Slope greening method

Through aerial image classification and soil data processing, combined with rock type identification network, an automated slope greening solution is generated, which solves the problems of low efficiency and lack of targetedness in the existing technology, and improves the effect and efficiency of slope greening and restoration.

CN118918391AActive Publication Date: 2024-11-08SHANXI METALLURGICAL GEOTECHNICAL ENG INVESTIGATION
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202411400345.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-09
Publication Date
2024-11-08
Estimated Expiration
2044-10-09

AI Technical Summary

Technical Problem

The existing slope greening methods are inefficient and lack targeted, resulting in poor greening effect and low restoration efficiency, waste of resources and high costs.

Method used

By obtaining aerial images for slope classification, obtaining soil data and performing data conversion, interpolation and grid division, a greening sub-scheme for different types of slopes is generated, and a slope transformation scheme is generated in combination with the rock type identification network to realize automatic generation of greening schemes. , improve targetedness and efficiency.

Benefits of technology

The automatic generation of slope greening solutions has been realized, efficiency and accuracy have been improved, and corresponding solutions have been proposed for different types of slopes, which have improved the greening effect and restoration efficiency, and reduced resource and labor costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118918391A_ABST
    Figure CN118918391A_ABST
Patent Text Reader

Abstract

The invention relates to the field of greening, and discloses a side slope greening method, which comprises the following steps: acquiring an aerial image of a side slope to be greened, and classifying the side slope according to the aerial image to obtain a side slope type of the side slope to be greened; if the slope type of the to-be-greened slope is a soil slope, acquiring a soil data set of the to-be-greened slope; interpolating the soil data set to obtain an updated soil data set; generating a greening sub-scheme corresponding to each to-be-greened sub-region according to the corresponding soil data subset; if the slope type of the slope to be greened is a rock slope, inputting the aerial image into a rock type identification network to obtain a rock type; according to the rock type and the gradient information of the to-be-greened slope, a slope surface reconstruction scheme for the to-be-greened slope is generated; and sending the slope surface reconstruction scheme to a greening personnel terminal, so that greening personnel reconstruct the slope surface of the to-be-greened slope according to the slope surface reconstruction scheme. Therefore, large-scale popularization and application of slope greening are realized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of greening, and in particular to a slope greening method. Background Art

[0002] The process of ore mining is often accompanied by the destruction of the original vegetation, which will lead to soil erosion and affect the ecological balance. Especially for the slopes, if the slopes are left exposed, landslides are prone to occur, which will cause safety risks. Therefore, while ore mining is in progress, it is necessary to green the exposed slopes.

[0003] In the existing slope greening process, the following technical problems often occur: First, in view of the complex actual situation, manual surveys are needed to understand different slope types; on this basis, greening personnel need to choose appropriate slope greening methods based on experience. This method relies on the experience of personnel and is difficult to achieve large-scale promotion and application, resulting in low efficiency of slope greening. Second, the existing slope greening methods lack specificity when dealing with different types of slopes, making it difficult to formulate targeted greening plans, resulting in uneven greening effects; Third, the lack of a targeted repair plan after slope greening not only affects the greening effect of the slope, but also reduces the efficiency of repair. If the repair effect is not good, secondary repair may be carried out, resulting in waste of resources and increased repair costs. Summary of the invention

[0004] The summary of the invention is used to introduce concepts in a brief form, which will be described in detail in the detailed implementation section below. The summary of the invention is not intended to identify the key features or essential features of the technical solution claimed for protection, nor is it intended to limit the scope of the technical solution claimed for protection.

[0005] The present invention proposes a slope greening method to solve one or more of the technical problems mentioned in the above background technology part.

[0006] The present invention provides a slope greening method, comprising: obtaining an aerial image of a slope to be greened, and classifying the slope according to the aerial image to obtain the slope type of the slope to be greened; if the slope type of the slope to be greened is a soil slope, obtaining a soil data set of the slope to be greened, wherein each soil data in the soil data set includes data acquisition point location information, soil looseness coefficient and soil nutrient content; converting the data acquisition point location information contained in each soil data in the soil data set to obtain image location coordinates corresponding to the aerial image, and mapping each soil data in the aerial image according to the image location coordinates to obtain a pixel point identifier corresponding to each soil data; interpolating the soil data set to obtain an updated soil data set, and the number of soil data in the updated soil data set is the same as that of the aerial image. The ratio of the areas of the images is greater than or equal to a preset ratio threshold; the aerial image is gridded according to the updated soil data set to obtain multiple aerial image areas, and the slope to be greened is divided into multiple sub-areas to be greened according to the multiple aerial image areas; according to the corresponding soil data subset, a greening sub-plan corresponding to each sub-area to be greened is generated; if the slope type of the slope to be greened is a rock slope, the aerial image is input into a rock type recognition network to obtain the rock type; according to the rock type and the slope information of the slope to be greened, a slope surface renovation plan for the slope to be greened is generated, and the renovation plan includes a slope greening substrate ratio and a substrate construction plan; the slope surface renovation plan is sent to the greening personnel terminal, so that the greening personnel can renovate the slope surface of the slope to be greened according to the slope surface renovation plan.

[0007] Optionally, slope classification is performed based on the aerial image to obtain the slope type to be greened, including: Perform feature extraction on the aerial image to obtain the actual image features of the aerial image; Acquire a pre-built slope image feature information library, wherein the slope image feature information in the slope image feature information library includes slope image features and corresponding slope types, and the slope image features include time features and image features; Determine the timestamp corresponding to the aerial image, encode the timestamp to obtain a coded timestamp, match the coded timestamp in a slope image feature information library to obtain multiple slope image feature information matching the coded timestamp, and the multiple slope image feature information constitute a candidate slope image feature information group; Matching the actual image features with the image features included in each slope image feature information in the candidate slope image feature information group to obtain a matching degree between the actual image features and the image features included in each slope image feature information; The slope image feature information with the highest matching degree is selected from the candidate slope image feature information group as the target slope image feature information, and the slope type corresponding to the target slope image feature information is determined as the slope type of the slope to be greened.

[0008] Optionally, the soil dataset is interpolated to obtain an updated soil dataset, including: According to the data collection time of each soil data in the soil data set, a corresponding weight is assigned to each soil data in the soil data set; For each soil data in the soil data set, according to the corresponding image position coordinates, determine the soil data closest to the soil data in the aerial image and use it as the adjacent soil data. The soil data and the adjacent soil data form a soil data group. Performing weighted summation on the two soil data contained in each soil data group according to corresponding weights to obtain inserted soil data, and adding the inserted soil data to the soil data set to obtain an updated soil data set; Determine whether the ratio of the number of soil data in the updated soil data set to the area of ​​the aerial image is greater than or equal to a preset ratio threshold; if it is greater than or equal to the preset ratio threshold, continue to execute subsequent processing steps; if it is less than the preset ratio threshold, continue to interpolate until the ratio of the number of soil data in the updated soil data set to the area of ​​the aerial image is greater than or equal to the preset ratio threshold.

[0009] Optionally, the aerial image is gridded according to the updated soil dataset to obtain multiple aerial image regions, including: The aerial image is divided according to a preset grid size to obtain a plurality of candidate image regions, wherein soil data corresponding to each pixel point included in each candidate image region constitutes a soil data subset; According to the corresponding soil data subset, the mean and variance of each candidate image area are calculated; for multiple candidate image areas in the aerial image, if the mean or variance of adjacent candidate image areas does not meet the first preset condition, the adjacent candidate image areas are merged into new candidate image areas, and after multiple iterations, until the mean or variance of the adjacent candidate image areas meets the first preset condition, multiple aerial image areas are obtained.

[0010] Optionally, a greening sub-plan corresponding to each sub-area to be greened is generated according to the corresponding soil data subset, including: According to the corresponding soil data subset, the soil characteristics corresponding to each sub-area to be greened are constructed; The soil characteristics corresponding to each sub-area to be greened are queried in a pre-constructed vegetation dictionary to obtain a plurality of candidate vegetation types corresponding to each sub-area to be greened, and the plurality of candidate vegetation types form a candidate vegetation type group; Obtain the purchase cost, cultivation cost, cultivation cycle, planting requirements and a pre-set set of constraints for each sub-area to be greened corresponding to each candidate vegetation type; According to the constraint condition group of each sub-area to be greened and the procurement cost, cultivation cost and cultivation period corresponding to each candidate vegetation type, the candidate vegetation type group corresponding to each sub-area to be greened is screened to obtain the target vegetation type corresponding to each sub-area to be greened; according to the target vegetation type and the planting requirements corresponding to the target vegetation type, the greening sub-plan corresponding to each sub-area to be greened is generated.

[0011] The present invention has the following beneficial effects: 1. Slope classification is performed by acquiring aerial images of the slopes to be greened. For soil slopes, soil data sets are acquired and data conversion, interpolation, grid division and other operations are performed. Greening sub-plans are generated according to the soil data subset corresponding to each sub-area. For rock slopes, aerial images are input into the rock type recognition network to obtain the rock type, and slope surface transformation plans are generated in combination with slope information and sent to the greening personnel terminal. The automatic generation of slope greening plans is realized, without relying on the experience of greening personnel for judgment and selection, which improves the efficiency and accuracy of slope greening. 2. Different processing methods are used for soil slopes and rock slopes. For soil slopes, a series of operations such as obtaining soil data sets, position information conversion and mapping, interpolation, and grid division are performed to generate greening sub-plans corresponding to each sub-area to be greened; for rock slopes, aerial images are input into the rock type recognition network, and slope surface transformation plans are generated in combination with slope information and sent to the greening personnel terminal. Corresponding solutions can be proposed for different types of slopes to be greened, thereby improving the problem that existing methods lack pertinence when dealing with different types of slopes and improving the overall greening effect; 3. In some embodiments, by adopting different repair schemes for the earthen slope greening sub-area and the rock slope greening sub-area in the slope greening area, the slope repair is made more targeted, which is beneficial to improving the efficiency of slope repair; by analyzing the density, distribution and growth status of vegetation, the dense growth area and replanting area in the earthen slope greening area and the rock to-be-replanted area in the rock slope greening area are accurately selected, which helps managers to quickly locate the areas that need to be repaired, reduce the time and labor costs of on-site investigations, improve the efficiency of slope greening, select suitable plants for replanting on the rock slope, and save resource costs and labor costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The above and other features, advantages and aspects of the embodiments of the present invention will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and that components and elements are not necessarily drawn to scale.

[0013] Figure 1 The present invention is a flow chart of a slope greening method. DETAILED DESCRIPTION

[0014] The present invention will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as being limited to the embodiments set forth herein. On the contrary, these embodiments are provided to provide a more thorough and complete understanding of the present invention. It should be understood that the drawings and embodiments of the present invention are only for exemplary purposes and are not intended to limit the scope of protection of the present invention.

[0015] It should also be noted that, for ease of description, only the parts related to the invention are shown in the drawings. In the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other.

[0016] It should be noted that the concepts such as "first" and "second" mentioned in the present invention are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0017] It should be noted that the modifications of "one" and "plurality" mentioned in the present invention are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0018] The names of the messages or information exchanged between multiple devices of the present invention are only for illustrative purposes, and are not used to limit the scope of these messages or information.

[0019] The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0020] like Figure 1 FIG. 1 is a flow chart of a slope greening method of the present invention, which specifically includes the following steps: Step 101: Acquire an aerial image of a slope to be greened, and classify the slope according to the aerial image to obtain the slope type of the slope to be greened.

[0021] In practice, the executor of a slope greening method is a cloud server for generating greening plans. The cloud server can communicate with various image acquisition devices (aerial drones). The aerial drone can collect images of the slope to be greened according to instructions, thereby obtaining aerial images of the slope to be greened and uploading them to the cloud server. Thus, the cloud server can classify the slopes according to the aerial images. Specifically, from the main composition of the slope, the slope can be divided into soil slopes and rock slopes. Different slope types require different greening plans. Therefore, it is necessary to classify the slopes through aerial images, and then obtain the slope type of the slope to be greened.

[0022] As an example, the slope type of the slope to be greened can be obtained through the following steps, including: Step 1: extract features from the aerial image to obtain the actual image features of the aerial image; specifically, the aerial image can be input into a convolutional neural network to obtain the actual image features of the aerial image. In addition, various algorithms such as color histogram can also be used to extract the actual image features.

[0023] Step 2, obtain a pre-constructed slope image feature information library, the slope image feature information in the slope image feature information library includes slope image features and corresponding slope types, and the slope image features include time features and image features. In practice, the slope image feature information library is constructed based on a large number of slope images collected in advance. Specifically, feature extraction can be performed on a large number of slope images respectively, and the image features obtained are combined with time features and corresponding slope types to obtain slope image feature information, thereby constructing a slope image feature information library. Among them, the time feature is constructed based on the shooting time of the slope image. Specifically, the month in the shooting time can be extracted and encoded to obtain the time feature.

[0024] Step three, determine the timestamp corresponding to the aerial image, encode the timestamp to obtain a coded timestamp, match the coded timestamp in the slope image feature information library, and obtain multiple slope image feature information matching the coded timestamp. The multiple slope image feature information constitute a candidate slope image feature information group.

[0025] Among them, a variety of encoding methods can be used to encode the timestamp, for example, a one-hot encoding method can be used to encode it to obtain an encoded timestamp.

[0026] Step 4: Match the actual image features with the image features included in each slope image feature information in the candidate slope image feature information group to obtain the matching degree between the actual image features and the image features included in each slope image feature information.

[0027] Step 5: Select the slope image feature information with the highest matching degree from the candidate slope image feature information group as the target slope image feature information, and determine the slope type corresponding to the target slope image feature information as the slope type of the slope to be greened. Slope types are divided into soil slopes and rock slopes.

[0028] Step 102: if the slope type of the slope to be greened is a soil slope, a soil data set of the slope to be greened is obtained, wherein each soil data in the soil data set includes location information of a data collection point, soil looseness coefficient, and soil nutrient content.

[0029] In some embodiments, for soil slopes, it is necessary to formulate corresponding greening plans according to different soil conditions. Therefore, it is necessary to obtain a soil data set for the slope to be greened. As needed, pre-collected historical soil data can be obtained from the database, or real-time soil data can be obtained by various sensors set on the slope to be greened. On this basis, historical soil data and real-time soil data together constitute a soil data set. Each soil data includes data collection point location information, soil loose coefficient and soil nutrient content. Optionally, each soil data also includes data collection time. Among them, the data collection point location information represents the location coordinates of the data collection point, and the soil loose coefficient represents the porosity of the soil. The soil loose coefficient can be calculated by measuring the soil specific gravity and the soil bulk density, or it can be obtained by querying local soil mapping data. The soil nutrient content can be the content of nutrients necessary for plant growth, including but not limited to: the content of nitrogen, phosphorus, potassium and other elements.

[0030] Step 103, converting the data collection point location information contained in each soil data in the soil data set to obtain image location coordinates corresponding to the aerial image, and mapping each soil data in the aerial image according to the image location coordinates to obtain the pixel point identification corresponding to each soil data.

[0031] Among them, the camera carried by the aerial photography drone can be calibrated in advance to obtain the camera calibration parameters, which include camera intrinsic parameters and camera extrinsic parameters. The camera extrinsic parameters include rotation matrix and translation matrix, and the camera intrinsic parameters include focal length and distortion coefficient. On this basis, the position information of the data acquisition point is converted using the camera intrinsic parameters and camera extrinsic parameters to obtain the image position coordinates corresponding to the aerial image. On this basis, each soil data can be mapped in the aerial image according to the image position coordinates, that is, determine which pixel point each soil data mapping corresponds to, so as to obtain the pixel point identification. In practice, each pixel point in the aerial image can be numbered. Therefore, the pixel point number can be used as the pixel point identification.

[0032] Step 104 , interpolating the soil data set to obtain an updated soil data set, wherein the ratio of the amount of soil data in the updated soil data set to the area of ​​the aerial image is greater than or equal to a preset ratio threshold.

[0033] In practice, whether it is historical soil data or real-time soil data, the number of corresponding data collection points is relatively small. Especially when the area of ​​the slope to be greened is large, it is difficult to reflect the soil conditions of the entire slope to be greened by relying solely on the soil dataset. Therefore, it is necessary to interpolate the soil dataset to obtain an updated soil dataset.

[0034] Specifically, various interpolation algorithms such as the inverse distance weighted average interpolation method (IDW) can be used to interpolate the soil data set, generate new soil data, and add the new soil data (i.e., insert soil data) to the soil data set to obtain an updated soil data set. Afterwards, it is determined whether the ratio of the number of soil data in the updated soil data set to the area of ​​the aerial image is greater than or equal to a preset ratio threshold; if it is less than the preset ratio threshold, the interpolation continues until the ratio of the number of soil data in the updated soil data set to the area of ​​the aerial image is greater than or equal to the preset ratio threshold.

[0035] Optionally, interpolating the soil dataset to obtain an updated soil dataset includes the following steps: Step 1: assign a corresponding weight to each soil data in the soil data set according to the data collection time of each soil data in the soil data set; specifically, the greater the difference between the data collection time and the current time, the smaller the assigned weight, and vice versa.

[0036] Step 2: For each soil data in the soil data set, determine the soil data closest to the soil data in the aerial image according to the corresponding image position coordinates and use it as the adjacent soil data. Each soil data and the adjacent soil data form a soil data group, thereby obtaining multiple soil data groups.

[0037] Step three, weighted sum the two soil data contained in each soil data group according to the corresponding weights to obtain the inserted soil data corresponding to each soil data group, add the inserted soil data to the soil data set, and obtain the updated soil data set. In practice, it is assumed that the two soil data contained in a soil data group are the first soil data and the second soil data, respectively, the second soil data is the adjacent soil data of the first soil data, the first soil data includes the first data collection point location information, the first soil loose coefficient and the first soil nutrient content, and the second soil data includes the second data collection point location information, the second soil loose coefficient and the second soil nutrient content. On this basis, when performing weighted summation, the first data collection point location information and the second data collection point location information are weighted to obtain the inserted data collection point location information, the first soil loose coefficient and the second soil loose coefficient are weighted to obtain the inserted soil loose coefficient, and the first soil nutrient content and the second soil nutrient content are weighted to obtain the inserted soil nutrient content. On this basis, the inserted data collection point location information, the inserted soil loose coefficient and the inserted soil nutrient content constitute the inserted soil data.

[0038] Furthermore, the inserted soil data may also be mapped into the aerial image to obtain pixel point identifiers corresponding to the inserted soil data.

[0039] Step 4, determine whether the ratio of the number of soil data in the updated soil data set to the area of ​​the aerial image is greater than or equal to the preset ratio threshold; if it is greater than or equal to the preset ratio threshold, continue to execute subsequent processing steps; if it is less than the preset ratio threshold, continue to interpolate until the ratio of the number of soil data in the updated soil data set to the area of ​​the aerial image is greater than or equal to the preset ratio threshold.

[0040] Step 105 , gridding the aerial image according to the updated soil data set to obtain a plurality of aerial image regions, and dividing the slope to be greened into a plurality of sub-regions to be greened according to the plurality of aerial image regions.

[0041] Specifically, first, the aerial image is divided according to a preset grid size to obtain multiple candidate image areas, wherein the soil data corresponding to each pixel point included in each candidate image area constitutes a soil data subset. In practice, each candidate image area corresponds to a grid. Afterwards, the mean and variance of each candidate image area are calculated according to the corresponding soil data subset. Then, for multiple candidate image areas in the aerial image, if the mean or variance of adjacent candidate image areas does not meet the first preset condition, the adjacent candidate image areas are merged into new candidate image areas, and after multiple iterations, until the mean or variance of the adjacent candidate image areas meets the first preset condition, one or more candidate image areas after the iteration are completed constitute multiple aerial image areas. Among them, the first preset condition is that the difference between the means of adjacent candidate image areas is greater than the first preset threshold and the difference between the variances of adjacent candidate image areas is greater than the second preset threshold.

[0042] On this basis, through the mapping relationship between the aerial image and the slope to be greened, the slope to be greened can be divided into multiple sub-areas to be greened according to multiple aerial image areas, and each aerial image area corresponds to a sub-area to be greened.

[0043] Step 106: Generate a greening sub-plan corresponding to each sub-area to be greened according to the corresponding soil data subset.

[0044] In practice, the corresponding greening sub-scheme can be generated according to the soil data subset corresponding to each sub-region to be greened. Specifically, the soil characteristics can be first constructed according to the mean of the soil data subset corresponding to each sub-region to be greened. For example, since the soil data includes the soil loose coefficient and the soil nutrient content, the mean of the soil loose coefficient can be encoded as a dimension of the soil characteristics, and the mean of the content of each nutrient element in the soil nutrient content can also be used as a dimension of the soil characteristics, thereby forming the soil characteristics. On this basis, the soil characteristics are queried in the pre-constructed vegetation dictionary to obtain multiple candidate vegetation types corresponding to each sub-region to be greened. In practice, a large number of vegetation types and soil characteristics corresponding to each vegetation type are stored in the vegetation dictionary. Therefore, the soil characteristics can be queried in the vegetation dictionary to query multiple soil characteristics whose similarity with the soil characteristics of each sub-region to be greened is less than the preset similarity threshold, and the vegetation types corresponding to the multiple soil characteristics queried are determined as multiple candidate vegetation types corresponding to each sub-region to be greened. Afterwards, the purchase cost, cultivation cost, cultivation cycle, planting requirements and a pre-set set of constraints for each sub-area to be greened corresponding to each candidate vegetation type are obtained; the purchase cost is the purchase cost of a single vegetation.

[0045] According to the constraint condition group of each sub-area to be greened and the procurement cost, cultivation cost and cultivation period corresponding to each candidate vegetation type, the candidate vegetation type group corresponding to each sub-area to be greened is screened to obtain the target vegetation type corresponding to each sub-area to be greened. Then, according to the target vegetation type and the planting requirements corresponding to the target vegetation type, the greening sub-plan corresponding to each sub-area to be greened is generated.

[0046] For example, the constraint group includes procurement cost constraints, cultivation cost constraints, cultivation cycle constraints, etc. The procurement cost constraint is, for example, that the overall procurement cost is less than 1 million. On this basis, for each candidate vegetation type among the multiple candidate vegetation types corresponding to each sub-area to be greened, the standard planting quantity is calculated according to the standard planting density of the candidate vegetation type and the area of ​​the sub-area to be greened, and the predicted procurement cost of each candidate vegetation type is calculated according to the standard planting quantity and the corresponding procurement cost, and it is determined whether the predicted procurement cost of each candidate vegetation type meets the procurement cost constraint of the sub-area to be greened. If the predicted procurement cost of a candidate vegetation type does not meet the procurement cost constraint of the sub-area to be greened, the candidate vegetation type is removed from the candidate vegetation type group. On this basis, the candidate vegetation types in the candidate vegetation type group can continue to be screened by the cultivation cost constraint and the cultivation cycle constraint, and the candidate vegetation types that do not meet the cultivation cost constraint and the cultivation cycle constraint are continued to be removed, and the remaining candidate vegetation types that meet each constraint in the constraint group are the target vegetation types.

[0047] On this basis, the target vegetation type and the planting requirements corresponding to each sub-area to be greened are determined as the greening sub-scheme corresponding to each sub-area to be greened. The greening sub-schemes corresponding to each sub-area to be greened constitute the greening scheme of the slope to be greened.

[0048] In these embodiments, by obtaining an aerial image of the slope to be greened and determining the slope type of the slope to be greened, the shortcomings of manual survey and reliance on personal experience in the traditional way are avoided. For soil slopes, by processing the soil data set and mapping the soil data set to the aerial image, further interpolation and gridding are performed, and finally a specific greening sub-scheme is generated based on the corresponding soil data subset. The degree of manual participation is reduced and the processing efficiency is improved.

[0049] Step 107: If the slope type of the slope to be greened is a rock slope, the aerial image is input into a rock type recognition network to obtain the rock type.

[0050] In practice, due to the different types of rocks, they need to be transformed in different ways to meet the needs of greening. Therefore, the aerial image can be input into the rock type recognition network to obtain the rock type. Among them, the network structure of the rock type recognition network can be composed of a convolutional neural network and a prediction network, and the prediction network includes a softmax layer. According to needs, the training sample set can be used for pre-training to obtain a rock type recognition network that meets a certain prediction accuracy. The training samples in the training sample set include sample rock images and corresponding rock types.

[0051] Step 108, generating a slope surface reconstruction plan for the slope to be greened according to the rock type and the slope information of the slope to be greened, the reconstruction plan including the slope greening base material ratio and the base material construction plan.

[0052] In some embodiments, the ratio of slope greening substrate includes: the ratio of cement, soil, humus, fertilizer, water retaining agent, concrete additives and mixed greening seeds. The substrate construction plan includes: substrate spray thickness, substrate spray times, composite mesh laying parameters. In practice, the rock type and slope information can be input into the scheme generation model to obtain the slope surface transformation plan. Among them, the scheme generation model includes an input layer, a first feature fusion network, a second feature fusion network, a substrate ratio generation sub-model and a substrate construction scheme generation sub-model, wherein the input layer takes the rock type and slope information of the slope to be greened as input and outputs the initial features.

[0053] The substrate ratio generation submodel is a BP neural network with a preset number of hidden layers. The substrate ratio generation submodel is used to generate the slope greening substrate ratio. The substrate construction plan generation submodel is an elastic network ElasticNet, and the substrate construction plan is used to generate the substrate construction plan. If needed, the BP neural network and the elastic network ElasticNet can be jointly trained.

[0054] Step 109, sending the slope surface reconstruction plan to the greening personnel terminal, so that the greening personnel can reconstruct the slope surface of the greening slope according to the slope surface reconstruction plan.

[0055] In some embodiments, the slopes are classified by acquiring aerial images of the slopes to be greened. For soil slopes, soil data sets are acquired and data conversion, interpolation, grid division and other operations are performed, and greening sub-plans are generated according to the soil data subset corresponding to each sub-area; for rock slopes, aerial images are input into the rock type recognition network to obtain the rock type, and the slope surface transformation plan is generated in combination with the slope information and sent to the greening personnel terminal. The automatic generation of slope greening plans is realized, without relying on the experience of greening personnel for judgment and selection, and the efficiency and accuracy of slope greening are improved.

[0056] In some embodiments, in order to further solve the second technical problem described in the background technology section, that is, "the existing slope greening method lacks pertinence when dealing with different types of slopes, and it is difficult to formulate a highly targeted greening plan, resulting in uneven greening effects", in some embodiments of the present invention, slope classification is performed according to aerial images to obtain the slope type of the slope to be greened, including: Perform feature extraction on the aerial image to obtain the actual image features of the aerial image; Acquire a pre-built slope image feature information library, wherein the slope image feature information in the slope image feature information library includes slope image features and corresponding slope types, and the slope image features include time features and image features; Determine the timestamp corresponding to the aerial image, encode the timestamp to obtain a coded timestamp, match the coded timestamp in a slope image feature information library to obtain multiple slope image feature information matching the coded timestamp, and the multiple slope image feature information constitute a candidate slope image feature information group; Matching the actual image features with the image features included in each slope image feature information in the candidate slope image feature information group to obtain a matching degree between the actual image features and the image features included in each slope image feature information; The slope image feature information with the highest matching degree is selected from the candidate slope image feature information group as the target slope image feature information, and the slope type corresponding to the target slope image feature information is determined as the slope type of the slope to be greened.

[0057] Optionally, the soil dataset is interpolated to obtain an updated soil dataset, including: According to the data collection time of each soil data in the soil data set, a corresponding weight is assigned to each soil data in the soil data set; For each soil data in the soil data set, according to the corresponding image position coordinates, determine the soil data closest to the soil data in the aerial image and use it as the adjacent soil data. The soil data and the adjacent soil data form a soil data group. Performing weighted summation on the two soil data contained in each soil data group according to corresponding weights to obtain inserted soil data, and adding the inserted soil data to the soil data set to obtain an updated soil data set; Determine whether the ratio of the number of soil data in the updated soil data set to the area of ​​the aerial image is greater than or equal to a preset ratio threshold; if it is greater than or equal to the preset ratio threshold, continue to execute subsequent processing steps; if it is less than the preset ratio threshold, continue to interpolate until the ratio of the number of soil data in the updated soil data set to the area of ​​the aerial image is greater than or equal to the preset ratio threshold.

[0058] Optionally, the aerial image is gridded according to the updated soil dataset to obtain multiple aerial image regions, including: The aerial image is divided according to a preset grid size to obtain a plurality of candidate image regions, wherein soil data corresponding to each pixel point included in each candidate image region constitutes a soil data subset; According to the corresponding soil data subset, the mean and variance of each candidate image area are calculated; for multiple candidate image areas in the aerial image, if the mean or variance of adjacent candidate image areas does not meet the first preset condition, the adjacent candidate image areas are merged into new candidate image areas, and after multiple iterations, until the mean or variance of the adjacent candidate image areas meets the first preset condition, multiple aerial image areas are obtained.

[0059] Optionally, a greening sub-plan corresponding to each sub-area to be greened is generated according to the corresponding soil data subset, including: According to the corresponding soil data subset, the soil characteristics corresponding to each sub-area to be greened are constructed; The soil characteristics corresponding to each sub-area to be greened are queried in a pre-constructed vegetation dictionary to obtain a plurality of candidate vegetation types corresponding to each sub-area to be greened, and the plurality of candidate vegetation types form a candidate vegetation type group; Obtain the purchase cost, cultivation cost, cultivation cycle, planting requirements and a pre-set set of constraints for each sub-area to be greened corresponding to each candidate vegetation type; According to the constraint condition group of each sub-area to be greened and the procurement cost, cultivation cost and cultivation period corresponding to each candidate vegetation type, the candidate vegetation type group corresponding to each sub-area to be greened is screened to obtain the target vegetation type corresponding to each sub-area to be greened; according to the target vegetation type and the planting requirements corresponding to the target vegetation type, the greening sub-plan corresponding to each sub-area to be greened is generated.

[0060] In some embodiments, different processing methods are used for soil slopes and rock slopes. For soil slopes, a series of operations such as obtaining soil data sets, converting and mapping position information, interpolation, and grid division are performed to generate greening sub-plans corresponding to each sub-area to be greened; for rock slopes, aerial images are input into a rock type recognition network, and slope surface transformation plans are generated in combination with slope information and sent to the greening personnel terminal. Corresponding solutions can be proposed for different types of slopes to be greened, thereby improving the problem that existing methods lack specificity in dealing with different types of slopes and improving the overall greening effect.

[0061] In some embodiments, in order to further solve the technical problem 3 described in the background technology part, that is, "the lack of a targeted repair plan after slope greening not only affects the greening effect of the slope, but also reduces the repair efficiency. If the repair effect is not good, secondary repair may be carried out, resulting in a waste of resources and an increase in the repair cost", some embodiments of the present invention also include: Step 1: Divide the slope greening area into a soil slope area and a rock slope area according to the slope type, and divide the soil slope area and the rock slope area into a plurality of soil slope greening sub-areas and a plurality of rock slope greening sub-areas according to preset rules; Step 2: Obtain a first vegetation aerial image group corresponding to multiple soil slope greening sub-areas and obtain a second vegetation aerial image group corresponding to multiple rock slope greening sub-areas; each vegetation aerial image in the first vegetation aerial image group corresponds to a soil slope greening sub-area in multiple soil slope greening sub-areas; each vegetation aerial image in the second vegetation aerial image group corresponds to a rock slope greening sub-area in multiple rock slope greening sub-areas; Step 3: input the first vegetation aerial image group into a pre-trained deep learning network to obtain first vegetation image feature information corresponding to each vegetation aerial image in the first vegetation aerial image group, where the first vegetation image feature information includes vegetation color morphology features and vegetation density features; Step 4. For the first vegetation image feature information corresponding to each vegetation aerial image in the first vegetation aerial image group, obtain a pre-constructed vegetation feature information database, where the vegetation feature information database includes vegetation health levels corresponding to vegetation color and morphological features and vegetation coverage rates corresponding to vegetation density features; match the first vegetation image feature information corresponding to each vegetation aerial image in the first vegetation aerial image group in the vegetation feature information database to obtain vegetation health levels and vegetation coverage rates corresponding to each vegetation aerial image in the first vegetation aerial image group; wherein the vegetation health levels are divided into primary health and secondary health; match the vegetation health levels in the first vegetation aerial image group to the vegetation health levels and vegetation coverage rates. The sub-region after greening of the soil slope corresponding to the aerial image of vegetation with a health level of the first level is determined as the first level healthy region, and the first level healthy region group is obtained; the first level healthy region in the first level healthy region group whose vegetation coverage is greater than or equal to the preset coverage threshold is determined as the dense growth region, and the pruning information corresponding to the dense growth region is generated; the aerial image group of vegetation is screened according to the vegetation coverage corresponding to the aerial image of vegetation corresponding to the second level healthy vegetation health level, so as to select the sub-region after greening of the soil slope corresponding to the aerial image of vegetation with a vegetation coverage less than the preset vegetation coverage threshold as the replanting region, and generate the replanting information corresponding to the replanting region; Step 5: Input the second vegetation aerial image group into the pre-trained vegetation prediction model to obtain the vegetation growth prediction type corresponding to each vegetation aerial image in the second vegetation aerial image group; if the vegetation growth prediction type is an obstructed growth type, determine the rock slope greening sub-area corresponding to the vegetation aerial image corresponding to the obstructed growth type as an obstructed growth area to obtain an obstructed growth area group; obtain the safety level corresponding to each obstructed growth area in the obstructed growth area group, determine the obstructed growth area corresponding to the safety level greater than the preset safety level in the obstructed growth area group as the rock to-be-replanted area, obtain the candidate plant group corresponding to the rock to-be-replanted area, and each candidate plant in the candidate plant group has a corresponding survival rate and landscape degree; obtain the pre-configured plant evaluation system. The plant scoring table includes multiple survival rates, scores corresponding to each survival rate, multiple landscape levels, and scores corresponding to each landscape level; by querying the survival rate and landscape level corresponding to each candidate plant in the plant scoring table, the survival rate score and landscape level score corresponding to each candidate plant are obtained; weights are respectively assigned to the survival rate score and landscape level score corresponding to the candidate plants, and the survival rate score and landscape level score corresponding to each candidate plant are weighted and summed up by the weights to obtain the total score corresponding to each candidate plant; according to the total score corresponding to each candidate plant, the candidate plant with the highest total score is selected from the candidate plant group as the standard replanting plant for the rock replanting area, and the replanting information of the rock replanting area is generated.

[0062] Step six, the pruning information corresponding to the dense growth area, the replanting information corresponding to the replanting area, and the replanting information of the rock to-be-replanted area are sent to the slope management terminal, so that the staff corresponding to the slope management terminal can perform pruning and replanting operations.

[0063] In some embodiments, the soil slope area and the rock slope area are divided into a plurality of soil slope greening sub-areas and a plurality of rock slope greening sub-areas according to preset rules. Taking the soil slope area as an example, the soil slope area is divided into a plurality of soil slope greening sub-areas of the same area according to the preset rules. The aerial photography drone can collect images of the plurality of soil slope greening sub-areas and the plurality of rock slope greening sub-areas, thereby obtaining a first vegetation aerial image group and a second vegetation aerial image group. Each vegetation aerial image in the first vegetation aerial image group corresponds to a soil slope greening sub-area in the plurality of soil slope greening sub-areas. Specifically, the first vegetation aerial image group may include vegetation aerial image 1 and vegetation aerial image 2, wherein vegetation aerial image 1 corresponds to soil slope greening sub-area 1 in the plurality of soil slope greening sub-areas, and vegetation aerial image 1 is the actual image of soil slope greening sub-area 1. The second vegetation aerial image corresponds to the second soil slope greening sub-area among the multiple soil slope greening sub-areas, and the second vegetation aerial image is the actual image of the second soil slope greening sub-area.

[0064] In some embodiments, the first vegetation aerial image group is input into a pre-trained deep learning network, the structure of which is a convolutional neural network, which has been pre-trained with a large number of vegetation data sets, and is used to obtain the first vegetation image feature information corresponding to each vegetation aerial image, and the first vegetation image feature information includes vegetation color and morphological features and vegetation density features. A pre-constructed vegetation feature information library is obtained, and the vegetation feature information library includes vegetation health levels corresponding to vegetation color and morphological features and vegetation coverage rates corresponding to vegetation density features, wherein the vegetation health levels are divided into primary health and secondary health. Specifically, when the vegetation color and morphological features are dark green in color and upright in shape, the corresponding vegetation health level is primary health, and when the vegetation color and morphological features are yellow-green in color and relatively upright in shape, the corresponding vegetation health level is secondary health. The primary health area in the primary health area group whose vegetation coverage rate is greater than or equal to the preset coverage rate threshold is determined as a dense growth area, and pruning information corresponding to the dense growth area is generated. Specifically, the first-level healthy area group corresponds to areas with good vegetation growth. The dense growth area is an area with high vegetation coverage. Pruning the first-level healthy area with a vegetation coverage rate greater than or equal to the preset coverage rate threshold is more conducive to vegetation growth. When the soil slope greening sub-area one is a dense growth area, the pruning information corresponding to the dense growth area may be that the soil slope greening sub-area one needs to be properly pruned. For the soil slope greening sub-area corresponding to the vegetation aerial image corresponding to the vegetation health level of the second level health, the health level is low. The soil slope greening sub-area corresponding to the vegetation aerial image with a vegetation coverage rate less than the preset vegetation coverage rate threshold is selected for replanting, and the replanting information corresponding to the replanting area is generated. When the soil slope greening sub-area two is a replanting area, the replanting information corresponding to the replanting area may be that the soil slope greening sub-area two needs to be replanted.

[0065] In some embodiments, the second vegetation aerial image group is input into a pre-trained vegetation prediction model. The structure of the vegetation prediction model is a convolutional neural network, which is pre-trained with a large number of vegetation data sets and is used to predict vegetation growth prediction types. The vegetation growth prediction types can be obstructed growth types and normal growth types. Among them, plants of obstructed growth types grow poorly and often find it difficult to maintain a normal growth state, so it is necessary to replant the obstructed growth type areas. The safety level corresponding to each obstructed growth area in the obstructed growth area group is obtained from the rock slope management library. The safety levels are divided into level one, level two, and level three. The higher the safety level, the higher the safety of the management personnel when working in the rock area. The rock area with a lower safety level has a high risk of operation for the management personnel, and it is not recommended for the management personnel to perform operations. Assuming that the preset safety level is level 2, the obstructed growth areas corresponding to level 1 and level 2 are determined as rock replanting areas, and the candidate plant group corresponding to the rock replanting area is obtained from the vegetation database. The candidate plant group may include candidate plant 1 and candidate plant 2. Each candidate plant in the candidate plant group has a corresponding survival rate and landscape degree. The landscape degree represents the ornamental value of the plant. The higher the landscape degree, the stronger the ornamental value. Specifically, the survival rate corresponding to candidate plant 1 is 90%, and the landscape degree is level 2. The survival rate corresponding to candidate plant 2 is 85%, and the landscape degree is level 1. A pre-configured plant scoring table is obtained from the plant database. The plant scoring table includes multiple survival rates, scores corresponding to each survival rate, multiple landscape degrees, and scores corresponding to each landscape degree. Specifically, the survival rate of 90% can be scored as 90 points, and the landscape degree of level 2 can be scored as 80 points; the survival rate of 85% can be scored as 85 points, and the landscape degree of level 1 can be scored as 95 points. A weight of 70% is assigned to the survival rate score corresponding to the candidate plants, and a weight of 30% is assigned to the landscape level score corresponding to the candidate plants. The total score corresponding to the candidate plant 1 is 87 points, and the total score corresponding to the candidate plant 2 is 88 points. The total score corresponding to the candidate plant 2 is the highest in the candidate plant group. The candidate plant 2 is used as the standard replanting plant for the rock replanting area. The replanting information for the rock replanting area can be that the sub-area 1 needs to be replanted after the rock slope is greened, and the replanting plant is the candidate plant 2. The pruning information corresponding to the dense growth area, the replanting information corresponding to the replanting area, and the replanting information for the rock replanting area are sent to the slope management terminal, so that the staff corresponding to the slope management terminal can perform pruning and replanting operations.

[0066] In some embodiments, in some embodiments, by adopting different repair schemes for the earth slope greening sub-area and the rock slope greening sub-area in the slope greening area, the slope repair is made more targeted, which is beneficial to improving the efficiency of slope repair; by analyzing the density, distribution and growth status of vegetation, the dense growth area and replanting area in the earth slope greening area and the rock to-be-replanted area in the rock slope greening area are accurately selected, which helps managers to quickly locate areas that need to be repaired, reduce the time and labor costs of on-site investigations, improve the efficiency of slope greening, select suitable plants for replanting on the rock slope, and save resource costs and labor costs.

[0067] The above descriptions are only some preferred embodiments of the present invention and the explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the present invention is not limited to the technical solutions formed by the specific combination of the above technical features, but also should cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present invention (but not limited to) to form a technical solution.

Claims

1. A slope greening method, characterized in that: include: Acquire an aerial image of the slope to be greened, and classify the slope according to the aerial image to obtain the slope type of the slope to be greened; If the slope type of the slope to be greened is a soil slope, a soil data set of the slope to be greened is obtained, wherein each soil data in the soil data set includes location information of a data collection point, soil looseness coefficient, and soil nutrient content; Converting the data collection point position information contained in each soil data in the soil data set to obtain image position coordinates corresponding to the aerial image, and mapping each soil data in the aerial image according to the image position coordinates to obtain a pixel point identifier corresponding to each soil data; interpolating the soil data set to obtain an updated soil data set, wherein a ratio of the amount of soil data in the updated soil data set to the area of ​​the aerial image is greater than or equal to a preset ratio threshold; Gridding the aerial image according to the updated soil data set to obtain a plurality of aerial image regions, and dividing the slope to be greened into a plurality of sub-regions to be greened according to the plurality of aerial image regions; Generate a greening sub-plan corresponding to each sub-area to be greened according to the corresponding soil data subset; If the slope type of the slope to be greened is a rock slope, input the aerial image into a rock type recognition network to obtain the rock type; According to the rock type and the slope information of the slope to be greened, a slope surface transformation plan for the slope to be greened is generated, wherein the transformation plan includes a slope greening substrate ratio and a substrate construction plan; The slope surface modification plan is sent to a terminal of a greening staff, so that the greening staff can modify the slope surface of the slope to be greened according to the slope surface modification plan.

2. The slope greening method according to claim 1, characterized in that: The step of classifying the slope according to the aerial image to obtain the type of the slope to be greened includes: Performing feature extraction on the aerial image to obtain actual image features of the aerial image; Acquire a pre-built slope image feature information library, wherein the slope image feature information in the slope image feature information library includes slope image features and corresponding slope types, and the slope image features include time features and image features; Determine a timestamp corresponding to the aerial image, encode the timestamp to obtain a coded timestamp, match the coded timestamp in the slope image feature information library to obtain a plurality of slope image feature information matching the coded timestamp, wherein the plurality of slope image feature information constitute a candidate slope image feature information group; Matching the actual image feature with the image feature included in each slope image feature information in the candidate slope image feature information group to obtain a matching degree between the actual image feature and the image feature included in each slope image feature information; The slope image feature information with the highest matching degree is selected from the candidate slope image feature information group as the target slope image feature information, and the slope type corresponding to the target slope image feature information is determined as the slope type of the slope to be greened.

3. The slope greening method according to claim 2, characterized in that: The interpolating the soil dataset to obtain an updated soil dataset includes: assigning a corresponding weight to each soil data in the soil data set according to a data collection time of each soil data in the soil data set; For each soil data in the soil data set, according to the corresponding image position coordinates, determine the soil data in the aerial image that is closest to the soil data and use it as the adjacent soil data, and the soil data and the adjacent soil data form a soil data group; Performing weighted summation on two soil data included in each soil data group according to corresponding weights to obtain inserted soil data, and adding the inserted soil data to the soil data set to obtain an updated soil data set; Determine whether the ratio of the number of soil data in the updated soil data set to the area of ​​the aerial image is greater than or equal to a preset ratio threshold; if it is greater than or equal to the preset ratio threshold, continue to execute subsequent processing steps; if it is less than the preset ratio threshold, continue to interpolate until the ratio of the number of soil data in the updated soil data set to the area of ​​the aerial image is greater than or equal to the preset ratio threshold.

4. The slope greening method according to claim 3, characterized in that: The step of gridding the aerial image according to the updated soil dataset to obtain a plurality of aerial image regions includes: The aerial image is divided according to a preset grid size to obtain a plurality of candidate image regions, wherein soil data corresponding to each pixel point included in each candidate image region constitutes a soil data subset; According to the corresponding soil data subset, the mean and variance of each candidate image area are calculated; for the multiple candidate image areas in the aerial image, if the mean or variance of adjacent candidate image areas does not meet the first preset condition, the adjacent candidate image areas are merged into new candidate image areas, and after multiple iterations, until the mean or variance of the adjacent candidate image areas meets the first preset condition, the multiple aerial image areas are obtained.

5. The slope greening method according to claim 4, characterized in that: The step of generating a greening sub-plan corresponding to each sub-area to be greened according to the corresponding soil data subset includes: According to the corresponding soil data subset, the soil characteristics corresponding to each sub-area to be greened are constructed; The soil characteristics corresponding to each sub-area to be greened are queried in a pre-constructed vegetation dictionary to obtain a plurality of candidate vegetation types corresponding to each sub-area to be greened, and the plurality of candidate vegetation types form a candidate vegetation type group; Obtain the purchase cost, cultivation cost, cultivation cycle, planting requirements and a pre-set set of constraints for each sub-area to be greened corresponding to each candidate vegetation type; According to the constraint condition group of each sub-area to be greened and the procurement cost, cultivation cost and cultivation period corresponding to each candidate vegetation type, the candidate vegetation type group corresponding to each sub-area to be greened is screened to obtain the target vegetation type corresponding to each sub-area to be greened; according to the target vegetation type and the planting requirements corresponding to the target vegetation type, the greening sub-plan corresponding to each sub-area to be greened is generated.

Citation Information

Patent Citations

  • Road slope vegetation recovery growth and construction method

    CN115936189A

  • Monitoring and maintenance management method for intelligent community greening species

    CN117372868A

  • Methods of greening management in smart cities, system, and storage mediums thereof

    US20230004903A1