is used. As those skilled in the art know, the preset clustering probability and the preset grayscale value variance are both preset by those skilled in the art according to actual needs, and will not be elaborated here.

[0040] S7. If a≠n, let a=a+1 and proceed to step S6; if a=n, take the first sub-image in LB as the target sub-image, and mark the target sub-image with the color corresponding to the average occurrence probability of the target event corresponding to the target sub-image, wherein the average occurrence probability of the target event corresponding to the first sub-image is taken as the average occurrence probability of the target event corresponding to its target sub-image.

[0041] Specifically, the target event is a locust plague event occurring at a target time.

[0042] Specifically, different target events correspond to different colors based on their average probability of occurrence.

[0043] Furthermore, the greater the average probability of the target event occurring, the darker its corresponding color, and the greater the likelihood that a locust plague event will occur in the actual geographical area corresponding to the target sub-image at the target time point.

[0044] Through the above steps, a grayscale image of the target map is obtained based on the locust-related dataset. A superpixel algorithm is used to segment the grayscale image of the target map to obtain a first sub-image list. A second sub-image set is then obtained. A key clustering probability list is obtained based on the locust clustering probability corresponding to the second sub-image set. Starting from the first key clustering probability in the key clustering probability list, if the key clustering probability is not less than a preset clustering probability, it indicates that the locusts in the actual geographical areas corresponding to the two second sub-images in the second sub-image set corresponding to the key clustering probability are likely a single locust swarm. If the preset grayscale value variance corresponding to the second sub-image set corresponding to the key clustering probability is not greater than a preset grayscale value variance, it indicates that the locust-related datasets collected from the actual geographical areas corresponding to the two second sub-images in the second sub-image set are very similar. Therefore, when the key clustering probability is not less than the preset clustering probability and the preset grayscale value variance corresponding to the second sub-image set corresponding to the key clustering probability is not greater than the preset grayscale value variance, the two first sub-images corresponding to the two second sub-images are merged into one first sub-image to update the first sub-image and facilitate unified management.

[0045] Furthermore, when the first sub-image is updated, the adjacency relationships between the first sub-images are also updated. At this point, it is necessary to re-acquire the second sub-image set. Otherwise, it indicates that the locusts in the actual geographical areas corresponding to the two second sub-images in the second sub-image set corresponding to the key clustering probability are likely not a locust colony, or the locust-related datasets collected from the actual geographical areas corresponding to the two second sub-images are dissimilar. Therefore, the first sub-images corresponding to the two second sub-images cannot be merged. Thus, the next key clustering probability is judged to determine whether there are any first sub-images that need to be merged. When all the first sub-images in the first sub-image list do not need to be merged, the first sub-images are merged. The first sub-image in the image list is selected as the target sub-image. When the target sub-image is obtained, it is marked with the color corresponding to the average occurrence probability of the target event. Taking into account the aggregation behavior of locusts and mainly relying on locust-related datasets to determine the target sub-image, the target sub-image is marked with the color corresponding to the average occurrence probability of the target event to achieve locust disaster prediction. The data volume is relatively small, and the target sub-image can be quickly identified and colored, so that users can more intuitively observe the probability of locust disasters occurring in the actual geographical area corresponding to each target sub-image, and achieve comprehensive and accurate locust disaster prediction.

[0046] Specifically, step S1 includes the following steps S11-S12: S11. Input the locust-related dataset corresponding to each initial grid region into the preset regression model to obtain the probability of the target event corresponding to each initial grid region and construct a list of the probability of the target event A={A1, A2, ..., A...} i A m}, where A i Let be the probability of the target event occurring in the i-th initial grid region, where i ranges from 1 to m, and m is the number of initial grid regions.

[0047] Specifically, the average probability of the target event corresponding to the first sub-image is the average of the probability of the target event corresponding to all initial grid regions belonging to the first sub-image; the initial grid region belonging to the first sub-image can be understood as the initial grid region that is completely covered by the first sub-image in space.

[0048] Optionally, the preset regression model is a model obtained by those skilled in the art through training a linear regression model for the task of predicting the probability of the occurrence of a target event. The linear regression model has a simple structure, high computational efficiency, is suitable for processing large-scale datasets, and its model parameters are easy to interpret. It can intuitively reflect the relationship between input features (such as locust-related datasets) and output (probability of the occurrence of a target event).

[0049] S12, according to Ai Obtain the grayscale value H corresponding to the i-th initial grid region. i And according to H1, H2, ..., H i H m Generate a grayscale image of the target map, where H i Meets the following conditions: H i =(A i -A min ) / (A max -A min )×255,A min Let A1, A2, ..., A i A m The minimum probability of the target event occurring, A max Let A1, A2, ..., A i A m The probability of the occurrence of the largest target event in the image is known to those skilled in the art. Any existing method for generating a grayscale image based on grayscale values ​​falls within the protection scope of this invention, and will not be elaborated further here.

[0050] By inputting the locust-related dataset into the preset regression model through the above steps, the probability of the target event can be obtained quickly. The preset regression model can intuitively reflect the relationship between the locust-related dataset and the probability of the target event, which is conducive to improving the accuracy of obtaining the probability of the target event. The probability of the target event is mapped to the interval [0, 255] so that the gray values ​​corresponding to the initial grid area can be further used to generate a grayscale map of the target map, which can more intuitively show the probability of locust disaster in the actual geographical area corresponding to each initial grid area.

[0051] Specifically, step S4 includes the following steps: S41-S42 Obtaining C j : S41. If the initial grid region belongs to the second sub-image, then the initial grid region is used as the key grid region corresponding to the second sub-image to obtain B. j1 The corresponding list of key grid regions F j1 and B j2 The corresponding list of key grid regions F j2 F j1 ={F 1 j1 F 2 j1 F x j1 F p(j1) j1}, F xj1 For B j1 The corresponding x-th key grid region, where x ranges from 1 to p(j1), and p(j1) is B. j1 The number of corresponding critical grid regions, F j2 ={F 1 j2 F 2 j2 F y j2 F q(j2) j2}, F y j2 For B j2 The corresponding y-th key grid region, where y takes values ​​from 1 to q(j2), and q(j2) is B. j2 The number of corresponding key grid regions.

[0052] Specifically, the initial grid region belonging to the second sub-image can be understood as the initial grid region being completely covered by the second sub-image in space.

[0053] S42, according to F x j1 and F y j2 Get C j C j Meets the following conditions: C j =W1×(α×(|U j1 -U j2 | / U max_j ))+β j +W2×(KS j ×(1-(M j / M max W1 is the first preset weight value, α is the preset adjustment parameter, and U j1 For F j1 The corresponding average probability of occurrence of the target event, U j2 For F j2 The corresponding average probability of occurrence of the target event, U max_j For B j The corresponding maximum probability of occurrence of the target event, β j For B j The corresponding environmental feature similarity, W2 is the second preset weight value, KS j For B j The corresponding locust dispersal ability value, M j For B j The corresponding obstacle density in the border area, M maxThe preset maximum obstacle density is defined as follows: as those skilled in the art know, the first preset weight value, the preset adjustment parameter, the second preset weight value, and the preset maximum obstacle density are all preset by those skilled in the art according to actual needs. For example, the first preset weight value is 0.5, the preset adjustment parameter is 0.3, the second preset weight value is 0.5, and the preset maximum obstacle density is 0.8. Further details will not be elaborated here.

[0054] Specifically, U j1 U j2 and U max_j Each of the following conditions must be met: U j1 =∑ p(j1) x=1 R x j1 / p(j1), R x j1 For F x j1 The probability of the corresponding target event occurring.

[0055] U j2 =∑ q(j2) y=1 R y j2 / q(j2), R y j2 For F y j2 The probability of the corresponding target event occurring.

[0056] U max_j =max(R 1 j1 R 2 j1 , ..., R x j1 , ..., R p(j1) j1 R 1 j2 R 2 j2 , ..., R y j2 , ..., R q(j2) j2 ), max() is the function to get the maximum value.

[0057] Through the above steps, key grid region lists corresponding to the two second sub-images are obtained respectively. Based on these lists, the locust aggregation probability corresponding to the second sub-image group is obtained. The higher the locust aggregation probability, the more likely the locusts in the two actual geographical areas corresponding to the two second sub-images in the second sub-image group are to form a locust swarm. The locust aggregation probabilities corresponding to all second sub-image groups are sorted in descending order to obtain a key aggregation probability list. Starting from the first key aggregation probability in the list, if the key aggregation probability is not less than a preset aggregation probability, it indicates that the second sub-image corresponding to the key aggregation probability is... The locusts in the actual geographical area corresponding to the two second sub-images in the image group are likely to be a locust swarm. If the preset gray value variance corresponding to the second sub-image group with the key clustering probability is not greater than the preset gray value variance, it means that the locust-related datasets collected from the actual geographical area corresponding to the two second sub-images in the second sub-image group are very similar. Therefore, when the key clustering probability is not less than the preset clustering probability and the preset gray value variance corresponding to the second sub-image group with the key clustering probability is not greater than the preset gray value variance, the two first sub-images corresponding to the two second sub-images are merged into one first sub-image to update the first sub-image, which facilitates unified management and helps to avoid resource waste.

[0058] Specifically, step S42 further includes the following steps S421-S423 to obtain β. j : S421, respectively for F x j1 All environmental feature parameters in the environmental feature parameter list of the corresponding locust-related dataset and F y j2 Vectorize all environmental feature parameters in the environmental feature parameter list of the corresponding locust-related dataset to obtain F. x j1 The corresponding first environmental feature vector K x j1 and F y j2 The corresponding second environmental feature vector K y j2 As those skilled in the art will know, the vectorization processing methods used to obtain the first environmental feature vector and the second environmental feature vector are the same method. Any vectorization processing method in the prior art is within the protection scope of this invention, and will not be described in detail here.

[0059] Specifically, the vector dimension of the first environmental feature vector is the same as the dimension of the environmental feature parameters in its corresponding environmental feature parameter list. This can be understood as: the vector values ​​in the first environmental feature vector correspond one-to-one with the environmental feature parameters in its corresponding environmental feature parameter list.

[0060] Specifically, the vector dimension of the second environmental feature vector is the same as the dimension of the environmental feature parameters in its corresponding environmental feature parameter list. This can be understood as: the vector values ​​in the second environmental feature vector correspond one-to-one with the environmental feature parameters in its corresponding environmental feature parameter list.

[0061] S422, K 1 j1 K 2 j1 , ..., K x j1 , ..., K p(j1) j1 Merge into B j The corresponding third environment feature vector L j1 ; K 1 j2 K 2 j2 , ..., K y j2 , ..., K q(j2) j2 Merge into B j The corresponding fourth environmental feature vector L j2 As those skilled in the art will know, the vector merging method used to obtain the third environmental feature vector and the vector merging method used to obtain the fourth environmental feature vector are the same method. Any vector merging method in the prior art that merges multiple vectors into the same vector is within the protection scope of this invention, such as the splicing method and the averaging method, which will not be elaborated here.

[0062] S423, L j1 With L j2 The vector similarity between them is used as β j .

[0063] Specifically, β j The larger L is j1 and L j2 The more similar they are.

[0064] Through the above steps, by vectorizing all environmental feature parameters in the list of environmental feature parameters corresponding to the key grid regions, the first environmental feature vector and the second environmental feature vector are obtained. Further, the first environmental feature vectors corresponding to all key grid regions in the first sub-image of the second sub-image group are merged to obtain the third environmental feature vector corresponding to the second sub-image group. The second environmental feature vectors corresponding to all key grid regions in the second sub-image of the second sub-image group are merged to obtain the fourth environmental feature vector corresponding to the second sub-image group. The vector similarity between the third and fourth environmental feature vectors corresponding to the second sub-image group is used as the environmental feature similarity of the second sub-image group. This considers each environmental feature parameter corresponding to each key grid region in the second sub-image, which helps improve the accuracy of obtaining the environmental feature similarity. Furthermore, obtaining the locust aggregation probability based on the environmental feature similarity helps improve the accuracy of obtaining the locust aggregation probability.

[0065] Specifically, step S42 also includes the following steps S4201-S4204: obtaining KS j : S4201, for F j1 and F j2 Deduplication was performed on all locust types in the locust parameter list of all key grid regions corresponding to the locust-related dataset to obtain B. j The corresponding list of target locust types N j ={N j1 N j2 ,…,N jr ,…,N js(j)}, N jr For B j The corresponding r-th target locust type, where r ranges from 1 to s(j), and s(j) is B. j The number of the corresponding target locust types.

[0066] S4202, F j1 and F j2 The locust parameter list in the locust-related dataset corresponding to each key grid region in the dataset is related to N. jr The sum of the number of eggs of the same locust type is used as N. jr The corresponding number of target insect eggs Q jr .

[0067] S4203, max(Q) j1 Q j2 Q jr Q js The corresponding target locust type is used as B. jThe corresponding key locust types.

[0068] S4204, B j The value corresponding to the total distance traveled by a locust swarm of the corresponding key locust type within a preset time period is used as KS. j As those skilled in the art will know, the preset duration is set by those skilled in the art according to actual needs, such as 1 day or 2 days. Any method in the prior art for obtaining the overall movement distance of a locust swarm within the preset duration is within the protection scope of this invention, and will not be described further here.

[0069] Specifically, the quotient obtained by dividing the total moving distance by the reference moving distance is taken as the value corresponding to the total moving distance, where the reference moving distance is 1. The unit of measurement for the reference moving distance is the same as that for the total moving distance. For example, if the total moving distance is 5 meters and the reference moving distance is 1 meter, then the value corresponding to the total moving distance is 5.

[0070] Through the above steps, the locust with the largest sum of the number of eggs among all locust types corresponding to the two second sub-images in the second sub-image group is taken as the key locust type corresponding to the second sub-image group. This can identify the locust types that may cause locust plagues. The value corresponding to the overall movement distance of the locust swarm of the key locust type in the second sub-image group within a preset time period is taken as the locust diffusion ability value of the second sub-image group. Furthermore, obtaining the locust aggregation probability based on the locust diffusion ability value is beneficial to improving the accuracy of obtaining the locust aggregation probability.

[0071] Specifically, step S42 also includes the following steps S01-S03 to obtain M. j : S01, Obtain B j The corresponding key geographical area, B j The corresponding key geographical region is B. j The corresponding connected image region corresponds to the actual geographic region, B j The corresponding connected image region is B j1 and B j2 Using the shared edge as the reference, respectively towards B j1 and B j2 The image area obtained by inner expansion z is a preset map expansion distance, which is set by those skilled in the art according to actual needs, such as 0.1 cm or 0.2 cm, and will not be elaborated here.

[0072] S02, Obtain B j The total area S of all ground features within the corresponding key geographical area that can hinder locust migration j .

[0073] Specifically, geographical features that can hinder locust migration include: rivers, lakes, wetlands, mountains, dense forests, urban building complexes, industrial facilities, highways, railways, reservoirs, and dams.

[0074] S03, S j Divide by F j The quotient obtained from the area of ​​the corresponding key geographical region is M. j .

[0075] The steps described above obtain the key geographic region corresponding to the second sub-image group. This can be understood as obtaining the border area between the two actual geographic regions corresponding to the two second sub-images in the second sub-image group. The quotient obtained by dividing the total area of ​​all ground features that can hinder locust migration within the key geographic region by the area of ​​the key geographic region is used as the obstacle density of the border area. The higher the obstacle density of the border area, the larger the proportion of the total area of ​​all ground features that can hinder locust migration within the key geographic region, and the more difficult it is for locusts to migrate from one side of the key geographic region to the other. Conversely, the lower the obstacle density of the border area, the smaller the proportion of the total area of ​​all ground features that can hinder locust migration within the key geographic region, and the easier it is for locusts to migrate from one side of the key geographic region to the other, which may cause locust aggregation. Therefore, obtaining the locust aggregation probability based on the obstacle density of the border area is beneficial to improving the accuracy of obtaining the locust aggregation probability.

[0076] Specifically, step S6 includes the following steps: S61-S62 Obtaining E a : S61, D 1 a The union of the lists of key grid regions corresponding to the two second sub-images in D is used as D. 1 a The corresponding list of intermediate grid regions T a ={T a1 T a2 ,…,T ae ,…,T af(a)}, T ae D 1 a The corresponding e-th intermediate grid region, where e takes values ​​from 1 to f(a), and f(a) is D. 1 a The number of corresponding intermediate grid regions.

[0077] S62, V a1 V a2 , ..., V ae , ..., V af(a) The variance as E a V aeFor T ae The corresponding grayscale value.

[0078] Through the above steps, the variance of the gray values ​​corresponding to all key grid regions of the two second sub-images in the second sub-image group is taken as the gray value variance of the second sub-image group. If the preset gray value variance of the second sub-image group is not greater than the preset gray value variance, it indicates that the locust-related datasets collected from the actual geographical areas corresponding to the two second sub-images in the second sub-image group are very similar; otherwise, it indicates that the locust-related datasets collected from the actual geographical areas corresponding to the two second sub-images in the second sub-image group are not similar. Therefore, when the key clustering probability is not less than the preset clustering probability and the preset gray value variance of the second sub-image group corresponding to the key clustering probability is not greater than the preset gray value variance, the two first... Sub-images are merged into a first sub-image to facilitate updating the first sub-image and unified management; otherwise, it indicates that the locusts in the actual geographical areas corresponding to the two second sub-images in the second sub-image group corresponding to the key clustering probability are likely not a locust swarm, or the locust-related datasets collected in the actual geographical areas corresponding to the two second sub-images are not similar, and the first sub-images corresponding to the two second sub-images cannot be merged. The target sub-image is mainly determined by the locust-related dataset, and the target sub-image is marked with the color corresponding to the average occurrence probability of the target event corresponding to the target sub-image to achieve locust disaster prediction. The data volume is small, and the target sub-image can be quickly determined and colored to achieve comprehensive and accurate locust disaster prediction.

[0079] Embodiments of the present invention also provide a locust plague prediction system based on collected data, such as... Figure 2 As shown, the system includes: Grayscale image acquisition module 1 is used to acquire grayscale images of the target map based on the locust-related dataset corresponding to each initial grid region. The initial grid region is a square region obtained by dividing the target map according to a preset map distance.

[0080] Specifically, the side length of each initial grid area is the same as the preset map distance.

[0081] Specifically, the preset map distance is a distance pre-set by those skilled in the art based on the preset geographical distance and the map scale of the target map. For example, if the map scale of the target map is 1:100,000, that is, 1 centimeter on the target map is equivalent to 100,000 centimeters (i.e., 1 kilometer) on the ground, and the preset geographical distance is 1 kilometer, then the preset map distance is 1 centimeter. The preset geographical distance is set by those skilled in the art according to actual needs, such as 1 kilometer, 2 kilometers, 3 kilometers, which will not be elaborated here.

[0082] Furthermore, any two initial grid regions do not overlap.

[0083] Specifically, the target map is a map of the geographical area where locust plague prediction is to be conducted.

[0084] Specifically, the locust-related dataset includes a list of locust parameters, a list of environmental feature parameters, and a list of climate feature parameters. The locust parameter list includes several locust parameters collected from the actual geographic region corresponding to the initial grid area during the current time period. These parameters include locust type, the number of eggs for each locust type, the breeding cycle for each locust type, the distribution density of eggs, and the distribution characteristics of eggs. The environmental feature parameter list includes several environmental feature parameters collected from the actual geographic region corresponding to the initial grid area during the current time period. These parameters include soil type, vegetation cover, soil moisture, vegetation type, and water conditions. The climate feature parameter list includes several climate feature parameters collected from the actual geographic region corresponding to the initial grid area during the current time period. These parameters include temperature, humidity, and wind speed.

[0085] Specifically, the end time of the current time period is the current time point, and the length of the current time period is set by those skilled in the art according to actual needs, such as 10 days, 20 days, 30 days, which will not be elaborated here.

[0086] Specifically, the dimensions of the locust parameters in the locust parameter lists of the locust-related datasets corresponding to different initial grid regions remain consistent; this can be understood as: different locust parameter lists contain the same number and type of locust parameters.

[0087] Specifically, the dimensions of the environmental feature parameters in the environmental feature parameter lists of locust-related datasets corresponding to different initial grid regions remain consistent; this can be understood as: different environmental feature parameter lists contain the same number and type of environmental feature parameters.

[0088] Specifically, the dimensionality of the climate feature parameters in the climate feature parameter lists of the locust-related datasets corresponding to different initial grid regions remains consistent; this can be understood as: different climate feature parameter lists contain the same number and type of climate feature parameters.

[0089] The first sub-image acquisition module 2 is used to perform image segmentation on the grayscale image of the target map using a superpixel algorithm to obtain a first sub-image list LB including several first sub-images. The first sub-image is an image obtained by performing image segmentation on the grayscale image of the target map using a superpixel algorithm. Those skilled in the art know that any superpixel algorithm in the prior art is within the protection scope of this invention, such as SLIC, SEEDS and LSC, which will not be described in detail here.

[0090] The second sub-image group acquisition module 3 is used to obtain a second sub-image group set B={B1, B2, ..., B...} by taking two adjacent first sub-images in the grayscale image of the target map as two second sub-images in a second sub-image group. j B n}, B j Let B be the j-th second sub-image group, where j ranges from 1 to n, and n is the number of second sub-image groups; this can be understood as: B j1 and B j2 Both belong to the first sub-image list and B j1 and B j2 In the grayscale image of the target map, B is adjacent. j1 For B j The first second sub-image in the middle, B j2 For B j The second sub-image in the middle.

[0091] Key Cluster Probability List Acquisition Module 4 is used to obtain B j The corresponding locust aggregation probability C j And sort C1, C2, ..., C in descending order. j C n Sort the data to obtain the key cluster probability list D.

[0092] Specifically, D includes n key clustering probabilities, which are the sorted locust clustering probabilities.

[0093] The count value initialization module 5 is used to set a=1, where a is a preset count value.

[0094] The first sub-image update module 6 is used if D a ≥D 0 And E a ≤E 0 Then LB and D 1 a The two first sub-images corresponding to the two second sub-images in the image are merged into one first sub-image so that the LB is updated and the image enters the second sub-image group acquisition module 3; otherwise, the image enters the target sub-image acquisition module 7. a Let D be the a-th key cluster probability. 0 To preset the aggregation probability, D 1 a D a The corresponding second sub-image group, E a D 1 a The corresponding variance of grayscale values, E 0The preset grayscale value variance is used. As those skilled in the art know, the preset clustering probability and the preset grayscale value variance are both preset by those skilled in the art according to actual needs, and will not be elaborated here.

[0095] The target sub-image acquisition module 7 is used to set a=a+1 and enter the first sub-image update module 6 if a≠n; if a=n, the first sub-image in LB is used as the target sub-image, and the target sub-image is marked with the color corresponding to the average occurrence probability of the target event corresponding to the target sub-image, wherein the average occurrence probability of the target event corresponding to the first sub-image is used as the average occurrence probability of the target event corresponding to its corresponding target sub-image.

[0096] Specifically, the target event is a locust plague event occurring at a target time.

[0097] Specifically, different target events correspond to different colors based on their average probability of occurrence.

[0098] Furthermore, the greater the average probability of the target event occurring, the darker its corresponding color, and the greater the likelihood that a locust plague event will occur in the actual geographical area corresponding to the target sub-image at the target time point.

[0099] Specifically, grayscale image acquisition module 1 includes: The target event occurrence probability acquisition unit is used to input the locust-related dataset corresponding to each initial grid region into a preset regression model to obtain the target event occurrence probability corresponding to each initial grid region and construct a target event occurrence probability list A={A1, A2, ..., A...} i A m}, where A i Let be the probability of the target event occurring in the i-th initial grid region, where i ranges from 1 to m, and m is the number of initial grid regions.

[0100] Specifically, the average probability of the target event corresponding to the first sub-image is the average of the probability of the target event corresponding to all initial grid regions belonging to the first sub-image; the initial grid region belonging to the first sub-image can be understood as the initial grid region that is completely covered by the first sub-image in space.

[0101] Optionally, the preset regression model is a model obtained by those skilled in the art through training a linear regression model for the task of predicting the probability of the occurrence of a target event. The linear regression model has a simple structure, high computational efficiency, is suitable for processing large-scale datasets, and its model parameters are easy to interpret. It can intuitively reflect the relationship between input features (such as locust-related datasets) and output (probability of the occurrence of a target event).

[0102] The grayscale image generation unit is used to generate grayscale images based on A. iObtain the grayscale value H corresponding to the i-th initial grid region. i And according to H1, H2, ..., H i H m Generate a grayscale image of the target map, where H i Meets the following conditions: H i =(A i -A min ) / (A max -A min )×255,A min Let A1, A2, ..., A i A m The minimum probability of the target event occurring, A max Let A1, A2, ..., A i A m The probability of the occurrence of the largest target event in the image is known to those skilled in the art. Any existing method for generating a grayscale image based on grayscale values ​​falls within the protection scope of this invention, and will not be elaborated further here.

[0103] Specifically, the key clustering probability list acquisition module 4 also includes: The key grid region list acquisition unit is used to, if the initial grid region belongs to the second sub-image, then use the initial grid region as the key grid region corresponding to the second sub-image to obtain B. j1 The corresponding list of key grid regions F j1 and B j2 The corresponding list of key grid regions F j2 F j1 ={F 1 j1 F 2 j1 F x j1 F p(j1) j1}, F x j1 For B j1 The corresponding x-th key grid region, where x ranges from 1 to p(j1), and p(j1) is B. j1 The number of corresponding key grid regions, F j2 ={F 1 j2 F 2 j2 F y j2 F q(j2) j2}, F y j2 For B j2The corresponding y-th key grid region, where y takes values ​​from 1 to q(j2), and q(j2) is B. j2 The number of corresponding key grid regions.

[0104] Specifically, the initial grid region belonging to the second sub-image can be understood as the initial grid region being completely covered by the second sub-image in space.

[0105] The locust aggregation probability acquisition unit is used to obtain the probability based on F. x j1 and F y j2 Get C j C j Meets the following conditions: C j =W1×(α×(|U j1 -U j2 | / U max_j ))+β j +W2×(KS j ×(1-(M j / M max W1 is the first preset weight value, α is the preset adjustment parameter, and U j1 For F j1 The corresponding average probability of occurrence of the target event, U j2 For F j2 The corresponding average probability of occurrence of the target event, U max_j For B j The corresponding maximum probability of occurrence of the target event, β j For B j The corresponding environmental feature similarity, W2 is the second preset weight value, KS j For B j The corresponding locust dispersal ability value, M j For B j The corresponding obstacle density in the border area, M max The preset maximum obstacle density is defined as follows: as those skilled in the art know, the first preset weight value, the preset adjustment parameter, the second preset weight value, and the preset maximum obstacle density are all preset by those skilled in the art according to actual needs. For example, the first preset weight value is 0.5, the preset adjustment parameter is 0.3, the second preset weight value is 0.5, and the preset maximum obstacle density is 0.8. Further details will not be elaborated here.

[0106] Specifically, U j1 U j2 and U max_j They each meet the following conditions: U j1 =∑ p(j1) x=1 Rx j1 / p(j1), R x j1 For F x j1 The probability of the corresponding target event occurring.

[0107] U j2 =∑ q(j2) y=1 R y j2 / q(j2), R y j2 For F y j2 The probability of the corresponding target event occurring.

[0108] U max_j =max(R 1 j1 R 2 j1 , ..., R x j1 , ..., R p(j1) j1 R 1 j2 R 2 j2 , ..., R y j2 , ..., R q(j2) j2 ), max() is the function to get the maximum value.

[0109] Specifically, the locust aggregation probability acquisition unit includes: The environmental feature vector acquisition sub-unit is used to obtain F respectively. x j1 All environmental feature parameters in the environmental feature parameter list of the corresponding locust-related dataset and F y j2 Vectorize all environmental feature parameters in the environmental feature parameter list of the corresponding locust-related dataset to obtain F. x j1 The corresponding first environmental feature vector K x j1 and F y j2 The corresponding second environmental feature vector K y j2 As those skilled in the art will know, the vectorization processing method used to obtain the first environmental feature vector and the method used to obtain the second environmental feature vector are the same method. Any vectorization processing method in the prior art is within the protection scope of this invention, and will not be described in detail here.

[0110] Specifically, the vector dimension of the first environmental feature vector is the same as the dimension of the environmental feature parameters in its corresponding environmental feature parameter list. This can be understood as: the vector values ​​in the first environmental feature vector correspond one-to-one with the environmental feature parameters in its corresponding environmental feature parameter list.

[0111] Specifically, the vector dimension of the second environmental feature vector is the same as the dimension of the environmental feature parameters in its corresponding environmental feature parameter list. This can be understood as: the vector values ​​in the second environmental feature vector correspond one-to-one with the environmental feature parameters in its corresponding environmental feature parameter list.

[0112] Environmental feature vector merging subunit, used to combine K 1 j1 K 2 j1 , ..., K x j1 , ..., K p(j1) j1 Merge into B j The corresponding third environment feature vector L j1 ; K 1 j2 K 2 j2 , ..., K y j2 , ..., K q(j2) j2 Merge into B j The corresponding fourth environmental feature vector L j2 As those skilled in the art will know, the vector merging method used to obtain the third environmental feature vector and the vector merging method used to obtain the fourth environmental feature vector are the same method. Any vector merging method in the prior art that merges multiple vectors into the same vector is within the protection scope of this invention, such as the splicing method and the averaging method, which will not be elaborated here.

[0113] The environmental feature similarity acquisition subunit is used to obtain L j1 With L j2 The vector similarity between them is used as β j .

[0114] Specifically, β j The larger L is j1 and L j2 The more similar they are.

[0115] Specifically, the locust aggregation probability acquisition unit also includes: The target locust type acquisition subunit is used for F j1 and F j2Deduplication was performed on all locust types in the locust parameter list of all key grid regions corresponding to the locust-related dataset to obtain B. j The corresponding list of target locust types N j ={N j1 N j2 ,…,N jr ,…,N js(j)}, N jr For B j The corresponding r-th target locust type, where r ranges from 1 to s(j), and s(j) is B. j The number of the corresponding target locust types.

[0116] The target egg quantity acquisition sub-unit is used to obtain the F j1 and F j2 The locust parameter list in the locust-related dataset corresponding to each key grid region in the dataset is related to N. jr The sum of the number of eggs of the same locust type is used as N. jr The corresponding number of target insect eggs Q jr .

[0117] Key locust type acquisition subunit, used to max(Q) j1 Q j2 Q jr Q js The corresponding target locust type is used as B. j The corresponding key locust types.

[0118] The locust dispersal ability value acquisition sub-unit is used to obtain B. j The value corresponding to the total distance traveled by a locust swarm of the corresponding key locust type within a preset time period is used as KS. j As those skilled in the art will know, the preset duration is set by those skilled in the art according to actual needs, such as 1 day or 2 days. Any method in the prior art for obtaining the overall movement distance of a locust swarm within the preset duration is within the protection scope of this invention, and will not be described further here.

[0119] Specifically, the quotient obtained by dividing the total moving distance by the reference moving distance is taken as the value corresponding to the total moving distance, where the reference moving distance is 1. The unit of measurement for the reference moving distance is the same as that for the total moving distance. For example, if the total moving distance is 5 meters and the reference moving distance is 1 meter, then the value corresponding to the total moving distance is 5.

[0120] Specifically, the locust aggregation probability acquisition unit also includes: Key geographic area acquisition sub-cells are used to acquire B. j The corresponding key geographical area, Bj The corresponding key geographical region is B. j The corresponding connected image region corresponds to the actual geographic region, B j The corresponding connected image region is B j1 and B j2 Using the shared edge as the reference, respectively towards B j1 and B j2 The image area obtained by inner expansion z is a preset map expansion distance, which is set by those skilled in the art according to actual needs, such as 0.1 cm or 0.2 cm, and will not be elaborated here.

[0121] Area acquisition subunit, used to acquire B j The total area S of all ground features within the corresponding key geographical area that can hinder locust migration j .

[0122] Specifically, geographical features that can hinder locust migration include: rivers, lakes, wetlands, mountains, dense forests, urban building complexes, industrial facilities, highways, railways, reservoirs, and dams.

[0123] The obstacle density acquisition subunit in the border area is used to obtain S j Divide by F j The quotient obtained from the area of ​​the corresponding key geographical region is M. j .

[0124] Specifically, the first sub-image update module 6 includes: The intermediate grid region acquisition unit is used to obtain D 1 a The union of the lists of key grid regions corresponding to the two second sub-images in D is used as D. 1 a The corresponding list of intermediate grid regions T a ={T a1 T a2 ,…,T ae ,…,T af(a)}, T ae D 1 a The corresponding e-th intermediate grid region, where e takes values ​​from 1 to f(a), and f(a) is D. 1 a The number of corresponding intermediate grid regions.

[0125] The grayscale value variance acquisition unit is used to obtain V a1 V a2 , ..., V ae , ..., V af(a) The variance as E a V ae For Tae The corresponding grayscale value.

[0126] This invention provides a method and system for predicting locust plagues based on collected data. The method acquires a grayscale image of a target map from a locust-related dataset, uses a superpixel algorithm to segment the grayscale image to obtain a first sub-image list, further acquires a second sub-image set, obtains a key aggregation probability list based on the locust aggregation probabilities corresponding to the second sub-image sets, and determines whether to merge and update the first sub-images or directly use the first sub-images as target sub-images based on the variance of the grayscale values ​​corresponding to the key aggregation probabilities and the second sub-image sets corresponding to the key aggregation probabilities. When a target sub-image is acquired, it is marked with the color corresponding to the average occurrence probability of the target event corresponding to the target sub-image, where the target event is a locust plague event occurring at a target time point. This invention considers the aggregation behavior of locusts and mainly relies on a locust-related dataset to determine the target sub-images. It uses the color corresponding to the average occurrence probability of the target event corresponding to the target sub-images to mark them, thus achieving locust plague prediction. The data volume is relatively small, allowing for rapid determination and color marking of target sub-images, enabling comprehensive and accurate locust plague prediction.

[0127] Embodiments of the present invention also provide a non-transitory computer-readable storage medium that can be disposed in an electronic device to store a computer program related to implementing a method in the method embodiments, the computer program being loaded and executed by the processor to implement the method provided in the above embodiments.

[0128] Embodiments of the present invention also provide an electronic device, including: a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method provided in the above embodiments.

[0129] Embodiments of the present invention also provide a computer program product including program code, which, when the program product is run on an electronic device, causes the electronic device to perform the steps of the methods described above in various exemplary embodiments of the present invention.

[0130] While specific embodiments of the invention have been described in detail by way of examples, those skilled in the art should understand that the examples are for illustrative purposes only and are not intended to limit the scope of the invention. Those skilled in the art should also understand that various modifications can be made to the embodiments without departing from the scope and spirit of the invention.

Claims

1. A method for predicting locust plagues based on collected data, characterized in that, The method includes the following steps: S1. Obtain the grayscale image of the target map based on the locust-related dataset corresponding to each initial grid area; S2. Use the superpixel algorithm to perform image segmentation on the grayscale image of the target map to obtain a list of first sub-images LB, which includes several first sub-images; S3. Two adjacent first sub-images in the grayscale image of the target map are treated as two second sub-images in a second sub-image group to obtain the second sub-image group set B = {B1, B2, ..., B...}. j B n }, B j Let j be the j-th second sub-image group, where j ranges from 1 to n, and n is the number of second sub-image groups; S4, Obtain B j The corresponding locust aggregation probability C j And sort C1, C2, ..., C in descending order. j C n Sort the data to obtain the key cluster probability list D; include: S41. If the initial grid region belongs to the second sub-image, then the initial grid region is used as the key grid region corresponding to the second sub-image to obtain B. j1 The corresponding list of key grid regions F j1 and B j2 The corresponding list of key grid regions F j2 F j1 ={F 1 j1 F 2 j1 F x j1 F p(j1) j1 }, F x j1 For B j1 The corresponding x-th key grid region, where x ranges from 1 to p(j1), and p(j1) is B. j1 The number of corresponding critical grid regions, F j2 ={F 1 j2 F 2 j2 F y j2 F q(j2) j2 }, F y j2 For B j2 The corresponding y-th key grid region, where y takes values ​​from 1 to q(j2), and q(j2) is B. j2 The number of corresponding key grid regions, B j1 For B j The first second sub-image in the middle, B j2 For B j The second sub-image in the middle; S42, according to F x j1 and F y j2 Get C j C j The following conditions must be met: C j =W1×(α×(|U j1 -U j2 | / U max_j ))+β j +W2×(KS j ×(1-(M j / M max W1 is the first preset weight value, α is the preset adjustment parameter, and U j1 For F j1 The corresponding average probability of occurrence of the target event, U j2 For F j2 The corresponding average probability of occurrence of the target event, U max_j For B j The corresponding maximum probability of occurrence of the target event, β j For B j The corresponding environmental feature similarity, W2 is the second preset weight value, KS j For B j The corresponding locust dispersal ability value, M j For B j The corresponding obstacle density in the border area, M max U is the preset maximum obstacle density; where U j1 U j2 and U max_j They each meet the following conditions: U j1 =∑ p(j1) x=1 R x j1 / p(j1), R x j1 For F x j1 The corresponding probability of the target event occurring; U j2 =∑ q(j2) y=1 R y j2 / q(j2), R y j2 For F y j2 The corresponding probability of the target event occurring; U max_j =max(R 1 j1 R 2 j1 , ..., R x j1 , ..., R p(j1) j1 R 1 j2 R 2 j2 , ..., R y j2 , ..., R q(j2) j2 `max()` is the function to get the maximum value. S5. Let a = 1, where a is a preset count value; S6, if D a ≥D 0 And E a ≤E 0 Then LB and D 1 a The two first sub-images corresponding to the two second sub-images in the image are merged into one first sub-image so that LB is updated and proceeding to step S3; otherwise, proceeding to step S7, D a Let D be the a-th key cluster probability. 0 To preset the aggregation probability, D 1 a D a The corresponding second sub-image group, E a D 1 a The corresponding variance of grayscale values, E 0 The preset grayscale value variance; S7. If a≠n, let a=a+1 and proceed to step S6. If a=n, take the first sub-image in LB as the target sub-image and mark the target sub-image with the color corresponding to the average occurrence probability of the target event.

2. The locust plague prediction method based on collected data according to claim 1, characterized in that, The initial grid area is a square area obtained by dividing the target map according to the preset map distance.

3. The locust plague prediction method based on collected data according to claim 1, characterized in that, The locust-related dataset includes a list of locust parameters, a list of environmental feature parameters, and a list of climate feature parameters. The locust parameter list includes several locust parameters collected from the actual geographic region corresponding to the initial grid area during the current time period. The locust parameters include locust type, the number of eggs corresponding to each locust type, the breeding cycle of each locust type, the distribution density of eggs, and the distribution characteristics of eggs. The environmental feature parameter list includes several environmental feature parameters collected from the actual geographic region corresponding to the initial grid area during the current time period. The climate feature parameter list includes several climate feature parameters collected from the actual geographic region corresponding to the initial grid area during the current time period.

4. The locust plague prediction method based on collected data according to claim 1, characterized in that, The target event is a locust plague event occurring at the target time point.

5. The locust plague prediction method based on collected data according to claim 3, characterized in that, Step S1 includes the following steps S11-S12: S11. Input the locust-related dataset corresponding to each initial grid region into the preset regression model to obtain the probability of the target event corresponding to each initial grid region and construct a list of the probability of the target event A={A1, A2, ..., A...} i A m }, where A i Let be the probability of the target event occurring in the i-th initial grid region, where i ranges from 1 to m, and m is the number of initial grid regions; S12, according to A i Obtain the grayscale value H corresponding to the i-th initial grid region. i And according to H1, H2, ..., H i H m Generate a grayscale image of the target map, where H i The following conditions must be met: H i =(A i -A min ) / (A max -A min )×255,A min Let A1, A2, ..., A i A m The minimum probability of the target event occurring, A max Let A1, A2, ..., A i A m The probability of the highest target event occurring.

6. The locust plague prediction method based on collected data according to claim 3, characterized in that, Step S42 also includes the following steps S421-S423 to obtain β. j : S421, respectively for F x j1 All environmental feature parameters in the environmental feature parameter list of the corresponding locust-related dataset and F y j2 Vectorize all environmental feature parameters in the environmental feature parameter list of the corresponding locust-related dataset to obtain F. x j1 The corresponding first environmental feature vector K x j1 and F y j2 The corresponding second environmental feature vector K y j2 ; S422, K 1 j1 K 2 j1 , ..., K x j1 , ..., K p(j1) j1 Merge into B j The corresponding third environment feature vector L j1 ; K 1 j2 K 2 j2 , ..., K y j2 , ..., K q(j2) j2 Merge into B j The corresponding fourth environmental feature vector L j2 ; S423, L j1 With L j2 The vector similarity between them is used as β j .

7. The locust plague prediction method based on collected data according to claim 3, characterized in that, Step S42 also includes the following steps S4201-S4204: obtaining KS j : S4201, for F j1 and F j2 Deduplication was performed on all locust types in the locust parameter list of all key grid regions corresponding to the locust-related dataset to obtain B. j The corresponding list of target locust types N j ={N j1 N j2 ,…,N jr ,…,N js(j) }, N jr For B j The corresponding r-th target locust type, where r ranges from 1 to s(j), and s(j) is B. j The corresponding number of target locust types; S4202, F j1 and F j2 The locust parameter list in the locust-related dataset corresponding to each key grid region in the dataset is related to N. jr The sum of the number of eggs of the same locust type is used as N. jr The corresponding number of target insect eggs Q jr ; S4203, max(Q) j1 Q j2 Q jr Q js The corresponding target locust type is used as B. j The corresponding key locust types; S4204, B j The value corresponding to the total distance traveled by a locust swarm of the corresponding key locust type within a preset time period is used as KS. j .

8. The locust plague prediction method based on collected data according to claim 1, characterized in that, Step S42 also includes the following steps S01-S03 to obtain M. j : S01, Obtain B j The corresponding key geographical area, B j The corresponding key geographical region is B. j The corresponding connected image region corresponds to the actual geographic region, B j The corresponding connected image region is B j1 and B j2 Using the shared edge as the reference, respectively towards B j1 and B j2 The image region obtained by inner expansion z, where z is the preset map expansion distance; S02, Obtain B j The total area S of all ground features within the corresponding key geographical area that can hinder locust migration j ; S03, S j Divide by F j The quotient obtained from the area of ​​the corresponding key geographical region is M. j .

9. A locust plague prediction system based on collected data, characterized in that, The system includes: The grayscale image acquisition module is used to acquire grayscale images of the target map based on the locust-related dataset corresponding to each initial grid area. The first sub-image acquisition module is used to perform image segmentation on the grayscale image of the target map using a superpixel algorithm to obtain a first sub-image list LB that includes several first sub-images; The second sub-image group acquisition module is used to obtain a second sub-image group set B={B1, B2, ..., B...} by taking two adjacent first sub-images in the grayscale image of the target map as two second sub-images in a second sub-image group. j B n }, B j Let j be the j-th second sub-image group, where j ranges from 1 to n, and n is the number of second sub-image groups; The key clustering probability list acquisition module is used to obtain B. j The corresponding locust aggregation probability C j And sort C1, C2, ..., C in descending order. j C n Sort the data to obtain a list D of key cluster probabilities; including: S41. If the initial grid region belongs to the second sub-image, then the initial grid region is used as the key grid region corresponding to the second sub-image to obtain B. j1 The corresponding list of key grid regions F j1 and B j2 The corresponding list of key grid regions F j2 F j1 ={F 1 j1 F 2 j1 F x j1 F p(j1) j1 }, F x j1 For B j1 The corresponding x-th key grid region, where x ranges from 1 to p(j1), and p(j1) is B. j1 The number of corresponding critical grid regions, F j2 ={F 1 j2 F 2 j2 F y j2 F q(j2) j2 }, F y j2 For B j2 The corresponding y-th key grid region, where y takes values ​​from 1 to q(j2), and q(j2) is B. j2 The number of corresponding key grid regions, B j1 For B j The first second sub-image in the middle, B j2 For B j The second sub-image in the middle; S42, according to F x j1 and F y j2 Get C j C j The following conditions must be met: C j =W1×(α×(|U j1 -U j2 | / U max_j ))+β j +W2×(KS j ×(1-(M j / M max W1 is the first preset weight value, α is the preset adjustment parameter, and U j1 For F j1 The corresponding average probability of occurrence of the target event, U j2 For F j2 The corresponding average probability of occurrence of the target event, U max_j For B j The corresponding maximum probability of occurrence of the target event, β j For B j The corresponding environmental feature similarity, W2 is the second preset weight value, KS j For B j The corresponding locust dispersal ability value, M j For B j The corresponding obstacle density in the border area, M max U is the preset maximum obstacle density; where U j1 U j2 and U max_j They each meet the following conditions: U j1 =∑ p(j1) x=1 R x j1 / p(j1), R x j1 For F x j1 The corresponding probability of the target event occurring; U j2 =∑ q(j2) y=1 R y j2 / q(j2), R y j2 For F y j2 The corresponding probability of the target event occurring; U max_j =max(R 1 j1 R 2 j1 , ..., R x j1 , ..., R p(j1) j1 R 1 j2 R 2 j2 , ..., R y j2 , ..., R q(j2) j2 `max()` is the function to get the maximum value. The count value initialization module is used to set a=1, where a is a preset count value; The first sub-image update module is used if D a ≥D 0 And E a ≤E 0 Then LB and D 1 a The two first sub-images corresponding to the two second sub-images in the algorithm are merged into one first sub-image so that the LB is updated and the algorithm enters the second sub-image group acquisition module; otherwise, the algorithm enters the target sub-image acquisition module. a Let D be the a-th key cluster probability. 0 To preset the aggregation probability, D 1 a D a The corresponding second sub-image group, E a D 1 a The corresponding variance of grayscale values, E 0 The preset grayscale value variance; The target sub-image acquisition module is used to set a=a+1 and enter the first sub-image update module if a≠n, and if a=n, the first sub-image in LB is used as the target sub-image, and the target sub-image is marked with the color corresponding to the average occurrence probability of the target event corresponding to the target sub-image.

Images