CN118135444A - Land planning resource monitoring method and system based on unmanned aerial vehicle remote sensing - Google Patents
Land planning resource monitoring method and system based on unmanned aerial vehicle remote sensing Download PDFInfo
- Publication number
- CN118135444A CN118135444A CN202410443219.2A CN202410443219A CN118135444A CN 118135444 A CN118135444 A CN 118135444A CN 202410443219 A CN202410443219 A CN 202410443219A CN 118135444 A CN118135444 A CN 118135444A
- Authority
- CN
- China
- Prior art keywords
- connected domain
- land
- remote sensing
- pixel
- clusters
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000012544 monitoring process Methods 0.000 title claims abstract description 45
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000004364 calculation method Methods 0.000 claims description 13
- 230000006870 function Effects 0.000 claims description 12
- 238000013528 artificial neural network Methods 0.000 claims description 11
- 238000013527 convolutional neural network Methods 0.000 claims description 10
- 239000003550 marker Substances 0.000 claims description 9
- 238000004422 calculation algorithm Methods 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims description 6
- 238000003064 k means clustering Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 claims description 2
- 238000010845 search algorithm Methods 0.000 claims description 2
- 238000012549 training Methods 0.000 claims description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of image recognition, in particular to a land planning resource monitoring method and system based on unmanned aerial vehicle remote sensing, comprising the following steps: according to the gray level difference and the distance between each pixel point in each connected domain and other pixel points and the gray level difference and the distance between each pixel point in each connected domain and the pixel points in the connected domain adjacent to each connected domain, a plurality of optimal initial clustering center points are selected from all the pixel points; obtaining a plurality of class clusters in a land remote sensing image, merging part of adjacent class clusters according to the difference between gray level distribution of all pixel points in the two class clusters, and marking the class clusters after merging as marked class clusters; and monitoring and planning the land resources according to a plurality of mark clusters in the land remote sensing image. The invention reduces the iteration times of clustering, improves the clustering efficiency and also improves the accuracy of land resource planning.
Description
Technical Field
The invention relates to the technical field of image recognition, in particular to a land planning resource monitoring method and system based on unmanned aerial vehicle remote sensing.
Background
The land planning resource monitoring refers to a process of monitoring and evaluating land utilization conditions and resource conditions by utilizing various technical means so as to support works such as land planning, resource management, environmental protection and the like; the land planning resource monitoring can provide scientific basis and decision support for decision makers, and promote reasonable utilization and sustainable development of land resources. The method is used for planning resources and monitoring the land by acquiring remote sensing images of the land through unmanned aerial vehicle technology.
When land resource planning and monitoring are carried out through the land remote sensing images, different types of land partition can be carried out on the land remote sensing images through a K-means clustering algorithm, and land planning is carried out through land monitoring of different types; however, when different types of land division are performed on the land remote sensing image through the K-means clustering algorithm, as initial clustering center points are randomly selected, some initial clustering center points need more times to obtain an optimal clustering result, so that the operation amount of a computer is increased, and the time required for calculation is increased; and the initial clustering center points are sensitive to the clustering structure, and the clustering results corresponding to different initial clustering center points have great difference, so that the selection of the initial clustering center points affects the final clustering result, and the accuracy of land resource planning is reduced.
Disclosure of Invention
The invention provides a land planning resource monitoring method and system based on unmanned aerial vehicle remote sensing, which are used for solving the existing problems.
The land planning resource monitoring method and system based on unmanned aerial vehicle remote sensing adopt the following technical scheme:
The embodiment of the invention provides a land planning resource monitoring method based on unmanned aerial vehicle remote sensing, which comprises the following steps:
Acquiring a land remote sensing image;
Detecting connected domains of the land remote sensing image to obtain a plurality of connected domains, and screening a plurality of preferred initial clustering center points from all the pixel points according to gray level differences and distances between each pixel point in each connected domain and other pixel points in each connected domain and gray level differences and distances between each pixel point in each connected domain and pixel points in the connected domain adjacent to each connected domain;
Clustering all pixel points in the land remote sensing image according to a plurality of preferred initial clustering center points to obtain a plurality of class clusters, merging part of adjacent class clusters according to the difference between gray level distribution of all pixel points in the two class clusters, and marking the class clusters after merging as marked class clusters;
And obtaining the area occupation ratio of each type of land according to a plurality of mark clusters in the land remote sensing image, and carrying out land resource monitoring and planning according to the area occupation ratio of each type of land.
Further, the step of selecting a plurality of preferred initial cluster center points from all the pixel points according to the gray scale difference and the distance between each pixel point in each connected domain and other pixel points, and the gray scale difference and the distance between each pixel point in each connected domain and the pixel points in the connected domain adjacent to each connected domain, includes the following specific steps:
Obtaining the possibility that each pixel point in each connected domain is an initial clustering center point according to the gray difference between each pixel point in each connected domain and all other pixel points, the distance between each pixel point in each connected domain and the center point of the connected domain and the gray difference between each pixel point in each connected domain and all pixel points in adjacent connected domains;
When the possibility that each pixel point in each connected domain is an initial clustering center point is greater than or equal to a preset first threshold epsilon, marking the pixel point as a candidate initial clustering center point;
According to the distance between each candidate initial clustering central point in each connected domain and all candidate initial clustering central points in the connected domain adjacent to each connected domain and the possibility that each pixel point in each connected domain is an initial clustering central point, the possibility that each candidate initial clustering central point in each connected domain is used as a preferable initial clustering central point is obtained;
and selecting all candidate initial cluster center points in each connected domain as one candidate initial cluster center point with the highest possibility of the preferred initial cluster center point, and recording the candidate initial cluster center point as the preferred initial cluster center point.
Further, the method for obtaining the possibility that each pixel point in each connected domain is an initial clustering center point according to the gray level difference between each pixel point in each connected domain and all other pixel points, the distance between each pixel point in each connected domain and the center point of the connected domain, and the gray level difference between each pixel point in each connected domain and all pixel points in adjacent connected domains, includes the following specific calculation method:
Where H i,j denotes a gray value of a j-th pixel in the i-th connected domain, H i,r denotes a gray value of a r-th pixel in the i-th connected domain, n denotes the number of all pixels in each connected domain, d i,j denotes a distance between the j-th pixel in the i-th connected domain and a center point of the connected domain, H i,e denotes a mean value of gray values of all pixels in an e-th adjacent connected domain of the i-th connected domain, m denotes the number of all adjacent connected domains of each connected domain, i is an absolute value symbol, norm () denotes a linear normalization function, and Q i,j denotes a possibility that the j-th pixel in the i-th connected domain is an initial cluster center point.
Further, the method for obtaining the probability that each candidate initial cluster center point in each connected domain is used as the preferred initial cluster center point according to the distance between each candidate initial cluster center point in each connected domain and all candidate initial cluster center points in the connected domain adjacent to each connected domain and the probability that each pixel point in each connected domain is the initial cluster center point comprises the following specific calculation methods:
where d i,c,e,v represents the distance between the c-th candidate initial cluster center point in the i-th connected domain and the v-th candidate initial cluster center point in the e-th adjacent connected domain, M represents the number of all candidate initial cluster center points in any one of the connected domains adjacent to each connected domain, N represents the number of all connected domains adjacent to each connected domain, Q i,j represents the likelihood that the j-th candidate initial cluster center point in the i-th connected domain is the initial cluster center point, Q i,min represents the minimum value of the likelihood that all initial cluster center points in the i-th connected domain are the initial cluster center points, Q i,max represents the maximum value of the likelihood that all initial cluster center points in the i-th connected domain are the initial cluster center points, and W i,c represents the likelihood that the c-th candidate initial cluster center point in the i-th connected domain is the preferred initial cluster center point.
Further, the clustering is performed on all pixel points in the land remote sensing image according to a plurality of preferred initial clustering center points to obtain a plurality of clusters, and the method comprises the following specific steps:
And clustering all pixel points in the land remote sensing image through a K-means clustering algorithm according to all the preferred initial clustering center points to obtain a plurality of class clusters.
Further, according to the difference between the gray level distribution of all the pixel points in the two class clusters, combining part of adjacent class clusters, and marking the combined class clusters as marked class clusters, comprising the following specific steps:
Obtaining the merging possibility between the two class clusters according to the difference between the gray level distribution of all the pixel points in the two class clusters and the gray level difference of all the pixel points in the two class clusters;
and merging adjacent class clusters with the merging possibility larger than a preset second threshold value beta between the two class clusters, and marking all the class clusters after merging as marked class clusters.
Further, the merging possibility between the two class clusters is obtained according to the difference between the gray level distribution of all the pixel points in the two class clusters and the gray level difference of all the pixel points in the two class clusters, and the specific calculation method comprises the following steps:
Wherein σ s represents the standard deviation of the gray values of all the pixels in the s-th cluster, σ z represents the standard deviation of the gray values of all the pixels in the z-th cluster, Representing the average value of gray values of all pixel points in the s-th class cluster,/>Representing the average value of the gray values of all pixel points in the z-th cluster, H s representing the gray value of the cluster center point of the s-th cluster, H z representing the gray value of the cluster center point of the z-th cluster, exp () representing an exponential function based on a natural constant, i being an absolute value symbol, and P s,z representing the merging possibility between the s-th cluster and the z-th cluster.
Further, the area occupation ratio of each type of land is obtained according to a plurality of mark clusters in the land remote sensing image, and the land resource monitoring and planning are performed according to the area occupation ratio of each type of land, comprising the following specific steps:
collecting a large number of land remote sensing images, acquiring areas corresponding to a plurality of mark clusters in each land remote sensing image, recording images of the areas corresponding to the plurality of mark clusters in each land remote sensing image as each reference image, and training a CNN neural network by using the large number of reference images to obtain a trained CNN neural network, wherein a loss function is a cross entropy loss function;
Inputting an image of a region corresponding to a plurality of marker clusters into a land remote sensing image through a trained CNN neural network, and outputting lands of different categories;
The specific calculation method of the area ratio of each type of land comprises the following steps:
Wherein MJ represents the total area of the land remote sensing image, MJ u,L represents the area of the area corresponding to the L-th mark cluster of the land of the u-th type, G represents the number of all mark clusters corresponding to each type of land, and mu u represents the area occupation ratio of the land of the u-th type;
and carrying out land planning resource monitoring according to the area ratio of each type of land and combining with a GeoNode platform.
Further, the detecting the connected domain of the land remote sensing image to obtain a plurality of connected domains comprises the following specific steps:
And detecting the connected domain of the land remote sensing image based on a depth-first search algorithm to obtain a plurality of connected domains in the land remote sensing image.
The invention also provides a land planning resource monitoring system based on unmanned aerial vehicle remote sensing, which comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the processor realizes the steps of any land planning resource monitoring method based on unmanned aerial vehicle remote sensing when executing the computer program.
The technical scheme of the invention has the beneficial effects that: according to the gray level difference and the distance between each pixel point in each connected domain and other pixel points and between each pixel point in each connected domain and the pixel points in the connected domain adjacent to each connected domain, a plurality of optimal initial clustering center points are screened out from all the pixel points, and the accuracy of the initial clustering center point selection is improved; clustering all pixel points in the land remote sensing image according to a plurality of preferred initial clustering center points to obtain a plurality of class clusters, merging part of adjacent class clusters according to the difference between gray level distribution of all pixel points in the two class clusters, and marking the class clusters after merging as mark class clusters, thereby improving the clustering accuracy; the area occupation ratio of each type of land is obtained according to a plurality of mark clusters in the land remote sensing image, and land resource monitoring and planning are carried out according to the area occupation ratio of each type of land, so that the iteration times of clustering are reduced, the clustering efficiency is improved, and the land resource planning accuracy is also improved.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of steps of a land planning resource monitoring method based on unmanned aerial vehicle remote sensing;
Fig. 2 is a flow chart of land planning resource monitoring for unmanned aerial vehicle remote sensing.
Detailed Description
In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following detailed description is given below of the land planning resource monitoring method and system based on unmanned aerial vehicle remote sensing according to the invention, which are specific embodiments, structures, features and effects thereof, with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following specifically describes a specific scheme of the land planning resource monitoring method and system based on unmanned aerial vehicle remote sensing provided by the invention with reference to the accompanying drawings.
Referring to fig. 1, a flowchart of a land planning resource monitoring method based on unmanned aerial vehicle remote sensing according to an embodiment of the present invention is shown, and the method includes the following steps:
Step S001: and collecting a land remote sensing image.
The land resources include cultivated land, garden land, woodland, grassland, construction land and the like, so that land remote sensing images are acquired according to unmanned aerial vehicles, and land resources of different types are divided through the land remote sensing images, so that the monitoring of the land resource changes of different types is completed.
Specifically, a panoramic camera is installed on the unmanned aerial vehicle to collect an image of a land resource, the image of the land resource is subjected to gray-scale and Gaussian filtering denoising pretreatment, and the pretreated image is recorded as a land remote sensing image.
The graying and gaussian filtering denoising processes are known in the art, and detailed descriptions thereof are omitted herein.
Thus, a land remote sensing image is obtained.
Step S002: and detecting the connected domains of the land remote sensing image to obtain a plurality of connected domains, and screening a plurality of preferred initial clustering center points from all the pixel points according to the gray level difference and the distance between each pixel point in each connected domain and other pixel points in each connected domain and the gray level difference and the distance between each pixel point in each connected domain and the pixel points in the connected domain adjacent to each connected domain.
It should be noted that, since the land remote sensing image includes a plurality of land resource categories of different categories, that is, the same land resource category may be an area in the land remote sensing image, the connected domain in the land remote sensing image is obtained for further analysis.
Specifically, performing connected domain detection on the land remote sensing image based on a depth-first search (DFS) algorithm to obtain a plurality of connected domains in the land remote sensing image; among them, depth-first search (DFS) based algorithms are known techniques, and detailed descriptions thereof are omitted herein.
It should be further noted that, because the land remote sensing image contains a plurality of connected domains, each connected domain has at least one initial clustering center point; the greater the likelihood that the pixels in or around each connected domain are clustered into a class. When the gray scale difference between each pixel point and all other pixel points in each connected domain is smaller, the probability that each pixel point in the connected domain is an initial clustering center point is larger, and otherwise, the probability that each pixel point in the connected domain is an initial clustering center point is smaller. In order to distinguish the differences between different classes, the gray value of the pixel point in each class cluster is also greatly different from the gray values of the pixel points in other class clusters, so that the gray value of each pixel point in each connected domain is further analyzed by the gray values of the pixel points in surrounding connected domains.
Specifically, a reference coordinate system is constructed by taking a lower left pixel point of the land remote sensing image as a coordinate origin, taking a horizontal right pixel point as a horizontal axis and taking a vertical upward pixel point as a vertical axis. Respectively summing up the abscissa and the ordinate of all the pixel points in each connected domain and dividing the sum by the number of pixels to obtain the center point coordinate of each connected domain; thus, the center point of each connected domain is obtained.
According to the difference between the gray value of each pixel point in each connected domain and the gray values of all other pixel points, the distance between each pixel point in each connected domain and the center point of the connected domain, and the difference between the gray value of each pixel point in each connected domain and the average value of the gray values of all pixel points in adjacent connected domains, the possibility that each pixel point in each connected domain is the initial clustering center point is obtained, as an embodiment, the specific calculation method is as follows:
Where H i,j denotes a gray value of a j-th pixel in the i-th connected domain, H i,r denotes a gray value of a r-th pixel in the i-th connected domain, n denotes the number of all pixels in each connected domain, d i,j denotes a distance between the j-th pixel in the i-th connected domain and a center point of the connected domain, H i,e denotes a mean value of gray values of all pixels in an e-th adjacent connected domain of the i-th connected domain, m denotes the number of all adjacent connected domains of each connected domain, i is an absolute value symbol, norm () denotes a linear normalization function, and Q i,j denotes a possibility that the j-th pixel in the i-th connected domain is an initial cluster center point. The distances in this embodiment are all euclidean distances.
Wherein,Representing the difference between the gray value of each pixel point in each connected domain and the gray values of all other pixel points, when the difference is larger, the pixel point is indicated to be less likely to be an initial clustering center point, otherwise, the pixel point is indicated to be more likely to be the initial clustering center point; when the distance between each pixel point in each connected domain and the central point of the connected domain is larger, the probability that the pixel point is the initial clustering central point is smaller, otherwise, the probability that the pixel point is the initial clustering central point is larger; /(I)The difference between the gray value of each pixel point in each connected domain and the average value of the gray values of all the pixel points in the connected domains adjacent to each connected domain is represented, when the difference is smaller, the pixel point is indicated to be less likely to be the initial clustering center point, otherwise, the pixel point is indicated to be more likely to be the initial clustering center point.
A first threshold epsilon is preset, where epsilon=0.88 is taken as an example in this embodiment, and this embodiment is not particularly limited, where epsilon may be determined according to the specific implementation. And when the possibility that each pixel point in each connected domain is an initial clustering center point is greater than or equal to a preset first threshold epsilon, marking the pixel point as a candidate initial clustering center point.
It should be noted that, when the distance between each candidate initial cluster center point and other candidate initial cluster center points is larger, the probability that the candidate initial cluster center point is used as an initial cluster center point of a class cluster is larger, whereas when the distance between each candidate initial cluster center point and other candidate initial cluster center points is smaller, it is noted that many candidate initial cluster center points exist around the candidate initial cluster center point, the probability that the candidate initial cluster center point is used as an initial cluster center point of a class cluster is smaller.
Specifically, according to the distance between each candidate initial clustering center point in each connected domain and all candidate initial clustering center points in the connected domain adjacent to each connected domain and the possibility that each pixel point in each connected domain is an initial clustering center point, the possibility that each candidate initial clustering center point in each connected domain is used as a preferred initial clustering center point is obtained, and as an embodiment, the specific calculation method is as follows:
where d i,c,e,v represents the distance between the c-th candidate initial cluster center point in the i-th connected domain and the v-th candidate initial cluster center point in the e-th adjacent connected domain, M represents the number of all candidate initial cluster center points in any one of the connected domains adjacent to each connected domain, N represents the number of all connected domains adjacent to each connected domain, Q i,j represents the likelihood that the j-th candidate initial cluster center point in the i-th connected domain is the initial cluster center point, Q i,min represents the minimum value of the likelihood that all initial cluster center points in the i-th connected domain are the initial cluster center points, Q i,max represents the maximum value of the likelihood that all initial cluster center points in the i-th connected domain are the initial cluster center points, and W i,c represents the likelihood that the c-th candidate initial cluster center point in the i-th connected domain is the preferred initial cluster center point.
The greater the distance between each candidate initial cluster center point in each connected domain and all candidate initial cluster center points in the connected domain adjacent to each connected domain, the greater the possibility that the candidate initial cluster center point is used as a preferred initial cluster center point, and conversely, the smaller the possibility that the candidate initial cluster center point is used as a preferred initial cluster center point. When the probability that each initial cluster center point in each connected domain is an initial cluster center point is higher, the probability that the candidate initial cluster center point is a preferred initial cluster center point is higher, and conversely, the probability that the candidate initial cluster center point is a preferred initial cluster center point is lower.
And selecting all candidate initial cluster center points in each connected domain as one candidate initial cluster center point with the highest possibility of the preferred initial cluster center point, and recording the candidate initial cluster center point as the preferred initial cluster center point.
So far, all the preferred initial cluster center points are obtained.
Step S003: clustering all pixel points in the land remote sensing image according to a plurality of preferred initial clustering center points to obtain a plurality of class clusters, merging part of adjacent class clusters according to the difference between gray level distribution of all pixel points in the two class clusters, and marking the class clusters after merging as marked class clusters.
It should be noted that, when analyzing according to the connected domain, a small river may appear in the middle of the land resource of the same cultivated land, so that two areas of the same type are divided into two or more areas, and therefore, clustering is performed through the selected preferred initial clustering center point, and similar clusters are combined according to the gray level difference of the pixel points in different clusters after clustering and the difference between the gray level distribution of all the pixel points in different clusters.
Specifically, all pixel points in the land remote sensing image are clustered through a K-means clustering algorithm according to all the preferred initial clustering center points, and a plurality of class clusters are obtained. The K-means clustering algorithm is a well-known technique, and will not be described in detail here.
According to the difference between the gray level distribution of all the pixel points in the two class clusters and the gray level difference of all the pixel points in the two class clusters, the merging possibility between the two class clusters is obtained, and as an embodiment, the specific calculation method is as follows:
Wherein σ s represents the standard deviation of the gray values of all the pixels in the s-th cluster, σ z represents the standard deviation of the gray values of all the pixels in the z-th cluster, Representing the average value of gray values of all pixel points in the s-th class cluster,/>Representing the average value of the gray values of all pixel points in the z-th cluster, H s representing the gray value of the cluster center point of the s-th cluster, H z representing the gray value of the cluster center point of the z-th cluster, exp () representing an exponential function based on a natural constant, i being an absolute value symbol, and P s,z representing the merging possibility between the s-th cluster and the z-th cluster.
Wherein,Representing the difference between the gray distribution of all the pixels in the two clusters, the smaller the difference between the gray distribution of all the pixels in the two clusters, the closer to 1 the ratio of the standard deviation of the gray values of all the pixels in the two clusters. Therefore, when the difference between the gray scale distribution of all the pixels in the two class clusters is smaller, the merging possibility between the two class clusters is larger, and conversely, the merging possibility between the two class clusters is smaller. /(I)Representing the difference between the average values of the gray values of all the pixels in the two class clusters, the smaller the difference is, the greater the merging possibility between the two class clusters is, and conversely, the smaller the merging possibility between the two class clusters is. When the difference between the gray values of the cluster center points of the two class clusters is smaller, the merging possibility between the two class clusters is larger, and conversely, the merging possibility between the two class clusters is smaller.
A second threshold β is preset, where β=0.9 is taken as an example in this embodiment, and this embodiment is not particularly limited, where β may be determined according to the specific implementation.
And merging adjacent class clusters with the merging possibility larger than a preset second threshold value beta between the two class clusters, and marking all the class clusters after merging as marked class clusters.
So far, a plurality of marker class clusters are obtained.
Step S004: and monitoring and planning the land resources according to a plurality of mark clusters in the land remote sensing image.
It should be noted that, in practice, some lands of the same category may not be all together, and may be divided due to lands of other categories, so that land resources of the same category in the plurality of marker category clusters may correspond to a plurality of marker category clusters, and thus, gradient resource category division of the plurality of marker category clusters is achieved through a neural network.
Specifically, a large number of land remote sensing images are collected, areas corresponding to a plurality of mark clusters in each land remote sensing image are obtained, images of the areas corresponding to the plurality of mark clusters in each land remote sensing image are recorded as each reference image, a CNN neural network is trained by using the large number of reference images, and a trained CNN neural network is obtained, wherein the loss function is a cross entropy loss function. The CNN neural network structure is a known technology, and will not be described in detail herein.
And inputting images of areas corresponding to a plurality of marker clusters into the land remote sensing image through the trained CNN neural network, and outputting lands of different categories.
According to the occupied area of each type of land and the total area of the land remote sensing image, the area occupied ratio of each type of land is obtained, and as an embodiment, the specific calculation method comprises the following steps:
Where MJ represents the total area of the land remote sensing image, MJ u,L represents the area of the region corresponding to the L-th marker cluster of the u-th type land, G represents the number of all marker clusters corresponding to each type land, and μ u represents the area ratio of the u-th type land.
Thus, the area ratio of each type of land is obtained.
And carrying out land planning resource monitoring according to the area ratio of each type of land and combining with a GeoNode platform. The flow chart of land planning resource monitoring of unmanned aerial vehicle remote sensing is shown in fig. 2.
It should be noted that, in this embodiment, the exp (-x) model is only used to indicate that the result output by the negative correlation and constraint model is within the (0, 1) interval, and in the implementation, other models with the same purpose may be replaced, and this embodiment is only described by taking the exp (-x) model as an example, and is not limited to this, where x refers to the input of the model.
The embodiment provides a land planning resource monitoring system based on unmanned aerial vehicle remote sensing, which comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the processor realizes the land planning resource monitoring method based on unmanned aerial vehicle remote sensing in the steps S001 to S004 when executing the computer program.
This embodiment is completed.
The above description is only of the preferred embodiments of the present invention and is not intended to limit the invention, but any modifications, equivalent substitutions, improvements, etc. within the principles of the present invention should be included in the scope of the present invention.
Claims (10)
1. The land planning resource monitoring method based on unmanned aerial vehicle remote sensing is characterized by comprising the following steps of:
Acquiring a land remote sensing image;
Detecting connected domains of the land remote sensing image to obtain a plurality of connected domains, and screening a plurality of preferred initial clustering center points from all the pixel points according to gray level differences and distances between each pixel point in each connected domain and other pixel points in each connected domain and gray level differences and distances between each pixel point in each connected domain and pixel points in the connected domain adjacent to each connected domain;
Clustering all pixel points in the land remote sensing image according to a plurality of preferred initial clustering center points to obtain a plurality of class clusters, merging part of adjacent class clusters according to the difference between gray level distribution of all pixel points in the two class clusters, and marking the class clusters after merging as marked class clusters;
And obtaining the area occupation ratio of each type of land according to a plurality of mark clusters in the land remote sensing image, and carrying out land resource monitoring and planning according to the area occupation ratio of each type of land.
2. The land planning resource monitoring method based on unmanned aerial vehicle remote sensing according to claim 1, wherein the selecting a plurality of preferred initial clustering center points from all the pixel points according to the gray scale difference and the distance between each pixel point and other pixel points in each connected domain and the gray scale difference and the distance between each pixel point in each connected domain and the pixel points in the connected domain adjacent to each connected domain comprises the following specific steps:
Obtaining the possibility that each pixel point in each connected domain is an initial clustering center point according to the gray difference between each pixel point in each connected domain and all other pixel points, the distance between each pixel point in each connected domain and the center point of the connected domain and the gray difference between each pixel point in each connected domain and all pixel points in adjacent connected domains;
When the possibility that each pixel point in each connected domain is an initial clustering center point is greater than or equal to a preset first threshold epsilon, marking the pixel point as a candidate initial clustering center point;
According to the distance between each candidate initial clustering central point in each connected domain and all candidate initial clustering central points in the connected domain adjacent to each connected domain and the possibility that each pixel point in each connected domain is an initial clustering central point, the possibility that each candidate initial clustering central point in each connected domain is used as a preferable initial clustering central point is obtained;
and selecting all candidate initial cluster center points in each connected domain as one candidate initial cluster center point with the highest possibility of the preferred initial cluster center point, and recording the candidate initial cluster center point as the preferred initial cluster center point.
3. The land planning resource monitoring method based on unmanned aerial vehicle remote sensing according to claim 2, wherein the obtaining the possibility that each pixel point in each connected domain is an initial clustering center point according to the gray scale difference between each pixel point and all other pixel points in each connected domain, the distance between each pixel point in each connected domain and the center point of the connected domain, and the gray scale difference between each pixel point in each connected domain and all pixel points in adjacent connected domains comprises the following specific calculation method:
Where H i,j denotes a gray value of a j-th pixel in the i-th connected domain, H i,r denotes a gray value of a r-th pixel in the i-th connected domain, n denotes the number of all pixels in each connected domain, d i,j denotes a distance between the j-th pixel in the i-th connected domain and a center point of the connected domain, H i,e denotes a mean value of gray values of all pixels in an e-th adjacent connected domain of the i-th connected domain, m denotes the number of all adjacent connected domains of each connected domain, i is an absolute value symbol, norm () denotes a linear normalization function, and Q i,j denotes a possibility that the j-th pixel in the i-th connected domain is an initial cluster center point.
4. The land planning resource monitoring method based on unmanned aerial vehicle remote sensing according to claim 2, wherein the specific calculation method is as follows according to the distance between each candidate initial clustering center point in each connected domain and all candidate initial clustering center points in the connected domain adjacent to each connected domain, and the possibility that each pixel point in each connected domain is an initial clustering center point, so as to obtain the possibility that each candidate initial clustering center point in each connected domain is used as a preferred initial clustering center point:
where d i,c,e,v represents the distance between the c-th candidate initial cluster center point in the i-th connected domain and the v-th candidate initial cluster center point in the e-th adjacent connected domain, M represents the number of all candidate initial cluster center points in any one of the connected domains adjacent to each connected domain, N represents the number of all connected domains adjacent to each connected domain, Q i,j represents the likelihood that the j-th candidate initial cluster center point in the i-th connected domain is the initial cluster center point, Q i,min represents the minimum value of the likelihood that all initial cluster center points in the i-th connected domain are the initial cluster center points, Q i,max represents the maximum value of the likelihood that all initial cluster center points in the i-th connected domain are the initial cluster center points, and W i,c represents the likelihood that the c-th candidate initial cluster center point in the i-th connected domain is the preferred initial cluster center point.
5. The land planning resource monitoring method based on unmanned aerial vehicle remote sensing according to claim 1, wherein the clustering of all pixel points in the land remote sensing image according to a plurality of preferred initial clustering center points to obtain a plurality of clusters comprises the following specific steps:
And clustering all pixel points in the land remote sensing image through a K-means clustering algorithm according to all the preferred initial clustering center points to obtain a plurality of class clusters.
6. The land planning resource monitoring method based on unmanned aerial vehicle remote sensing according to claim 1, wherein the merging of the partially adjacent class clusters according to the difference between the gray level distribution of all the pixel points in the two class clusters, and the marking of the merged class clusters as the marked class clusters comprises the following specific steps:
Obtaining the merging possibility between the two class clusters according to the difference between the gray level distribution of all the pixel points in the two class clusters and the gray level difference of all the pixel points in the two class clusters;
and merging adjacent class clusters with the merging possibility larger than a preset second threshold value beta between the two class clusters, and marking all the class clusters after merging as marked class clusters.
7. The land planning resource monitoring method based on unmanned aerial vehicle remote sensing according to claim 6, wherein the merging possibility between two clusters is obtained according to the difference between gray level distribution of all pixel points in the two clusters and the gray level difference of all pixel points in the two clusters, and the specific calculation method comprises the following steps:
Wherein σ s represents the standard deviation of the gray values of all the pixels in the s-th cluster, σ z represents the standard deviation of the gray values of all the pixels in the z-th cluster, Representing the average value of gray values of all pixel points in the s-th class cluster,/>Representing the average value of the gray values of all pixel points in the z-th cluster, H s representing the gray value of the cluster center point of the s-th cluster, H z representing the gray value of the cluster center point of the z-th cluster, exp () representing an exponential function based on a natural constant, i being an absolute value symbol, and P s,z representing the merging possibility between the s-th cluster and the z-th cluster.
8. The land planning resource monitoring method based on unmanned aerial vehicle remote sensing according to claim 1, wherein the land planning resource monitoring and planning is performed according to the area occupation ratio of each type of land by obtaining the area occupation ratio of each type of land according to a plurality of mark clusters in the land remote sensing image, and comprises the following specific steps:
collecting a large number of land remote sensing images, acquiring areas corresponding to a plurality of mark clusters in each land remote sensing image, recording images of the areas corresponding to the plurality of mark clusters in each land remote sensing image as each reference image, and training a CNN neural network by using the large number of reference images to obtain a trained CNN neural network, wherein a loss function is a cross entropy loss function;
Inputting an image of a region corresponding to a plurality of marker clusters into a land remote sensing image through a trained CNN neural network, and outputting lands of different categories;
The specific calculation method of the area ratio of each type of land comprises the following steps:
Wherein MJ represents the total area of the land remote sensing image, MJ u,L represents the area of the area corresponding to the L-th mark cluster of the land of the u-th type, G represents the number of all mark clusters corresponding to each type of land, and mu u represents the area occupation ratio of the land of the u-th type;
and carrying out land planning resource monitoring according to the area ratio of each type of land and combining with a GeoNode platform.
9. The land planning resource monitoring method based on unmanned aerial vehicle remote sensing according to claim 1, wherein the performing connected domain detection on the land remote sensing image to obtain a plurality of connected domains comprises the following specific steps:
And detecting the connected domain of the land remote sensing image based on a depth-first search algorithm to obtain a plurality of connected domains in the land remote sensing image.
10. Land planning resource monitoring system based on unmanned aerial vehicle remote sensing, comprising a memory, a processor and a computer program stored in the memory and running on the processor, characterized in that the processor, when executing the computer program, realizes the steps of the land planning resource monitoring method based on unmanned aerial vehicle remote sensing according to any one of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410443219.2A CN118135444A (en) | 2024-04-12 | 2024-04-12 | Land planning resource monitoring method and system based on unmanned aerial vehicle remote sensing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410443219.2A CN118135444A (en) | 2024-04-12 | 2024-04-12 | Land planning resource monitoring method and system based on unmanned aerial vehicle remote sensing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118135444A true CN118135444A (en) | 2024-06-04 |
Family
ID=91243680
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410443219.2A Pending CN118135444A (en) | 2024-04-12 | 2024-04-12 | Land planning resource monitoring method and system based on unmanned aerial vehicle remote sensing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118135444A (en) |
-
2024
- 2024-04-12 CN CN202410443219.2A patent/CN118135444A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113378632B (en) | Pseudo-label optimization-based unsupervised domain adaptive pedestrian re-identification method | |
| CN110738247B (en) | Fine-grained image classification method based on selective sparse sampling | |
| CN112734775B (en) | Image labeling, image semantic segmentation and model training methods and devices | |
| US10275683B2 (en) | Clustering-based person re-identification | |
| CN109978893A (en) | Training method, device, equipment and the storage medium of image, semantic segmentation network | |
| CN110889318B (en) | Lane detection method and device using CNN | |
| CN114092832B (en) | High-resolution remote sensing image classification method based on parallel hybrid convolutional network | |
| CN107633226B (en) | Human body motion tracking feature processing method | |
| US20080071708A1 (en) | Method and System for Data Classification Using a Self-Organizing Map | |
| CN108875816A (en) | Merge the Active Learning samples selection strategy of Reliability Code and diversity criterion | |
| CN104765768A (en) | Mass face database rapid and accurate retrieval method | |
| CN109635010B (en) | User characteristic and characteristic factor extraction and query method and system | |
| CN111126396A (en) | Image recognition method and device, computer equipment and storage medium | |
| CN112633382A (en) | Mutual-neighbor-based few-sample image classification method and system | |
| CN112232371A (en) | American license plate recognition method based on YOLOv3 and text recognition | |
| CN116206185A (en) | Lightweight small target detection method based on improved YOLOv7 | |
| CN112685504A (en) | Production process-oriented distributed migration chart learning method | |
| CN111539910B (en) | Rust area detection method and terminal equipment | |
| CN114972759A (en) | Remote sensing image semantic segmentation method based on hierarchical contour cost function | |
| CN111352926B (en) | Method, device, equipment and readable storage medium for data processing | |
| CN114897085A (en) | Clustering method based on closed subgraph link prediction and computer equipment | |
| CN112115996B (en) | Image data processing method, device, equipment and storage medium | |
| CN113743443A (en) | Image evidence classification and identification method and device | |
| CN110287970B (en) | Weak supervision object positioning method based on CAM and covering | |
| CN114219051B (en) | Image classification method, classification model training method and device and electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination |