Urban multi-center extraction method and system based on night light remote sensing image

Abstract

The method and system for extracting multiple urban centers based on nighttime light remote sensing images belong to the technical field of remote sensing image processing and analysis. The present invention solves the problem that the existing remote sensing image data analysis method mainly stays at the analysis of the central position, and does not analyze the influence of each urban center on the surrounding area. The method comprises the following steps: step S1, reading the urban night light image and saving it as an image matrix I, removing the brightness area with a pixel area of ​​≤4*4 in the image matrix I; step S2, obtaining the maximum neighborhood matrix of the local pixel maximum value of the image matrix I with the brightness area removed, defining the peak matrix using the maximum domain matrix, and obtaining the number of peaks; step S3, determining the number of sub-models in the Gaussian mixture model using the number of peaks, wherein each pixel point of the image matrix I with a brightness area of ​​≤4*4 removed in step S1 is sorted as training data and input into the model, and the center point and the coverage range of the center point after training are extracted as the extracted urban center.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing image processing and analysis, and in particular to a method and system for extracting urban multi-centers based on nighttime light remote sensing images. Background Art

[0002] With the acceleration of urbanization, the spatial agglomeration effect of urban elements has been continuously strengthened, and the urban structure has gradually evolved. Some large cities have developed a multi-center structure, that is, from the initial "single center" to a "multi-center" or more decentralized urban spatial structure. The multi-center urban development model can effectively improve the economic level and social cohesion, alleviate population congestion, and improve the urban environment. The study of the multi-center structure of the city is of great significance to the sustainable development of the city, monitoring the urban center, and analyzing the dynamic pattern of the urban center at different spatial scales. With the development and application of geographic information technology and remote sensing satellite technology, remote sensing technology products such as night light data have begun to be used to identify urban spatial structure. Based on night light data, the overall structure of the city is analyzed by virtue of its large spatial scale advantage.

[0003] At present, the research on urban spatial structure uses remote sensing image methods, mainly combining remote sensing images and other data to achieve comprehensive analysis. Using remote sensing image data to analyze urban structure is more convenient and faster, and the data is more objective, but due to the low resolution of open source night light image data, the defects of using current remote sensing image methods are: current research mainly stays in the analysis of the central location, and does not analyze the impact of each city center on the surrounding areas.

[0004] In summary, it is more convenient and faster to analyze urban structure using existing remote sensing image data, and the data is more objective. However, due to the low resolution of open source night light image data, the current research method mainly focuses on the analysis of central locations, and does not analyze the impact of each city center on the surrounding areas.

[0005] Therefore, in the prior art, the analysis of urban structure based on existing remote sensing image data mainly stays at the center, and the impact of each urban center on the surrounding areas is not analyzed. For example, Chen et al. published an article titled "A New Approach for Detecting Urban Centers and Their Spatial Structure With Nighttime Light Remote Sensing" in 2017 in IEEE Transactions on Geoscience and Remote Sensing. They proposed a new method for identifying urban spatial structure, the local contour tree algorithm, by combining remote sensing image data with night light images, and demonstrated the feasibility of the method through empirical evidence. The article demonstrated the feasibility of the method through empirical evidence, but did not clearly give an analysis of the analysis of urban structure based on remote sensing image data mainly staying at the center, and did not propose a solution to the problem of analyzing the impact of each urban center on the surrounding areas. In 2021, He et al. published an article titled "Evaluation of Polycentric Spatial Structure in the Urban Agglomeration of the Pearl River Delta (PRD) Based on Multi-Source Big Data Fusion" in Remote Sens, proposing a new method to identify the polycentric spatial structure of the Pearl River Delta urban agglomeration based on the fusion of night light data, point of interest data and Tencent migration data. The urban polycentric identification accuracy calculated by this method is 92.84%, and the Kappa value is 0.8971, which is higher than the polycentric identification result obtained by the traditional relative threshold. This article solves the problem of low accuracy in urban polycentric identification, and its results cannot provide an analysis of the urban structure mainly staying in the central position for remote sensing image data analysis, and does not propose improvement directions or degrees of improvement for the impact of each urban center on the surrounding areas. Summary of the invention

[0006] The present invention solves the problem that the existing remote sensing image data analysis method mainly stays at the analysis of the central position, and does not analyze the influence of each city center on the surrounding areas.

[0007] The method for extracting urban multi-centers based on nighttime light remote sensing images of the present invention comprises the following steps:

[0008] Step S1, reading the city night light image and saving it as an image matrix I, removing the brightness area with a pixel area ≤ 4*4 in the image matrix I;

[0009] Step S2, obtaining a maximum neighborhood matrix of the local pixel maximum value of the image matrix I with the brightness area removed, defining a peak matrix using the maximum neighborhood matrix, and obtaining the number of peaks;

[0010] Step S3, using the number of peaks to determine the number of sub-models in the Gaussian mixture model, the step S1 removes the image matrix I of the brightness area with a pixel area ≤ 4*4, and organizes each pixel point as training data and inputs it into the model, extracting the trained center point and the center point coverage range as the extracted city center.

[0011] Further, in one embodiment of the present invention, in step S1, removing the brightness area with a pixel area ≤ 4*4 in the image matrix I comprises the following steps:

[0012] Step S101, using kernel to perform morphological opening operation on the image matrix I to remove the brightness area in the image matrix I;

[0013] The kernel is:

[0014]

[0015] Represents the morphological opening structure element.

[0016] Further, in one embodiment of the present invention, in step S2, the step of obtaining the maximum neighborhood matrix of the local pixel maximum value of the image matrix without the brightness area comprises the following steps:

[0017] Step S201, the maximum neighborhood matrix is ​​I out , initialized to I out =I*0;

[0018] Step S202, set a 3*3 sliding window

[0019]

[0020] Where (i, j) is the center point of the sliding window;

[0021] Step S203, after initialization, out The expanded matrix is

[0022] Step S204, using a sliding window to traverse the expanded matrix I fill For each pixel in, calculate I out (i-1, j-1) = max(window(i, j)), i∈(1, r), j∈(1, c), r represents the number of rows of the image matrix, and c represents the number of columns of the image matrix.

[0023] Further, in one embodiment of the present invention, in step S2, the method for obtaining the number of peaks comprises the following steps:

[0024] Step S205, the peak matrix is ​​I diff =I out -I;

[0025] Step S206, statistics I diff The number of points K whose median value is 0 is the number of peaks.

[0026] Further, in one embodiment of the present invention, in step S3, the process of determining the number of sub-models in the Gaussian mixture model by using the number of peaks includes the following steps:

[0027] Step S301: Gaussian mixture model

[0028]

[0029] Where K is the number of peaks, which also represents the number of sub-models in the Gaussian mixture model.

[0030]

[0031] represents the probability of the kth submodel in the Gaussian mixture model, φ(x|θ k ) represents the Gaussian distribution density function of the kth sub-model, and θ k =(μ k , σ k ).

[0032] Further, in one embodiment of the present invention, in step S3, the process of organizing each pixel point of the image matrix into training data and inputting it into the model includes the following steps:

[0033] Step S302, organize each pixel of the image matrix I into training data with three attributes, expressed as (index_x, index_y, I(i, j)), where index_x represents the row number where the pixel is located, index_y represents the column number where the pixel is located, and the image matrix I is organized into r*c training data, and then the training data is input into the Gaussian mixture model for training.

[0034] Further, in one embodiment of the present invention, in step S3, the process of extracting the trained center point and the center point coverage as the extracted city center includes the following steps:

[0035] Step S303, extracting each sub-model parameter of the trained Gaussian mixture model

[0036] (α k, μ k1 , μ k2 , σ k1 , σ k2 ),

[0037] In the formula (μ k1 , μ k2 ) represents the position of the kth center in the night light image;

[0038] Step S304, use the trained Gaussian mixture model to mark the image matrix I, obtain the sub-model category to which each pixel belongs, and then count the number of pixels according to the category to obtain the number of pixels covered by each sub-model. The number of pixels is the city center point, and the number of covered pixels is the coverage range of the city center point.

[0039] The urban multi-center extraction system based on nighttime light remote sensing images of the present invention comprises the following modules:

[0040] Module S1, used for reading the city night light image and saving it as an image matrix I, and removing the brightness area specified in the image matrix I;

[0041] Module S2 is used to obtain the maximum neighborhood matrix of the local pixel maximum value of the image matrix I after removing the brightness area and the abnormal points, and define the peak matrix using the maximum neighborhood matrix to obtain the number of peaks;

[0042] Module S3 is used to determine the number of sub-models in the Gaussian mixture model using the number of peaks. The module S1 removes the image matrix I with brightness areas whose pixel area is ≤4*4, organizes each pixel point into training data and inputs it into the model, extracts the center point after training and the coverage range of the center point as the extracted city center.

[0043] Furthermore, in one embodiment of the present invention, the urban multi-center extraction system based on nighttime light remote sensing images, in the module S1, the brightness area and abnormal points specified by the image matrix are removed, and the system includes the following modules:

[0044] Module S101, used for performing morphological opening operation on the image matrix I by using kernel to remove the brightness area in the image matrix I;

[0045] The kernel is:

[0046]

[0047] Represents the morphological opening structure element.

[0048] The computer-readable storage medium described in the present invention stores a computer program thereon, and when the computer program is executed by a processor, the steps of the method described in any one of the above methods are implemented.

[0049] A computer device described in the present invention includes a memory and a processor, wherein a computer program is stored in the memory, and when the processor runs the computer program stored in the memory, the steps of the method described in any one of the above methods are executed.

[0050] The present invention solves the problem that the existing remote sensing image data analysis method mainly stays at the analysis of the central location, and does not analyze the impact of each city center on the surrounding areas. Specific beneficial effects include:

[0051] 1. The present invention considers the night light distribution of the city center as Gaussian distribution and uses Gaussian mixture model to fit the night light distribution of the target area. In the fitted Gaussian mixture model, the center point of each sub-model represents the city center, and the coverage of each sub-model represents the influence range of each center. This extraction method can not only identify multiple centers of the city, but also represent the influence range of each center.

[0052] 2. The present invention calculates the coverage of each center through Gaussian mixture model parameters. The result can be used for horizontal comparison to analyze the differences between the centers, and can also be used for vertical comparison of multi-phase data to analyze the development direction of each urban center and the expansion of the affected area.

[0053] The present invention is applicable to the field of night light remote sensing image recognition and analysis technology. In view of the fact that night light brightness has a strong correlation with human activities, the present invention analyzes urban structure based on night light remote sensing data and solves the technical problem of identifying urban multi-center structure. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0055] Figure 1 It is a flow chart of the method and system for extracting urban multi-centers based on nighttime light remote sensing images described in the present invention.

[0056] Figure 2 The remote sensing image of night lights in Guangzhou in January 2020 as described in Implementation Example 8.

[0057] Figure 3 Light distribution diagram in the three-dimensional coordinate system described in Implementation Example 8.

[0058] Figure 4 A three-dimensional graph of the fitted Gaussian mixture model described in Implementation Example 8. DETAILED DESCRIPTION

[0059] The following will clearly and completely describe various embodiments of the present invention in conjunction with the accompanying drawings. The embodiments described with reference to the accompanying drawings are exemplary and intended to be used to explain the present invention, but should not be understood as limiting the present invention.

[0060] Implementation method 1: The method for extracting urban multi-centers based on nighttime light remote sensing images described in this implementation method is as follows: Figure 1 This embodiment can be better understood by including the following steps:

[0061] Step S1, reading the city night light image and saving it as an image matrix I, removing the brightness area with a pixel area ≤ 4*4 in the image matrix I;

[0062] Step S2, obtaining a maximum neighborhood matrix of the local pixel maximum value of the image matrix I with the brightness area removed, defining a peak matrix using the maximum neighborhood matrix, and obtaining the number of peaks;

[0063] Step S3, using the number of peaks to determine the number of sub-models in the Gaussian mixture model, the step S1 removes the image matrix I of the brightness area with a pixel area ≤ 4*4, and organizes each pixel point as training data and inputs it into the model, extracting the trained center point and the center point coverage range as the extracted city center.

[0064] In this implementation, in step S1, the night light remote sensing image is cropped according to the vector data of the target area, and the cropped night light remote sensing image is read and saved as an image matrix. The image matrix is ​​defined as I, and the matrix size is r*c, where r represents the number of rows of the image matrix, c represents the number of columns of the image matrix, and I(i, j) represents the pixel value of the pixel located at the i row and j column.

[0065] Specifically, considering the continuous brightness area in the night light image as the central area will affect the recognition result. Therefore, the brightness area with a pixel area ≤ 4*4 and abnormal points are removed to eliminate the interference of the fitting model.

[0066] Implementation method 2: This implementation method further limits the urban multi-center extraction method based on nighttime light remote sensing images described in implementation method 1. In this implementation method, in step S1, removing the brightness area with a pixel area ≤ 4*4 in the image matrix I includes the following steps:

[0067] Step S101, using kernel to perform morphological opening operation on the image matrix I to remove the brightness area in the image matrix I;

[0068] The kernel is:

[0069]

[0070] Represents the morphological opening structure element.

[0071] In this implementation, the non-brightness area of ​​the image matrix I is 0, and the brightness area is greater than 0, that is, the higher the brightness, the larger the pixel value. After the morphological opening operation, the brightness area with a brightness area ≤ 4*4 can be removed.

[0072] Implementation method 3: This implementation method further limits the urban multi-center extraction method based on nighttime light remote sensing images described in implementation method 1. In this implementation method, in step S2, the maximum neighborhood matrix of the local pixel maximum value of the image matrix without the brightness area is obtained, including the following steps:

[0073] Step S201, the maximum neighborhood matrix is ​​I out And initialize, I out =I*0;

[0074] Step S202, set a 3*3 sliding window

[0075]

[0076] Where (i, j) is the center point of the sliding window;

[0077] Step S203, after initialization, out The expanded matrix is

[0078] Step S204, using a sliding window to traverse the expanded matrix I fill For each pixel in, calculate I out (i-1, j-1) = max(window(i, j)), i∈(1, r), j∈(1, c), r represents the number of rows of the image matrix, and c represents the number of columns of the image matrix.

[0079] In this implementation, a sliding window is used to replace the pixel value at the center of each window with the maximum pixel value in the window, and finally a maximum neighborhood matrix recording the local maximum pixel value can be obtained. out I is the same as I, which is a pixel point with a matrix size of r*c. out The center point (i, j) is the maximum value of all windows.

[0080] Embodiment 4: This embodiment further limits the method for extracting urban multi-centers based on nighttime light remote sensing images described in Embodiment 1. In this embodiment, in step S2, the method for obtaining the number of peaks includes the following steps:

[0081] Step S205, the peak matrix is ​​I diff =Iout -I;

[0082] Step S206, statistics I diff The number of points K whose median value is 0 is the number of peaks.

[0083] In this embodiment, in the night light image, the city center often appears as the brightest area locally, and the light brightness of each city center and nearby areas presents a peak-like shape, with the peak corresponding to the city center. Based on the maximum neighborhood matrix for finding the maximum value of pixels in the local area, the number of centers is solved.

[0084] The method used to calculate the number of peaks is the difference algorithm, because this algorithm is relatively simple and convenient to solve the complex problem of calculating local maximum values ​​and multiple variables.

[0085] Implementation mode 5: This implementation mode further limits the urban multi-center extraction method based on nighttime light remote sensing images described in implementation mode 1. In this implementation mode, in step S3, the process of determining the number of sub-models in the Gaussian mixture model by using the number of peaks includes the following steps:

[0086] Step S301: Gaussian mixture model

[0087]

[0088] Where K is the number of peaks, which also represents the number of sub-models in the Gaussian mixture model.

[0089]

[0090] represents the probability of the kth submodel in the Gaussian mixture model, φ(x|θ k ) represents the Gaussian distribution density function of the kth sub-model, and θ k =(μ k , σ k ).

[0091] In this implementation, the Gaussian mixture model is an existing model, and the determined number of peaks K represents the number of sub-models in the Gaussian mixture model.

[0092] Implementation 6. This implementation further defines the method for extracting urban multi-centers based on nighttime light remote sensing images described in Implementation 1. In this implementation, in step S3, the process of arranging each pixel point of the image matrix as training data and inputting it into the model includes the following steps:

[0093] Step S302, organize each pixel of the image matrix I into training data with three attributes, expressed as (index_x, index_y, I(i, j)), where index_x represents the row number where the pixel is located, index_y represents the column number where the pixel is located, and the image matrix I is organized into r*c training data, and then the training data is input into the Gaussian mixture model for training.

[0094] Implementation 7. This implementation further limits the method for extracting multiple city centers based on nighttime light remote sensing images described in Implementation 1. In this implementation, in step S3, the process of extracting the trained center point and the center point coverage as the extracted city center includes the following steps:

[0095] Step S303, extracting each sub-model parameter of the trained Gaussian mixture model

[0096] (α k , μ k1 , μ k2 , σ k1 , σ k2 ),

[0097] In the formula (μ k1 , μ k2 ) represents the position of the kth center in the night light image;

[0098] Step S304, use the trained Gaussian mixture model to mark the image matrix I, obtain the sub-model category to which each pixel belongs, and then count the number of pixels according to the category to obtain the number of pixels covered by each sub-model. The number of pixels is the city center point, and the number of covered pixels is the coverage range of the city center point.

[0099] In this implementation, in the fitted Gaussian mixture model, the center point of each sub-model represents the city center, and the coverage of each sub-model represents the influence range of each center.

[0100] Implementation 8: This implementation is based on the city multi-center extraction method based on nighttime light remote sensing images described in the present invention, and provides a practical implementation in combination with specific objects:

[0101] Figure 2 This is a remote sensing image of night light in Guangzhou in January 2020. Now consider the pixel subscript as the horizontal and vertical coordinates, and the pixel value as the height, and we can get the following: Figure 3 The three-dimensional image of the night light image shown contains multiple peaks, each of which corresponds to Figure 2 A local brighter area in the original night light image. In the original night light image, the location with the larger brightness value is Figure 3 The higher the height of the medium wave peak.

[0102] Figure 4 This is a 3D image of the Gaussian mixture model after fitting the light distribution. Figure 4 The light distribution of the original night light image is basically fitted, and the positions of the peaks and troughs are Figure 3 It is basically the same and smoother, which ensures that while obtaining the location of the city center, the influence range of each center can also be calculated.

[0103] As mentioned above, the urban multi-center extraction method based on night light remote sensing images and Gaussian mixture models, on the basis of identifying urban multi-centers, represents the influence range of each center. Analyzing the coverage or influence range of urban centers and the spatial expansion characteristics is also an important aspect of characterizing urban structure and development.

[0104] Implementation method 9: The urban multi-center extraction system based on nighttime light remote sensing images described in this implementation method includes the following modules:

[0105] Module S1, used for reading the city night light image and saving it as an image matrix I, and removing the brightness area specified in the image matrix I;

[0106] Module S2 is used to obtain the maximum neighborhood matrix of the local pixel maximum value of the image matrix I after removing the brightness area and the abnormal points, and define the peak matrix using the maximum neighborhood matrix to obtain the number of peaks;

[0107] Module S3 is used to determine the number of sub-models in the Gaussian mixture model using the number of peaks. The module S1 removes the image matrix I with brightness areas whose pixel area is ≤4*4, organizes each pixel point into training data and inputs it into the model, extracts the center point after training and the coverage range of the center point as the extracted city center.

[0108] Embodiment 10. This embodiment further defines the urban multi-center extraction system based on nighttime light remote sensing images described in Embodiment 9. In the module S1, the brightness area and abnormal points specified by the image matrix are removed. The system includes the following modules:

[0109] Module S101, used for performing morphological opening operation on the image matrix I by using kernel to remove the brightness area in the image matrix I;

[0110] The kernel is:

[0111]

[0112] Represents the morphological opening structure element.

[0113] Embodiment eleven: A computer-readable storage medium described in this embodiment stores a computer program, and when the computer program is executed by a processor, the steps of the method described in any one of embodiments one to seven are implemented.

[0114] Embodiment 12: A computer device described in this embodiment includes a memory and a processor, wherein a computer program is stored in the memory, and when the processor runs the computer program stored in the memory, the steps of the method described in any one of embodiments 1 to 7 are executed.

Claims

1. A method for extracting urban multi-centers based on nighttime light remote sensing images, characterized in that: The following steps are involved: Step S1, reading the city night light image and saving it as an image matrix I, removing the brightness area with a pixel area ≤ 4*4 in the image matrix I; Step S2, obtaining a maximum neighborhood matrix of the local pixel maximum value of the image matrix I with the brightness area removed, defining a peak matrix using the maximum neighborhood matrix, and obtaining the number of peaks; Step S3, using the number of peaks to determine the number of sub-models in the Gaussian mixture model, removing the brightness area of ​​the image matrix I with a pixel area of ​​≤4*4 in step S1, and arranging each pixel point as training data and inputting it into the model, extracting the center point and the coverage of the center point after training as the extracted city center; In step S3, the process of determining the number of sub-models in the Gaussian mixture model by using the number of peaks includes the following steps: Step S301: Gaussian mixture model Where K is the number of peaks, which also represents the number of sub-models in the Gaussian mixture model. represents the probability of the kth submodel in the Gaussian mixture model, φ(x|θ k ) represents the Gaussian distribution density function of the kth sub-model, and θ k =(μ k , σ k ).

2. The method for extracting urban multi-centers based on nighttime light remote sensing images according to claim 1 is characterized in that: In step S1, removing the brightness area with a pixel area ≤ 4*4 in the image matrix I comprises the following steps: Step S101, using kernel to perform morphological opening operation on the image matrix I to remove the brightness area in the image matrix I; The kernel is: Represents the morphological opening structure element.

3. The method for extracting urban multi-centers based on nighttime light remote sensing images according to claim 1 is characterized in that: In step S2, the step of obtaining the maximum neighborhood matrix of the local pixel maximum value of the image matrix without the brightness area comprises the following steps: Step S201, the maximum neighborhood matrix is ​​I out , initialized to I out =I*0; Step S202, set a 3*3 sliding window Where (i, j) is the center point of the sliding window; Step S203, after initialization, out The expanded matrix is Step S204, using a sliding window to traverse the expanded matrix I fill For each pixel in, calculate I out (i-1, j-1) = max(window(i, j)), i∈(1, r), j∈(1, c), r represents the number of rows of the image matrix, and c represents the number of columns of the image matrix.

4. The method for extracting urban multi-centers based on nighttime light remote sensing images according to claim 3 is characterized in that: In step S2, the method for obtaining the number of peaks includes the following steps: Step S205, the peak matrix is ​​I diff =I out -I; Step S206, statistics I diff The number of points K whose median value is 0 is the number of peaks.

5. The method for extracting urban multi-centers based on nighttime light remote sensing images according to claim 1 is characterized in that: In step S3, the process of organizing each pixel point of the image matrix into training data and inputting it into the model includes the following steps: Step S302, organize each pixel of the image matrix I into training data with three attributes, expressed as (index_x, index_y, I(i, j)), where index_x represents the row number of the pixel, index_y represents the column number of the pixel, and (i, j) is the center point of the sliding window. The image matrix I is organized into r*c training data, and then the training data is input into the Gaussian mixture model for training, r represents the number of rows of the image matrix, and c represents the number of columns of the image matrix.

6. The method for extracting urban multi-centers based on nighttime light remote sensing images according to claim 5 is characterized in that: In step S3, the process of extracting the trained center point and the center point coverage as the extracted city center includes the following steps: Step S303, extracting each sub-model parameter of the trained Gaussian mixture model (a k ,m k1 ,m k2 ,s k1 ,s k2 ), In the formula (μ k1 , μ k2 ) represents the position of the kth center in the night light image; Step S304, use the trained Gaussian mixture model to mark the image matrix I, obtain the sub-model category to which each pixel belongs, and then count the number of pixels according to the category to obtain the number of pixels covered by each sub-model. The number of pixels is the city center point, and the number of covered pixels is the coverage range of the city center point.

7. The urban multi-center extraction system based on night light remote sensing images is characterized by: The system includes the following modules: Module S1, used for reading the city night light image and saving it as an image matrix I, and removing the brightness area specified in the image matrix I; Module S2 is used to obtain the maximum neighborhood matrix of the local pixel maximum value of the image matrix I after removing the brightness area and the abnormal points, and define the peak matrix using the maximum neighborhood matrix to obtain the number of peaks; Module S3 is used to determine the number of sub-models in the Gaussian mixture model by using the number of peaks, and to sort each pixel of the image matrix I in the module S1 without the brightness area of ​​pixel area ≤4*4 as training data and input it into the model, and to extract the center point and the coverage of the center point after training as the extracted city center; In the module S3, the process of determining the number of sub-models in the Gaussian mixture model by using the number of peaks includes the following steps: Module S301, Gaussian mixture model Where K is the number of peaks, which also represents the number of sub-models in the Gaussian mixture model. represents the probability of the kth submodel in the Gaussian mixture model, φ(x|θ k ) represents the Gaussian distribution density function of the kth sub-model, and θ k =(μ k , σ k ).

8. The urban multi-center extraction system based on nighttime light remote sensing images according to claim 7 is characterized in that: In the module S1, the brightness area and outliers specified by the image matrix are removed, and the system includes the following modules: Module S101, used for performing morphological opening operation on the image matrix I by using kernel to remove the brightness area in the image matrix I; The kernel is: Represents the morphological opening structure element.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor runs the computer program stored in the memory, the steps of the method according to any one of claims 1 to 6 are performed.

Images