Change detection method and system for contiguous inefficient industrial land based on remote sensing images

By acquiring remote sensing images at different times, identifying and combining pixels of bare land types, and assessing land value, the accuracy problem of detecting the evolution of contiguous inefficient industrial land is solved, and quantitative analysis of land evolution rate is achieved.

CN118658070BActive Publication Date: 2024-11-19ZHEJIANG URBAN & RURAL PLANNING DESIGN INST
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411142918.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-20
Publication Date
2024-11-19
Estimated Expiration
2044-08-20

AI Technical Summary

Technical Problem

Existing technologies struggle to accurately detect and assess the evolution of contiguous, inefficient industrial land, especially during urban planning and infrastructure improvements, where the deceptive nature of changes in remote sensing imagery leads to inaccurate assessments.

Method used

By acquiring remote sensing images at different times, land use types and pixel information are identified, pixels of bare land types are combined to form clusters, the value of bare land is assessed, and the land use evolution rate is calculated. Cluster analysis and value assessment models are used to improve detection accuracy.

Benefits of technology

It enables accurate detection of changes in contiguous inefficient industrial land and quantitative analysis of land use evolution trends, improving the accuracy of assessments and the completeness of time series data.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118658070B_ABST
    Figure CN118658070B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of remote sensing technology, and in particular to a method and system for detecting changes in contiguous inefficient industrial land based on remote sensing images. The method comprises: obtaining target images collected at different times for target land; for each target image, identifying each pixel corresponding to each land type in the target image and the pixel information of each pixel; wherein the land type includes built type, bare land type and unplanned type; the pixel information includes the number of pixels and the position of the pixels; combining the pixels of the bare land type in the target image to obtain at least one developable bare land cluster; determining the bare land value of each bare land cluster in the target image according to the pixel information of each land type in each bare land cluster; determining the land evolution rate of the target land according to the bare land value of each bare land cluster in each target image.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of remote sensing technology, and in particular to a method and system for detecting changes in contiguous inefficient industrial land based on remote sensing images. Background Art

[0002] Contiguous inefficient industrial land refers to one or more pieces of industrial land distributed geographically continuously, which perform poorly in terms of economic benefits, environmental impact or social contribution, or are urban stock construction land with scattered layout, unreasonable use and dilapidated buildings. This situation may occur on the edge of the city, in industrial parks or historical industrial concentration areas. High-resolution remote sensing technology has broad application potential in the detection of contiguous inefficient industrial land. This technology can provide detailed ground information to help identify and assess the status of industrial land.

[0003] However, with the reasons of urban planning, industrial upgrading, infrastructure improvement, etc., inefficient industrial land may be transformed, but the planning and upgrading of contiguous inefficient industrial land are not synchronized. When detecting changes in land plots, changes in remote sensing images of some plots are often confusing, resulting in the inability to accurately assess the evolution of inefficient industrial land. Summary of the invention

[0004] In order to solve the technical problem of improving the recognition accuracy of industrial land evolution, the purpose of the present invention is to provide a method for detecting changes in contiguous inefficient industrial land based on remote sensing images. The technical solution adopted is as follows:

[0005] Acquire target images collected at different times for the target land;

[0006] For each target image, identify each pixel corresponding to each land use type in the target image and the pixel information of each pixel; wherein the land use type includes built type, bare land type and unplanned type; the pixel information includes the number of pixels and the position of the pixels;

[0007] Combining the bare land type pixel points in the target image to obtain at least one developable bare land cluster;

[0008] Determine the bare land value of each bare land cluster in the target image according to the pixel point information of each land use type in each bare land cluster;

[0009] According to the bare land value of each bare land cluster in each target image, the land use evolution rate of the target land is determined.

[0010] In one embodiment, pixels of the bare ground type in the target image are combined to obtain at least one developable bare ground cluster, including:

[0011] Generate at least one bare ground connected region according to each pixel point of the bare ground type in the target image and a preset number of connections;

[0012] Cluster analysis is performed on the connected areas of each bare land to obtain at least one developable bare land cluster.

[0013] In one embodiment, cluster analysis is performed on each bare land connected area to obtain at least one developable bare land cluster, including:

[0014] Cluster analysis is performed on each bare land connected area according to the centroid distance to obtain at least one developable bare land cluster; wherein the distance between any two bare land connected areas in any developable bare land cluster is less than a distance threshold.

[0015] In one embodiment, determining the bare land value of each bare land cluster in the target image according to the pixel point information of each land use type in each bare land cluster includes:

[0016] Divide the target image into multiple candidate plots;

[0017] For each candidate plot, the land use value of the candidate plot is determined according to the pixel point information corresponding to each land use type in the candidate plot;

[0018] According to the land use value of each candidate plot and the distribution position of each candidate plot in the target image, cluster analysis is performed on each candidate plot to obtain at least one candidate plot cluster;

[0019] For each bare land cluster, select a target land cluster corresponding to the bare land cluster from among the candidate land clusters;

[0020] The bare land value of the bare land cluster is determined according to the land value of the land group of the target land group; the land value of the land group corresponding to any candidate land group is the normalized result of the average land value of all candidate land groups contained in the candidate land group.

[0021] In one embodiment, determining the land use value of the candidate land parcel according to the pixel point information of each land use type in the candidate land parcel includes:

[0022] According to the distribution of pixel points corresponding to buildings in the candidate plots, the supporting factors corresponding to the candidate plots are determined;

[0023] According to the pixel point information corresponding to the unplanned land use type in the candidate plot, the unplanned factor corresponding to the candidate is determined;

[0024] The land use value of the candidate plot is determined based on the pixel information corresponding to the built-up type in the candidate plot, the pixel information corresponding to the bare land type, the number of bare land connected areas, the unplanned factor and the supporting factor.

[0025] In one embodiment, selecting a target land parcel cluster corresponding to a bare land cluster from among the candidate land parcel clusters includes:

[0026] For each candidate plot cluster, the candidate plots corresponding to each bare land connected area in the candidate plot cluster are used as matching plots;

[0027] From each candidate plot cluster, the candidate plot cluster containing the largest number of matching plots is selected as the target plot cluster corresponding to the bare land cluster.

[0028] In one embodiment, determining the bare land value of the bare land cluster according to the land value of the land parcel group of the target land parcel cluster includes:

[0029] Determine the overlap area between the bare land cluster and the target plot cluster;

[0030] The bare land value of the bare land cluster is determined based on the overlapping area and the land value of the plot family of the target plot cluster.

[0031] In one embodiment, the land use value of the candidate plot is determined based on the pixel point information corresponding to the built type, the pixel point information corresponding to the bare land type, the number of bare land connected areas, the unplanned factor and the supporting factor in the candidate plot, including:

[0032] Determine the explicit land use value of the candidate plots based on the pixel information corresponding to the built-up type in the candidate plots, the pixel information corresponding to the bare land type, the number of bare land connected areas, the unplanned factor and the supporting factor;

[0033] According to the regional shape parameters of the connected areas of each bare land in the candidate plots, the implicit land use value of the candidate plots is determined;

[0034] The land value of the candidate plot is determined based on the explicit land value and implicit land value of the candidate plot.

[0035] In one embodiment, the regional shape parameters include the edge line of the bare land connected area and the slope of the edge line; and determining the land value of the candidate land parcel according to the explicit land value and the implicit land value of the candidate land parcel includes:

[0036] According to the regional shape parameters of each bare land connected area in the candidate plot, each edge line type and the distribution probability of each edge line type are determined;

[0037] Using the gray-level co-occurrence matrix, each bare land connected area is traversed to obtain the distribution entropy corresponding to each bare land connected area;

[0038] The implicit land value of the candidate plots is determined based on the distribution probability of each edge line type and the distribution entropy corresponding to each bare land connected area.

[0039] In one embodiment, if the target images at different times include a target image at a first time and a target image at a second time, and the second time is later than the first time, then according to the bare land value of each bare land cluster in each target image, determining the land use evolution rate of the target land includes:

[0040] Determine a value weight according to the bare land value of each bare land cluster in the target image at the first moment and the bare land value of each bare land cluster in the second target image;

[0041] The land use evolution rate of the target land is determined according to the pixel point information corresponding to each bare land cluster in the target image at the first moment, the pixel point information corresponding to each bare land cluster in the target image at the second moment, and the value weight.

[0042] In the second aspect, the present application provides a system for detecting changes in contiguous inefficient industrial land based on remote sensing images, the system comprising:

[0043] An acquisition module is used to acquire each target image collected at different times for the target land;

[0044] A pixel extraction module is used to identify, for each target image, each pixel corresponding to each land use type in the target image and the pixel information of each pixel; wherein the land use type includes a built type, a bare land type and an unplanned type; and the pixel information includes the number of pixels and the position of the pixels;

[0045] A clustering module, used for combining the pixels of the bare land type in the target image to obtain at least one developable bare land cluster;

[0046] A land use evaluation module, used to determine the bare land value of each bare land cluster in the target image according to the pixel point information of each land use type in each bare land cluster;

[0047] The evolution determination module is used to determine the land use evolution rate of the target land according to the bare land value of each bare land cluster in each target image.

[0048] The present invention has the following beneficial effects: acquiring each target image collected at different times for the target land, ensuring the data integrity in the time series, and identifying each pixel point and the pixel point information of each pixel point corresponding to each land use type for each target image. The land use type is clearly divided into built type, bare land type and unplanned type, which is helpful for subsequent classification analysis and value assessment. The pixel point information includes the number of pixels and the position of the pixels, which provides the possibility for accurate spatial analysis. The pixel points of the bare land type in the target image are combined to obtain at least one developable bare land cluster, thereby realizing the effective integration of the bare land; according to the pixel point information of each land use type in each bare land cluster, the bare land value of each bare land cluster in the target image is determined. The assessment of land value takes into account the influence of multiple land use types, making the assessment result more comprehensive and accurate. According to the bare land land value of each bare land cluster in each target image, the land evolution rate of the target land is determined. The quantitative analysis of the land evolution trend is realized, and the accuracy of the analysis is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0050] Figure 1 A schematic diagram of a flow chart of a method for detecting changes in contiguous inefficient industrial land based on remote sensing images provided by an embodiment of the present invention;

[0051] Figure 2 A schematic diagram of a flow chart for determining the bare land value of each bare land cluster in a target image provided by an embodiment of the present invention;

[0052] Figure 3 A schematic diagram of a first process for determining the land use value of a candidate land parcel provided by an embodiment of the present invention;

[0053] Figure 4 A schematic diagram of a second process for determining the land use value of a candidate land parcel provided by an embodiment of the present invention;

[0054] Figure 5 A schematic diagram of a flow chart of determining the land value of a candidate land parcel based on the explicit land value and implicit land value of the candidate land parcel provided in one embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a method for detecting changes in contiguous inefficient industrial land based on remote sensing images proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0056] 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.

[0057] The following specifically describes a method for detecting changes in contiguous inefficient industrial land based on remote sensing images provided by the present invention in conjunction with the accompanying drawings.

[0058] See also Figure 1 , which shows a method flow chart of a method for detecting changes in contiguous inefficient industrial land based on remote sensing images provided by an embodiment of the present invention, specifically comprising:

[0059] S101, acquiring target images collected at different times for the target land.

[0060] Among them, the target land is the object of analysis or processing, which may be a piece of land to be developed, an urban area, an industrial area, etc. Different moments refer to the image acquisition of the target land at multiple time points. These time points can be continuous or discrete, depending on the specific needs of the project. Images at different moments can reflect the changes of the target land over time, such as seasonal changes, construction progress, vegetation growth, etc. When acquiring target images, use image acquisition equipment such as satellites, drones, ground cameras, etc. to take photos or videos of the target land.

[0061] Exemplarily, a sunny day is selected as the condition for collecting the target image, because on a sunny day there is sufficient sunlight and the electromagnetic radiation reflected by the ground is strong, which is conducive to the generation of high-quality remote sensing images. Select the panchromatic image type, which can more realistically reflect the natural color and texture of the ground, which is conducive to the subsequent land use classification and identification. When the satellite passes over the target area, the onboard remote sensing sensor will start working to capture electromagnetic radiation reflected or spontaneously generated by the ground. By processing these electromagnetic radiation data, the remote sensing sensor can generate high-resolution remote sensing images of the target town. These images have rich ground object information and details, providing basic data for subsequent land use analysis. Using the satellite communication system, high-resolution remote sensing images are transmitted to the earth receiving station in real time. After receiving the image data, the receiving station will store it on the data server.

[0062] S102, for each target image, identifying each pixel corresponding to each land use type in the target image and pixel information of each pixel.

[0063] The land use type includes built-up type, bare land type and unplanned type. The pixel point information includes the number of pixels and the pixel point location.

[0064] Specifically, for each target image, it is necessary to identify the built type, bare land type and unplanned type. The built type usually refers to the area that has been developed and constructed, such as factories, roads, buildings, etc. The bare land type refers to the exposed ground that has not been developed and constructed, such as vacant land, unused land, etc. The unplanned type refers to the area that has not been planned or constructed yet, and these areas may be developed in the future.

[0065] Exemplarily, the built type includes building type and impervious surface type. The building type corresponds to the building area, which is the actual use area within the plot, representing different buildings and corresponding utilization types. The building area usually has unique color, texture and shape characteristics, which can be identified by the target image segmentation algorithm. The identified building area can be used to calculate the building's footprint, distribution, etc.

[0066] The impervious surface type corresponds to the impervious surface area, which is also the usable area within the plot, but different from the building area, it usually refers to the open ground paved with concrete. The impervious surface area can be used for stacking goods, as a concentrated activity area, etc. On the target image, the impervious surface area may show different color or texture characteristics from the surrounding environment, such as brighter tones or smoother surfaces.

[0067] The bare land type corresponds to the bare land area, which is divided into two parts: the area with remaining buildings and the unplanned area. The area with remaining buildings is usually the remaining part of the building demolition or construction process, which may appear as discretely distributed bare ground on the image.

[0068] Unplanned types correspond to unplanned areas, which are empty spaces where no construction has been carried out. They may appear as bare ground distributed in patches on the image. For example, unplanned areas can be green areas, which refer to areas of trees and vegetation that are connected in patches. They are the green areas within the plot. However, scattered areas of trees and vegetation may also be considered unplanned areas, depending on their distribution and density on the image. Identifying green areas is of great significance for evaluating the ecological and environmental quality of the plot and planning future greening construction.

[0069] It is understandable that for each land use type, the corresponding pixel points need to be extracted. The pixel point information includes the number of pixels, that is, the number of pixels occupied by each land use type in the image. It is also necessary to extract the position information of each pixel point, that is, their coordinate position in the target image.

[0070] S103, combining bare land type pixel points in the target image to obtain at least one developable bare land cluster.

[0071] Specifically, when combining the bare land type pixels in the target image, the basis for the combination can be the spatial proximity between the pixels, similar attribute characteristics (such as soil type, terrain slope, etc.) or other relevant planning criteria. By combining the pixels, at least one developable bare land cluster can be obtained. Each bare land cluster is composed of a group of interrelated bare land pixels with common development potential. The formation of bare land clusters helps to understand the distribution and development potential of bare land in the target land.

[0072] S104, determining the bare land value of each bare land cluster in the target image according to the pixel point information of each land use type in each bare land cluster.

[0073] Optionally, the pixel information belonging to the same bare land cluster is integrated to obtain key information such as the total number of pixels and land use type distribution of the cluster. A bare land value assessment model is established based on factors such as land use type, number of pixels, and location. The model may consider the market demand, development potential, and environmental impact factors of the land use type. The integrated data is input into the value assessment model to calculate the bare land value of each bare land cluster.

[0074] S105, determining the land use evolution rate of the target land according to the bare land value of each bare land cluster in each target image.

[0075] Optionally, perform a time series analysis on the land values ​​of bare land clusters in the target images collected at different times. Calculate the land evolution rate by comparing the land values ​​at different times. This may involve analysis of the growth rate and change trend of the land value. Visualize the land evolution rate in the form of charts, curves, etc. for intuitive understanding and analysis.

[0076] In an exemplary embodiment, bare ground type pixel points in a target image are combined to obtain at least one developable bare ground cluster, including: generating at least one bare ground connected region according to each bare ground type pixel point in the target image and a preset number of connections; and performing cluster analysis on each bare ground connected region to obtain at least one developable bare ground cluster.

[0077] Specifically, according to the preset number of connections (i.e., the connection relationship between a pixel and its adjacent pixels), the algorithm connects these bare land pixels into connected areas. These connected areas represent the continuous distribution of bare land in the image and are the basis for subsequent clustering analysis.

[0078] Specifically, cluster analysis is performed on each bare land connected area to obtain at least one developable bare land cluster, including: cluster analysis is performed on each bare land connected area according to centroid distance to obtain at least one developable bare land cluster.

[0079] Optionally, the algorithm first calculates the centroid of each bare ground connected region, which is the average of the coordinates of all pixels in the region and represents the center of the region. The distance between the centroids is used as the basis for cluster analysis, specifically comparing the distances between the centroids of different bare ground connected regions and grouping them according to these distances. During the clustering process, the algorithm will merge bare ground connected regions with close centroid distances into the same cluster. This process will continue until all bare ground connected regions are assigned to a cluster or a certain stop condition is met (such as the number of clusters reaches a preset value). In order to ensure that the bare ground connected regions in a cluster are relatively close in space, a distance threshold is set. In any developable bare ground cluster, the centroid distance between any two bare ground connected regions must be less than this distance threshold. If the centroid distance between two bare ground connected regions in a cluster exceeds the distance threshold, they may be reallocated to different clusters. It should be noted that the method for setting the distance threshold is a conventional technical means in clustering algorithms, and the specific setting process will not be repeated.

[0080] Further, such as Figure 2 As shown, according to the pixel information of each land use type in each bare land cluster, the bare land value of each bare land cluster in the target image is determined, including:

[0081] S201, dividing the target image into multiple candidate plots.

[0082] The target image is divided into multiple candidate plots, which are the basic units for subsequent analysis, and each plot contains a certain number of pixels.

[0083] S202: for each candidate land parcel, determine the land use value of the candidate land parcel according to the pixel point information corresponding to each land use type in the candidate land parcel.

[0084] For each candidate plot, the land use value of the plot is evaluated based on the pixel information corresponding to each land use type within it.

[0085] S203: Perform cluster analysis on each candidate land parcel according to the land use value of each candidate land parcel and the distribution position of each candidate land parcel in the target image to obtain at least one candidate land parcel cluster.

[0086] According to the land use value of each candidate plot and their distribution position in the target image, cluster analysis is performed. This process aims to merge plots with similar land use value and adjacent positions into the same plot cluster.

[0087] For example, the candidate plots are clustered according to the land use value of the plots. The minimum distance between the core points in the candidate plots and the difference in the land use value of the plots. The K-means clustering method is used, and the optimal number of clusters K is determined by the elbow method, so as to obtain several candidate plot clusters.

[0088] S204 , for each bare land cluster, selecting a target land cluster corresponding to the bare land cluster from among the candidate land clusters.

[0089] The plots in the same cluster are close to each other and have similar actual land use values. For each bare land cluster, it is necessary to select the corresponding target plot cluster from the previously formed plot clusters. This selection process may be based on factors such as the spatial relationship between bare land clusters and plot clusters, and the similarity of land use types.

[0090] S205, determining the bare land value of the bare land cluster according to the land value of the land parcel group of the target land parcel cluster.

[0091] The bare land value of the corresponding bare land cluster is determined according to the land value of the land group of the selected target land cluster.

[0092] Specifically, Figure 3 As shown, according to the pixel point information of each land use type in the candidate plot, the land use value of the candidate plot is determined, including:

[0093] S301, determining the supporting factors corresponding to the candidate plot according to the distribution of the pixel points corresponding to the buildings in the candidate plot.

[0094] It is understandable that the remaining bare land area within the candidate plot is indeed an important indicator, which represents the developable space of the plot. Generally speaking, the larger the bare land area, the stronger the plasticity of the plot and the higher the land value of the plot. However, if these bare land areas are scattered within the plot instead of being concentrated into one piece, the difficulty and cost of developing the plot will increase. Because the scattered bare land requires more infrastructure and transportation networks to connect, which will increase the complexity and investment of development.

[0095] The aggregation of surrounding functional points is also an important factor in determining the land value of a candidate plot. When the aggregation of functional points (such as suppliers, customers, public facilities, etc.) around the plot is high, the supporting maturity of the candidate plot is high, which usually means that the economic activities of the plot are frequent and the business environment is good. However, if the bare land area within the plot is scattered and the surrounding functional points are already highly aggregated, the multiple development and utilization of the plot will be limited. Because it is difficult for new development projects to find a suitable location to integrate into the existing functional layout, and they may also face greater social and economic resistance. Among them, the supporting factor can quantify the aggregation of functional points around the plot and the development potential within the plot.

[0096] The specific calculation method is as follows: First, it is necessary to identify and extract the building information within all industrial plots from high-resolution remote sensing images. This usually involves image classification technology, which uses algorithms to identify building pixels in the image and distinguish them from background pixels. Next, the buildings in the remote sensing image are marked as points of interest using POI data. POI data provides the geographic coordinates and related information of a specific location, such as building type, name, etc. By matching the buildings in the remote sensing image with the POI data, each building can be assigned a corresponding point of interest label, such as "store", "restaurant", "hotel", etc. Among all the marked points of interest, special attention should be paid to the factory-type points of interest, and they should be marked as core points. Core points represent the main production or processing facilities in the industrial plot and are an important part of the land value of the plot. Other types of points of interest, such as shops, restaurants, hotels, etc., are uniformly marked as functional points. Although these functional points contribute to the overall function of the plot, they are not core production facilities. When marking points of interest, it is necessary to recognize that the industrial planning area may contain multiple candidate plots of different types. Therefore, it is necessary to carefully distinguish different plots during the marking process and ensure that only buildings within the industrial plot are marked with points of interest. Finally, through comprehensive analysis of the POI marking results of industrial plots in remote sensing images, the functional layout and development potential of the plots can be evaluated.

[0097] Take the candidate plot as an industrial plot as an example to illustrate the calculation of supporting factors When: Select the core point in the candidate plot (if there are multiple, select one at random), and then calculate the average value of the Euclidean distance between the core point and its k neighboring function points. This average value represents the degree of aggregation of the function points around the candidate plot, that is, the supporting factor. The smaller the supporting factor, the more concentrated the function points around the candidate plot, and the higher the supporting maturity of the candidate plot; conversely, the larger the supporting factor, the more dispersed the function points around the candidate plot, and the lower the supporting maturity of the candidate plot.

[0098] When calculating the above k domain function points, the function points are sorted according to the distance from the core point according to the calculated distance. Then, the first k function points with the closest distance are selected as k neighborhood function points. The value of k is a pre-set parameter, which determines the size of the neighborhood range to be considered.

[0099] S302: Determine the unplanned factor corresponding to the candidate according to the pixel point information corresponding to the unplanned land use type in the candidate plot.

[0100] Specifically, the above-mentioned building area is recorded as a Class A area, the above-mentioned impermeable surface area is recorded as a Class B area, the above-mentioned bare land area is recorded as a Class C area, and the above-mentioned green area is recorded as a Class D area.

[0101] S303, determining the land use value of the candidate plot according to the pixel point information corresponding to the built type in the candidate plot, the pixel point information corresponding to the bare land type, the number of bare land connected areas, the unplanned factor and the supporting factor.

[0102] Specifically, Figure 4 As shown, according to the pixel point information corresponding to the built type in the candidate plot, the pixel point information corresponding to the bare land type, the number of bare land connected areas, the unplanned factor and the supporting factor, the land use value of the candidate plot is determined, including:

[0103] S401, determining the explicit land use value of the candidate plot according to the pixel point information corresponding to the built type in the candidate plot, the pixel point information corresponding to the bare land type, the number of bare land connected areas, the unplanned factor and the supporting factor.

[0104] S402, determining the implicit land use value of the candidate plots according to the regional shape parameters of the connected areas of each bare land in the candidate plots.

[0105] S403: Determine the land value of the candidate land parcel according to the explicit land value and implicit land value of the candidate land parcel.

[0106] Specifically, for each candidate plot, the total area of ​​pixels in the a, b, and c regions is calculated. This can be achieved through image classification technology, which classifies each pixel into the corresponding region type and counts the number of pixels in each region. The total area of ​​pixels in the a, b, and c regions is recorded as , , For the c-type region (bare land region), calculate the number of its connected domains. A connected domain refers to a region consisting of adjacent pixels of the same type. The number of connected domains in the c-type region is recorded as , matching factor The following formula (1) is used to calculate:

[0107] (1)

[0108] in, Represents the ratio of the total area of ​​type C pixels to the total area of ​​type A and B pixels. The larger the ratio, the larger the bare land area is compared to the used area, and the higher the land value is in theory. The actual land value of the bare land area is not only determined by the area, The larger the denominator of the value, the more dispersed the bare land is and the lower the actual land value of the land parcel is;

[0109] Represents the ratio of the greening distribution factor to the supporting factor. The larger the numerator, the more dispersed the greening distribution is, and the larger the denominator, the more mature the supporting facilities around the candidate plot. Therefore, the larger the ratio, the lower the actual land value of the plot. Therefore, the exponential function with the natural constant e as the base is used to normalize the ratio inversely;

[0110] therefore Represents the inefficiency weight, which is multiplied by the theoretical land value of the candidate plot , the land value R of each candidate plot is obtained; the candidate plot with higher land value has a higher potential evolution trend of the bare land within it.

[0111] Specifically, from each candidate plot cluster, a target plot cluster corresponding to the bare land cluster is selected, including: for each candidate plot cluster, the candidate plots corresponding to each bare land connected area in the candidate plot cluster are selected as matching plots; from each candidate plot cluster, the candidate plot cluster containing the largest number of matching plots is selected as the target plot cluster corresponding to the bare land cluster. The land value of the plot family corresponding to any candidate plot cluster is the normalized result of the average land value of all candidate plots contained in the candidate plot cluster, which is expressed by the formula: .

[0112] Among them, the candidate plot cluster refers to a group of plots with similar characteristics or attributes, which may have certain correlations in terms of geographical location, functional layout, development potential, etc.

[0113] For each candidate plot cluster, it is necessary to identify candidate plots corresponding to the bare land connected areas within it, which are called matching plots. The selection of matching plots is based on factors such as the distribution and area of ​​bare land areas within the plot and their relative position to other functional areas.

[0114] After determining the matching plots in each candidate plot cluster, it is necessary to count the number of matching plots contained in each cluster. From all candidate plot clusters, the cluster containing the largest number of matching plots is selected as the target plot cluster that best matches the bare land cluster. This selection process is based on the advantage of quantity, that is, the plot cluster containing more matching plots is considered to have higher similarity or complementarity with the bare land cluster in terms of functional layout, development potential, etc.

[0115] Exemplarily, the candidate plots where all the bare land connected domains are located in each bare land cluster are obtained. A candidate plot cluster containing the most of these candidate plots is found as the target plot cluster. The bare land cluster and the target plot cluster are regarded as a set of matching clusters.

[0116] Furthermore, the bare land value of the bare land cluster is determined according to the land value of the plot family of the target plot cluster, including: determining the overlapping area between the bare land cluster and the target plot cluster; and determining the bare land value of the bare land cluster according to the overlapping area and the land value of the plot family of the target plot cluster.

[0117] Specifically, the bare land value of the bare land cluster may be determined by proportional calculation according to the overlapping area and the land value of the land parcel family of the target land parcel cluster.

[0118] The specific calculation method may be to distribute the land use value of the target land cluster to the bare land cluster according to the area ratio, so as to obtain the bare land value of the bare land cluster. .

[0119] For example, the average functional value of all plots in each candidate plot cluster is obtained. , the following is used to calculate the bare land value of the bare land cluster: :

[0120] (2)

[0121] Where v represents the vth bare land cluster, r represents the rth land parcel cluster (target land parcel cluster) matched by the vth bare land cluster, represents the overlapping area between the vth bare land cluster and the rth land plot cluster (i.e. the overlapping area mentioned above), represents the area of ​​the vth bare land cluster; represents the bare land value of the vth bare land cluster; Represents the ratio of the overlapping area to the area of ​​the vth bare land cluster. The larger the value, the greater the encirclement of the vth bare land cluster. Represents the land use value of the plot family of the rth plot cluster. The larger the value, the higher the available value of the plot cluster. The larger the encirclement degree and the larger the available value, the higher the actual functional value of the contiguous bare land of the vth bare land cluster after the verification of the plot functional value. Therefore, the actual utilization rate of the contiguous bare land represented by the bare land cluster is higher. The bare land land value of all bare land clusters is obtained. According to the bare land land value of the bare land cluster, the real inefficient industrial land can be screened out, and its evolution can be observed.

[0122] It is understandable that each factory may make temporary and simple use of the bare land area within its industrial land. This use is for the bare land within the factory's own internal planning plot, such as temporarily stacking excess and discarded materials. Due to the different types, properties, uses, and quantities of materials, not all materials need to be built in separate warehouses. It may not be necessary to build a separate warehouse or open ground just for these material areas. Although the value of the plot may not be reflected on the surface, these bare lands do provide certain conveniences for nearby industrial areas. Temporary use meets the hidden needs of the factory, which is the same as the actual effect achieved by the use after planning and construction. Some inefficient industrial land plays its hidden value in certain periods. In order to accurately evaluate the actual evolution trend of the industrial planning area, it is necessary to increase the evolution rate of these plots with hidden value in the short term. This application proposes to calculate the implicit land value of each bare land cluster. The bare land area with hidden needs is relatively regular, which means that the user has goals and plans for the use of the plot.

[0123] Specifically, Figure 5 As shown, the regional shape parameters include the edge line of the bare land connected area and the slope of the edge line; according to the explicit land value and implicit land value of the candidate plot, the land value of the candidate plot is determined, including:

[0124] S501, determining each edge line type and the distribution probability of each edge line type according to the area shape parameters of each bare land connected area in the candidate plot.

[0125] Specifically, the edge lines of each bare land connected area are analyzed to determine their types (such as straight lines, curves, etc.) and the distribution probability of each edge line type is calculated, that is, the frequency of each edge line type appearing in the candidate plots.

[0126] S502, using the gray level co-occurrence matrix, traverse each bare land connected region to obtain the distribution entropy corresponding to each bare land connected region.

[0127] The gray-level co-occurrence matrix is ​​used to traverse each bare ground connected region, and the distribution entropy corresponding to each bare ground connected region is calculated, which is a quantitative indicator reflecting the complexity of regional texture.

[0128] S503: Determine the implicit land use value of the candidate land parcel according to the distribution probability of each edge line type and the distribution entropy corresponding to each bare land connected area.

[0129] The distribution probability and distribution entropy of edge line types are weighted summed or other appropriate comprehensive methods are adopted. The weight distribution of the distribution probability and distribution entropy of edge line types in implicit land use value is considered.

[0130] For example, calculate the implicit land value of the candidate plot , you can use the following steps:

[0131] 1) Each bare land cluster has several candidate plots where all the bare land connected domains are located, and Canny edge detection is performed on the bare land area in each candidate plot to obtain several edge lines in the bare land area in each candidate plot;

[0132] Obtain the average slope of all edge lines in the bare land area of ​​each candidate plot, classify all edge lines according to the average slope, classify those with the same average slope into one class, and obtain the distribution probability of each class of edge lines;

[0133] Calculate the kurtosis of the probability distribution of all types of edge lines (i.e., the kurtosis of the marginal distribution ), the smaller the kurtosis, the closer the marginal distribution probability of all slopes is, and there is no obvious peak, which is consistent with the characteristic that when the placement is unified, each type of placement direction has a high repetition rate; the kurtosis is normalized and recorded as J;

[0134] 2) Use the gray-level co-occurrence matrix to traverse each bare land area in the three directions of 0, 45, and 90 degrees to obtain three co-occurrence matrices. The three matrices are added together to obtain the target gray-level co-occurrence matrix;

[0135] According to the distribution entropy (ENT) of all elements in the target gray-level co-occurrence matrix of each bare land area, that is, the distribution entropy of the co-occurrence matrix The larger the entropy value, the more complex the grayscale distribution of the bare land area is. Conversely, the smaller the entropy value, the more regular the grayscale distribution of the bare land area is. The grayscale co-occurrence matrix can be used to quickly and concisely evaluate the regularity of the bare land area. The entropy value is normalized and recorded as H.

[0136] 3) Obtaining implicit land value , using the following formula (3):

[0137] (3)

[0138] Where v represents the vth bare land cluster, n represents the bare land connected area within the nth candidate plot contained in the bare land cluster, Represents the number of all candidate plots contained in the vth bare land cluster;

[0139] The product of the normalized marginal distribution kurtosis of the nth candidate plot in the vth bare land cluster and the distribution entropy of the co-occurrence matrix is ​​represented by a constant 1, which is different from the product and is used to represent the regularity of the bare land area of ​​the nth candidate plot in the vth bare land cluster. The regularity of the bare land areas of all candidate plots in the vth bare land cluster is averaged to obtain the implicit land use value of the vth bare land cluster.

[0140] In an exemplary embodiment, if the target images at different moments include a target image at a first moment and a target image at a second moment, and the second moment is later than the first moment, the land evolution rate of the target land is determined according to the bare land value of each bare land cluster in each target image, including: determining the value weight according to the bare land value of each bare land cluster in the target image at the first moment and the bare land value of each bare land cluster in the second target image; determining the land evolution rate of the target land according to the pixel point information corresponding to each bare land cluster in the target image at the first moment, the pixel point information corresponding to each bare land cluster in the target image at the second moment and the value weight.

[0141] For example, the land parcels in the classified utilization image are reduced in the observation scale, and the actual evolution of the land parcels still needs to be obtained from the original high-resolution remote sensing image; the high-resolution remote sensing image of the industrial planning area is collected once a month; the land use evolution rate of all the bare land cluster areas in the target image collected for the first time in the same area position in the high-resolution remote sensing image collected for the second time is calculated. , calculated using the following formula (4):

[0142] (4)

[0143] Where v represents the vth bare ground cluster in the tth collected remote sensing image, N represents the number of bare ground clusters, represents the area of ​​Class C (bare land) within the vth bare land cluster collected at the tth time, Represents the reduction in the area of ​​the c-type region of the v-th bare land cluster in the remote sensing image collected in the t+1th time compared with the last time; the larger the value, the faster the evolution rate of the inefficient industrial land in the v-th bare land cluster in the remote sensing images collected in two adjacent times;

[0144] represents the sum of the contiguous utilization rate and hidden utilization rate of the vth bare land cluster. A bare land cluster must have relatively low utilization rates to be considered a contiguous inefficient candidate plot. Therefore As the inefficiency weight, it is multiplied by the evolution rate of inefficient industrial land in the vth bare land cluster, and then the weighted average is taken for all bare land clusters to obtain the evolution rate of inefficient industrial land in the entire industrial planning area compared with the previous collection in the t+1th time.

[0145] Then, the evolution rate of inefficient industrial land for each remote sensing image acquisition compared with the last acquisition can be obtained, and projected on a two-dimensional plane to generate an evolution curve, with the horizontal axis representing the number of acquisitions and the vertical axis representing the evolution rate.

[0146] The evolution curve can be used to evaluate, observe and model the changes in contiguous inefficient industrial land in the target area.

[0147] In this embodiment, candidate plots are classified according to their functional values, which are then used to verify the contiguous utilization rate of bare land areas connected by distance, and then the hidden utilization rate of bare land areas is explored to jointly correct the evolution rate of inefficient industrial land in the industrial planning area, obtain an accurate evolution rate curve, and provide data support for subsequent plot planning and evaluation.

[0148] In one embodiment, a remote sensing image-based contiguous inefficient industrial land change detection system is provided, comprising:

[0149] An acquisition module is used to acquire each target image collected at different times for the target land;

[0150] A pixel extraction module is used to identify, for each target image, each pixel corresponding to each land use type in the target image and the pixel information of each pixel; wherein the land use type includes a built type, a bare land type and an unplanned type; and the pixel information includes the number of pixels and the position of the pixels;

[0151] A clustering module, used for combining the pixels of the bare land type in the target image to obtain at least one developable bare land cluster;

[0152] A land use evaluation module, used to determine the bare land value of each bare land cluster in the target image according to the pixel point information of each land use type in each bare land cluster;

[0153] The evolution determination module is used to determine the land use evolution rate of the target land according to the bare land value of each bare land cluster in each target image.

[0154] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0155] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A method for detecting changes in contiguous inefficient industrial land based on remote sensing images, characterized in that: The method comprises: Acquire target images collected at different times for the target land; For each target image, identify each pixel corresponding to each land use type in the target image and the pixel information of each pixel; wherein the land use type includes a built type, a bare land type and an unplanned type; and the pixel information includes the number of pixels and the position of the pixels; Combining the pixels of the bare land type in the target image to obtain at least one developable bare land cluster; Determining the bare land value of each bare land cluster in the target image according to the pixel point information of each land use type in each bare land cluster; Determining a land use evolution rate of the target land according to the bare land value of each bare land cluster in each target image; The step of combining the bare land type pixel points in the target image to obtain at least one developable bare land cluster includes: Generate at least one bare ground connected region according to each pixel point of the bare ground type in the target image and a preset number of connections; Perform cluster analysis on the connected areas of each bare land to obtain at least one developable bare land cluster; Determining the bare land value of each bare land cluster in the target image according to the pixel point information of each land use type in each bare land cluster includes: Dividing the target image into a plurality of candidate plots; For each candidate land parcel, determining the land use value of the candidate land parcel according to the pixel point information corresponding to each land use type in the candidate land parcel; Performing cluster analysis on each candidate land parcel according to the land use value of each candidate land parcel and the distribution position of each candidate land parcel in the target image to obtain at least one candidate land parcel cluster; For each bare land cluster, select a target land cluster corresponding to the bare land cluster from each candidate land cluster; Determine the bare land value of the bare land cluster according to the land value of the land group of the target land cluster; the land value of the land group corresponding to any candidate land cluster is the normalized result of the average land value of all candidate lands contained in the candidate land cluster; The determining the bare land value of the bare land cluster according to the land value of the land parcel group of the target land parcel cluster includes: Determining the overlap area between the bare land cluster and the target land cluster; Determining the bare land value of the bare land cluster according to the overlapped area and the land value of the land parcel group of the target land parcel cluster; If the target images at different times include a target image at a first time and a target image at a second time, and the second time is later than the first time, then determining the land use evolution rate of the target land according to the bare land value of each bare land cluster in each target image includes: Determining a value weight according to the bare land value of each bare land cluster in the target image at the first moment and the bare land value of each bare land cluster in the target image at the second moment; The land use evolution rate of the target land is determined according to the pixel point information corresponding to each bare land cluster in the target image at the first moment, the pixel point information corresponding to each bare land cluster in the target image at the second moment, and the value weight.

2. The method for detecting changes in contiguous inefficient industrial land based on remote sensing images according to claim 1 is characterized in that: The determining the land use value of the candidate land parcel according to the pixel point information of each land use type in the candidate land parcel comprises: Determine the supporting factor corresponding to the candidate plot according to the distribution of pixel points corresponding to the building in the candidate plot; Determine the unplanned factor corresponding to the candidate plot according to the pixel point information corresponding to the unplanned land type in the candidate plot; The land use value of the candidate plot is determined based on the pixel point information corresponding to the built type in the candidate plot, the pixel point information corresponding to the bare land type, the number of connected areas of the bare land, the unplanned factor and the supporting factor.

3. The method for detecting changes in contiguous inefficient industrial land based on remote sensing images according to claim 2 is characterized in that: The step of selecting a target land parcel cluster corresponding to the bare land cluster from among the candidate land parcel clusters includes: For each candidate plot cluster, the candidate plots corresponding to each bare land connected area in the candidate plot cluster are used as matching plots; From the candidate plot clusters, a candidate plot cluster containing the largest number of matching plots is selected as the target plot cluster corresponding to the bare land cluster.

4. The method for detecting changes in contiguous inefficient industrial land based on remote sensing images according to claim 2 is characterized in that: The determining of the land use value of the candidate plot according to the pixel point information corresponding to the built type in the candidate plot, the pixel point information corresponding to the bare land type, the number of the bare land connected areas, the unplanned factor and the supporting factor includes: Determine the explicit land value of the candidate land parcel according to the pixel point information corresponding to the built type in the candidate land parcel, the pixel point information corresponding to the bare land type, the number of connected areas of the bare land, the unplanned factor and the supporting factor; Determining the implicit land use value of the candidate land parcel according to the regional shape parameters of the connected areas of each bare land in the candidate land parcel; The land value of the candidate land parcel is determined according to the explicit land value and implicit land value of the candidate land parcel.

5. The method for detecting changes in contiguous inefficient industrial land based on remote sensing images according to claim 4 is characterized in that: The regional shape parameters include the edge line of the bare land connected area and the slope of the edge line; the determining the land value of the candidate land according to the explicit land value and the implicit land value of the candidate land includes: Determining each edge line type and the distribution probability of each edge line type according to the regional shape parameters of each bare land connected area in the candidate plot; Using the gray-level co-occurrence matrix, each bare land connected area is traversed to obtain the distribution entropy corresponding to each bare land connected area; The implicit land use value of the candidate plot is determined according to the distribution probability of each edge line type and the distribution entropy corresponding to each bare land connected area.

6. A contiguous inefficient industrial land change detection system based on remote sensing images, characterized in that: The system comprises: An acquisition module, used to acquire each target image collected at different times for the target land; A pixel extraction module is used to identify, for each target image, each pixel corresponding to each land use type in the target image and the pixel information of each pixel; wherein the land use type includes a built type, a bare land type and an unplanned type; and the pixel information includes the number of pixels and the position of the pixels; A clustering module, used for combining the pixels of the bare land type in the target image to obtain at least one developable bare land cluster; A land use evaluation module, used to determine the bare land value of each bare land cluster in the target image according to the pixel point information of each land use type in each bare land cluster; An evolution determination module, used for determining a land use evolution rate of the target land use according to the bare land value of each bare land cluster in each target image; The step of combining the bare land type pixel points in the target image to obtain at least one developable bare land cluster includes: Generate at least one bare ground connected region according to each pixel point of the bare ground type in the target image and a preset number of connections; Perform cluster analysis on the connected areas of each bare land to obtain at least one developable bare land cluster; Determining the bare land value of each bare land cluster in the target image according to the pixel point information of each land use type in each bare land cluster includes: Dividing the target image into a plurality of candidate plots; For each candidate land parcel, determining the land use value of the candidate land parcel according to the pixel point information corresponding to each land use type in the candidate land parcel; Performing cluster analysis on each candidate land parcel according to the land use value of each candidate land parcel and the distribution position of each candidate land parcel in the target image to obtain at least one candidate land parcel cluster; For each bare land cluster, select a target land cluster corresponding to the bare land cluster from each candidate land cluster; Determine the bare land value of the bare land cluster according to the land value of the land group of the target land cluster; the land value of the land group corresponding to any candidate land cluster is the normalized result of the average land value of all candidate lands contained in the candidate land cluster; The determining the bare land value of the bare land cluster according to the land value of the land parcel group of the target land parcel cluster includes: Determining the overlap area between the bare land cluster and the target land cluster; Determining the bare land value of the bare land cluster according to the overlapped area and the land value of the land parcel group of the target land parcel cluster; If the target images at different times include a target image at a first time and a target image at a second time, and the second time is later than the first time, then determining the land use evolution rate of the target land according to the bare land value of each bare land cluster in each target image includes: Determining a value weight according to the bare land value of each bare land cluster in the target image at the first moment and the bare land value of each bare land cluster in the target image at the second moment; The land use evolution rate of the target land is determined according to the pixel point information corresponding to each bare land cluster in the target image at the first moment, the pixel point information corresponding to each bare land cluster in the target image at the second moment, and the value weight.