Urban impervious surface inversion method and device, computer equipment and medium

Abstract

The invention provides an inversion method and device for an urban impervious surface, computer equipment and a medium, and belongs to the technical field of digital image processing.The method comprises the steps that remote sensing images of multiple resolutions of a target area are obtained, and a variable end member linear spectrum mixed decomposition model DELSMA is constructed; inputting the remote sensing image into DELSMA, and performing spectrum and spatial feature component extraction on the medium-resolution remote sensing image in an image layering stage of the DELSMA; carrying out feature component enhancement on the extracted spectral and spatial feature components based on the high-resolution remote sensing image; performing image layering on the enhanced feature component to obtain a bright color layer image and a dark color layer image; and introducing a variable end member method to respectively carry out end member extraction on the bright color layer image and the dark color layer image, and respectively carrying out FCLS spectrum mixed decomposition to obtain an output impervious surface inversion result. Through the steps, a high-precision impervious surface inversion result can be finally obtained, and powerful data support is provided for the fields of urban planning, environment monitoring and the like.

Description

Technical Field

[0001] The present invention belongs to the technical field of digital image processing, and in particular relates to an inversion method, device, computer equipment and medium for urban impervious surfaces. Background Art

[0002] The construction and development of cities has led to the gradual replacement of a large number of natural landscapes dominated by vegetation with artificial buildings, resulting in the emergence of impervious surfaces, a typical type of urban surface landscape. Compared with vegetation and bare soil, impervious surfaces have a smaller permeability ratio. They generally refer to the surface of artificial structures such as traffic roads, roofs, and squares covered with materials such as concrete and asphalt. Impervious surface percentage (ISP) refers to the percentage of impervious surface area per unit surface area, and its spatial distribution pattern is a continuous description of urban land use. Impervious surface is an important indicator for evaluating the degree of urban development, affecting the spatial pattern distribution, ecological construction and other aspects of the city.

[0003] Impervious surfaces can be used to monitor urban expansion. Impervious surfaces are an important component of urban landscapes and are widely used in land use coverage and classification. Analyzing the temporal changes and spatial layout of impervious surfaces is helpful for studying urban land use and expansion. Urban expansion is accompanied by changes in land use types, which are usually reflected in the increase of urban construction land and changes in other land use types. Urban expansion and construction usually show consistent changes with the increase of impervious surfaces. Therefore, it is feasible to reflect the expansion and development of cities by monitoring and identifying the evolution of impervious surfaces through remote sensing technology, and it can also be used to evaluate the carrying capacity of urban growth. Under the background of urbanization, changes in land cover have a huge impact on urban development and its ecological environment. Impervious surfaces are an important indicator for measuring the level of urban development, which can effectively reflect the urbanization process and changes in the urban ecological environment. Therefore, accurately obtaining urban impervious surface information and analyzing its spatiotemporal dynamic changes and driving mechanisms have important scientific significance and practical value for urban development planning and ecological construction.

[0004] The extraction of urban impervious surfaces has gone through a development process from manual drawing to computer automatic interpretation. The development of remote sensing interpretation theory and technology has laid the foundation for the continuous updating of impervious surface extraction methods. Aerial photography is the main remote sensing data source for estimating and drawing impervious surfaces. The four commonly used extraction methods include: area method, grid method, image classification technology and substitution method. According to the different principles of impervious surface remote sensing inversion, it can be summarized into three categories: spectral hybrid analysis method, index method and machine learning method.

[0005] For the spectral mixing analysis method, the existing methods mainly use the DELSMA model to analyze the impervious surface. However, the features extracted by the DELSMA model contain limited ground information, and there are problems of same-object different-spectrum interference and excessive end-member decomposition in the end-member selection process, resulting in poor accuracy of the final impervious surface inversion results and large errors in the impervious surface related data obtained. Summary of the invention

[0006] In order to solve the problem of poor accuracy of DELSMA model inversion results, the present invention provides an inversion method, device, computer equipment and medium for urban impervious surface.

[0007] In order to achieve the above object, the present invention provides the following technical solutions:

[0008] Firstly, an inversion method for urban impervious surface is provided, and the method comprises:

[0009] Acquire remote sensing images of the target area at multiple resolutions and construct the DELSMA variable endmember linear spectral mixture decomposition model;

[0010] The remote sensing image is input into DELSMA, and spectral and spatial characteristic components are extracted from the medium-resolution remote sensing image in the image layering stage of DELSMA; and characteristic component enhancement is performed on the extracted spectral and spatial characteristic components based on the high-resolution remote sensing image;

[0011] The enhanced feature components are layered to obtain a bright layer image and a dark layer image;

[0012] The endmembers of the bright layer image and the dark layer image are extracted by the variable endmember method, and the extracted spectral information is subjected to FCLS spectral mixing decomposition to obtain the impervious surface inversion result.

[0013] Optionally, the FCLS spectral mixture decomposition formula is:

[0014]

[0015] Among them, R b is the reflectance of the image b band in the pixel. The image b band refers to a preset specific spectral band in the remote sensing image. i,b is the reflectivity of endmember i in band b, that is, the spectral information obtained by endmember extraction, N is the number of endmembers, f i is the proportion of end member i in the pixel, e b is the standardized residual;

[0016] Based on R b Calculate the proportion of high reflectivity objects in the bright layer image f high亮and the proportion of low reflectivity objects f low亮 And the proportion of low reflectivity objects in dark layer pixels f low暗 , and then calculate the proportion of impervious surface. The specific formula is:

[0017] ISP 亮 =f high亮 +f low亮 ;

[0018] ISP 暗 =f low暗 ;

[0019] ISP DELSMA =ISP 亮 +ISP 暗 ;

[0020] Among them, ISP 亮 is the proportion of impervious surface in the bright layer image, ISP 暗 is the proportion of impervious surface in dark layer image, ISP DELSMA is the impervious surface ratio of the entire image.

[0021] Optionally, the method further comprises:

[0022] Constructing a standard deviation ellipse model, a spatial autocorrelation model and a geographic detector model based on the impervious surface inversion result;

[0023] The spatial distribution information, spatial correlation and driving factors of impervious surfaces were determined by standard deviation ellipse model, spatial autocorrelation model and geographic detector model.

[0024] The expansion driving mechanism of urban impervious surfaces is determined through the spatial distribution information, spatial correlation and driving factors.

[0025] Optionally, the standard deviation ellipse calculation formula is:

[0026]

[0027] Among them, SDE x and SDE y are the coordinates of the weighted center of the impervious surface, i.e., the center of the ellipse; x i and i are the coordinates of the i-th pixel in the impervious surface coverage map, and w is the coordinate of the arithmetic mean center of the impervious surface coverage map; i is the impervious surface coverage of pixel i; n is the total number of pixels; θ is the ellipse azimuth; and is the deviation between the center of the ith pixel and the weighted center of the impervious surface in the x and y directions; δ x and δ y Represents the major and minor semi-axes of the ellipse.

[0028] Optionally, the spatial autocorrelation model includes global spatial autocorrelation and local spatial autocorrelation;

[0029] The Moran index formula for the global spatial autocorrelation is:

[0030]

[0031] Where I represents the global Moran's I, n is the number of regions involved in the analysis; x i and x j are the attribute values ​​of region i and region j respectively, represents the average value of all regional attribute values; w ij is the spatial weight;

[0032] For a certain spatial unit i, the Moran index formula of the local spatial autocorrelation is:

[0033]

[0034] Among them, I i represents the local Moran's I index; the calculation formula of the intermediate variable S3 is:

[0035]

[0036] Optionally, the driving factor contribution rate formula of the geographic detector model is:

[0037]

[0038] D is the driving factor, I is the impervious surface expansion rate, P D,I is the contribution rate of D to I, n and σ 2 are the number of samples and variance of the impervious surface expansion rate under a certain driving factor, m is the number of categories of variable Y or factor X, n D,i is the number of D indicators on class i samples.

[0039] Secondly, an inversion device for urban impervious surface is provided, the device comprising:

[0040] The acquisition module acquires remote sensing images of the target area with multiple resolutions and constructs the DELSMA model with variable endmember linear spectral mixture decomposition;

[0041] A feature extraction module is used to input the remote sensing image into DELSMA, extract spectral and spatial feature components of the medium-resolution remote sensing image in the image layering stage of DELSMA; and enhance the extracted spectral and spatial feature components based on the high-resolution remote sensing image;

[0042] A stratification module is used to stratify the enhanced feature components to obtain a bright layer image and a dark layer image;

[0043] The endmember extraction module is used to extract endmembers from the bright layer image and the dark layer image respectively by using the variable endmember method, and to perform FCLS spectral mixing decomposition on the extracted spectral information to obtain the impervious surface inversion result.

[0044] In addition, a computer-readable storage medium is provided, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned inversion method for urban impervious surfaces is implemented.

[0045] Finally, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned inversion method for urban impervious surfaces when executing the program.

[0046] The inversion method of an urban impervious surface provided by the present invention has the following beneficial effects:

[0047] In the present invention, the spectral and spatial characteristic components are enhanced by high-resolution images, and the advantages of images with different resolutions are fully utilized to provide more comprehensive and richer information about the objects. While improving the accuracy and stability of the spectral and spatial characteristic components, the high consumption of computing resources by high-resolution images and the information interference caused by redundant information are avoided; then the enhanced characteristic components are image-layered into a bright layer and a dark layer. The images of these two layers respectively reflect the spectral and spatial characteristics of different objects. This step helps to better capture the spectral characteristics of the objects and reduce the interference of the same object with different spectra and the same spectrum of different objects. It provides more accurate data support for the subsequent inversion of impervious surfaces; the variable end member method is used for end member extraction, and the number of end members can be flexibly adjusted according to the image characteristics of different levels, so as to more accurately reflect the spectral characteristics of the objects; the FCLS spectral mixed decomposition can further decompose the mixed pixels and improve the accuracy of the inversion of the impervious surface. In this way, through the processing and analysis of the above steps, a high-precision inversion result of the impervious surface can be finally obtained. These results can provide strong data support for urban planning, environmental monitoring and other fields, and help to achieve more scientific and accurate management and decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiment of the present invention and its design scheme, the following briefly introduces the drawings required for this embodiment. 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 creative work.

[0049] Figure 1 A schematic diagram of a DELSMA model provided by the present invention according to an exemplary embodiment.

[0050] Figure 2 A schematic flow chart of an inversion method for urban impervious surfaces provided by the present invention according to an exemplary embodiment.

[0051] Figure 3 A schematic diagram of a research model for a driving mechanism of urban impervious surface expansion provided by the present invention according to an exemplary embodiment.

[0052] Figure 4 A local Moran index spatial autocorrelation quadrant diagram is provided according to an exemplary embodiment of the present invention.

[0053] Figure 5 The present invention is a schematic diagram of a geographic detector according to an exemplary embodiment.

[0054] Figure 6 A block diagram of an inversion device for urban impervious surfaces provided by the present invention according to an exemplary embodiment. DETAILED DESCRIPTION

[0055] In order to enable those skilled in the art to better understand the technical solution of the present invention and implement it, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the scope of protection of the present invention.

[0056] The present invention constructs a variable end-member linear spectral mixed decomposition model for inverting the impervious surface coverage of the study area, specifically as follows Figure 1 .

[0057] The technical solutions provided by various embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0058] First, the present invention provides an inversion method for urban impervious surfaces, specifically: Figure 1 As shown, the following steps are included:

[0059] S201, obtain remote sensing images of the target area with multiple resolutions, and construct a variable endmember linear spectral mixture decomposition model DELSMA.

[0060] In this step, it is necessary to collect multi-resolution remote sensing images of the target area obtained at different times and under different sensors. These images may come from different satellite platforms, such as Landsat, Sentinel-2, MODIS, etc., with different spatial resolutions (such as 30 meters, 10 meters, 1 meter, etc.) and spectral resolutions. In the present invention, urban areas are studied, and multispectral images with a resolution of 30 meters can be obtained from Landsat 8, images with a resolution of 10 meters can be obtained from Sentinel-2, and high-definition images with a resolution of 1 meter can be obtained from drones.

[0061] The acquired remote sensing images of the target area with multiple resolutions can first be preprocessed.

[0062] For example, the following preprocessing can be performed on remote sensing images:

[0063] Radiometric calibration: Convert the original digital value (DN value) of remote sensing data into physical quantities such as radiometric brightness value, reflectivity or surface temperature. This ensures that data collected at different times and by different sensors have consistent scales, which is helpful for subsequent comparison and quantitative analysis.

[0064] Atmospheric correction: Considering the effects of atmospheric absorption and scattering, the image is corrected to reduce the impact of the atmosphere on remote sensing data. After correction, the accuracy of the data can be improved, especially when conducting quantitative analysis.

[0065] Geometric correction: including geolocation, geometric precision correction, image registration and orthorectification. The purpose is to ensure the accuracy of remote sensing data in geographic coordinates and match the pixel position with the actual position on the earth's surface. This usually involves operations such as correcting ground distortion, projection transformation and registration.

[0066] Noise reduction: Reduce random and systematic noise in images through filtering techniques or other noise reduction methods. This helps improve image quality and make subsequent analysis more accurate.

[0067] Image cropping: cropping an image to remove unnecessary areas to reduce data volume and improve processing efficiency. Cropping is usually based on the study area or region of interest (ROI).

[0068] Image fusion: Fusion of images from different bands or sensors into a multispectral image. This can improve the spatial and spectral resolution of the image, providing more information for subsequent analysis.

[0069] Removing clouds and shadows: Using cloud and shadow masks in remote sensing data or specialized cloud detection algorithms to remove the effects of clouds and shadows. This ensures that the ground object information in the image is reliable and not interfered by clouds and shadows.

[0070] In another embodiment, a dynamic endmember linear spectral mixing analysis (DELSMA) is a linear spectral mixing decomposition model that takes into account the dynamic changes of land cover types, and allows endmembers (i.e., basic spectral characteristics of land cover types) to change with time and place, thereby improving the adaptability and accuracy of the model. In the present invention, when constructing DELSMA, it may be necessary to consider different land surface types in the city, such as vegetation, water bodies, buildings, roads, etc., and allow the spectral characteristics of these types of endmembers to change with seasons, weather and other conditions, and build a DELSMA model framework based on the above-mentioned multiple influencing factors.

[0071] S202: Input the remote sensing image into DELSMA, and extract spectral and spatial feature components of the medium-resolution remote sensing image in the image layering stage of DELSMA.

[0072] In this step, the preprocessed medium-resolution remote sensing image is input into the DELSMA model as input data, and then feature extraction is performed on the input data, and image stratification is performed based on the extracted features. The DELSMA model was originally designed for medium-resolution images, and its algorithm and parameter settings are more suitable for the data characteristics of medium-resolution images. Therefore, medium-resolution remote sensing images are used as input in this step. Among them, medium-resolution remote sensing images refer to remote sensing images with medium spatial resolution. The "spatial resolution" here refers to the size of the smallest ground unit that can be distinguished in the image. The actual distance on the ground (such as meters) is usually used to represent the size of the area represented by each pixel. For example, the 30-meter resolution multispectral images provided by the Landsat series of satellites are considered medium-resolution images.

[0073] Feature extraction of remote sensing images involves performing principal component analysis (PCA), independent component analysis (ICA) or other similar dimensionality reduction techniques on the images to extract the spectral and spatial characteristic components that best represent the surface cover type.

[0074] For example, for the 30-meter resolution image of Landsat 8, we can use the PCA method to extract the first few principal components, which may represent different types of surface cover such as vegetation, water bodies, soil, etc. The extracted characteristic components can be used for subsequent image processing and analysis.

[0075] S203, enhancing the extracted spectral and spatial characteristic components based on the high-resolution remote sensing image.

[0076] In this step, since the medium-resolution remote sensing image contains limited ground object information, the subsequent steps will extract less information and the recognition result will be less accurate. Therefore, the spatial detail information of the high-resolution remote sensing image can be used to enhance the extracted spectral and spatial feature components. This can be achieved by fusing or matching the high-resolution image with the low-resolution feature components to improve the spatial resolution and accuracy of the feature components. Among them, high-resolution remote sensing images refer to remote sensing images with higher spatial resolution than medium-resolution remote sensing images, which can provide more detailed and clearer ground feature information. The spatial resolution of high-resolution images is usually less than 10 meters, and can even reach sub-meter level (such as 1 meter or lower), which enables them to capture more ground details, such as the outline of buildings, the direction of roads, the distribution of vegetation, etc. High-resolution remote sensing images are of great value in accurate ground object recognition, urban change monitoring, disaster assessment, etc.

[0077] For example, the feature components of the 10-meter resolution image of Sentinel-2 and the 30-meter resolution image extracted by Landsat 8 can be fused to improve the spatial resolution of the feature components. The fusion methods may include simple pixel-level fusion, feature-based fusion, or decision tree-based fusion.

[0078] S204, performing image layering on the enhanced feature components to obtain a light layer image and a dark layer image.

[0079] In this step, further image layering is performed based on the enhanced feature components. This involves threshold segmentation of the image based on brightness or reflectance to distinguish between bright and dark layers. The bright layer represents the surface types with high reflectance, such as buildings and roads, while the dark layer represents the surface types with low reflectance, such as vegetation and water bodies.

[0080] For example, for the enhanced feature component, the present invention can set a reflectance threshold, classify pixels above the threshold as a bright layer, and pixels below the threshold as a dark layer, so that two image layers representing the bright and dark surfaces can be obtained.

[0081] For the layered images, the MNF transformation can be performed on the bright layer image and the dark layer image respectively, and the signal-to-noise ratio of the image and the accuracy of subsequent processing can be improved through the MNF transformation. Specifically, noise whitening is first performed, and the noise covariance matrix in the principal component is used to separate and readjust the noise in the data so that the transformed noise data has only the minimum variance and no correlation between bands. Then the principal component transformation is performed on the noise whitening data to further extract the main information in the data.

[0082] S205, introducing a variable endmember method to extract endmembers from the bright layer image and the dark layer image respectively, and performing FCLS spectral mixing decomposition on the extracted spectral information respectively to obtain an impervious surface inversion result.

[0083] In this step, PPI calculation is required first to identify pure pixels in the image. These pure pixels will serve as the basis for endmember extraction. Based on the results of PPI calculation, endmembers in the image, that is, representative land feature types, are extracted.

[0084] Among them, the PPI calculation formula is:

[0085]

[0086] X is the number of horizontal pixels, indicating the number of pixels in the horizontal direction of the screen, Y is the number of vertical pixels, indicating the number of pixels in the vertical direction of the screen, and Z is the diagonal size of the screen, indicating the physical size of the screen in inches.

[0087] In one embodiment, it is necessary to introduce a variable endmember method to extract endmembers from the bright layer image and the dark layer image respectively. In the bright layer and the dark layer image, the variable endmember method is applied to extract endmembers respectively. This means that in each image layer, the spectral characteristics of the endmembers are dynamically changing and need to be extracted according to the actual situation. After the endmembers are extracted, the FCLS (Fully Constrained Least Squares) method is used to perform spectral mixture decomposition. FCLS is a linear spectral mixture decomposition method, which assumes that the spectrum of each pixel is composed of a linear combination of several endmembers. Through FCLS decomposition, the abundance information of different endmembers in each pixel can be obtained. Based on this information, the proportion or coverage of the impervious surface can be calculated as the final output result.

[0088] Specifically, since the impervious surface inversion method may have the problems of same-object different-spectrum interference and excessive endmember decomposition in the endmember selection process, in order to prevent the above problems, this step introduces image layering technology and variable endmember method, which effectively improves the inversion accuracy of impervious surfaces in high-density urban areas. The expression is as follows:

[0089] The FCLS spectral mixture decomposition formula is:

[0090]

[0091] Among them, R b is the reflectance of the image b band in the pixel. The image b band refers to a preset specific spectral band in the remote sensing image. i,b is the reflectivity of endmember i in band b, that is, the spectral information obtained by endmember extraction, N is the number of endmembers, f iis the proportion of end member i in the pixel, e b is the standardized residual;

[0092] Based on R b Calculate the proportion of high reflectivity objects in the bright layer image f high亮 and the proportion of low reflectivity objects f low亮 And the proportion of low reflectivity objects in dark layer pixels f low暗 , and then calculate the proportion of impervious surface. The specific formula is:

[0093] ISP 亮 =f high亮 +f low亮 ;

[0094] ISP 暗 =f low暗 ;

[0095] ISP DELSMA =ISP 亮 +ISP 暗 ;

[0096] Among them, ISP 亮 is the proportion of impervious surface in the bright layer image, ISP 暗 is the proportion of impervious surface in dark layer image, ISP DELSMA is the impervious surface ratio of the entire image.

[0097] Based on the impervious surface inversion results obtained in the above steps, the present invention also constructs a research model of the driving mechanism of urban impervious surface expansion, specifically as follows: Figure 3 shown.

[0098] In this research model, it is necessary to study the overall expansion trend, expansion rate, internal hierarchical structure changes, spatial differentiation characteristics, expansion direction, landscape pattern characteristics and driving mechanism of the impervious surface. Therefore, based on the above requirements, the present invention also constructs a standard deviation ellipse model, a spatial autocorrelation model and a geographic detector model to study the impervious surface.

[0099] The standard deviation ellipse (SDE) is one of the classic methods for analyzing the directional characteristics of the spatial distribution of geographic elements. The key parameters include the major semiaxis, the minor semiaxis, and the deflection angle. In this study, the length of the major and minor semiaxis reflects the concentration density of the spatial distribution of the impervious surface, the deflection angle reflects the dominant direction of the spatial distribution of the impervious surface, and the ratio of the major and minor semiaxis reflects the expansion trend of the impervious surface in the dominant direction. The calculation formula of the standard deviation ellipse is as follows:

[0100]

[0101] Among them, SDE x and SDE yare the coordinates of the weighted center of the impervious surface, i.e., the center of the ellipse; x i and i are the coordinates of the i-th pixel in the impervious surface coverage map, and w is the coordinate of the arithmetic mean center of the impervious surface coverage map; i is the impervious surface coverage of pixel i; n is the total number of pixels; θ is the ellipse azimuth; and is the deviation between the center of the ith pixel and the weighted center of the impervious surface in the x and y directions; δ x and δ y Represents the major and minor semi-axes of the ellipse.

[0102] The spatial autocorrelation model includes global spatial autocorrelation and local spatial autocorrelation. Among them, global Moran's I is the most commonly used statistical variable for global spatial autocorrelation test. The global Moran's I value is between positive and negative 1. A value greater than 0 indicates that the region as a whole has positive spatial autocorrelation, and the spatial distribution is clustered. The closer the attribute value is to 1, the smaller the overall spatial difference. A value less than 0 indicates that the region as a whole has negative spatial autocorrelation, and the spatial distribution is discrete. The closer the attribute value is to -1, the greater the overall spatial difference. A value equal to 0 indicates that there is no autocorrelation in the spatial distribution, and it presents an independent random distribution. The Moran's index formula for the global spatial autocorrelation is:

[0103]

[0104] Where I represents the global Moran's I, n is the number of regions involved in the analysis; x i and i are the attribute values ​​of region i and region j respectively, represents the average value of all regional attribute values; w ij is the spatial weight (when region i and region j are adjacent, w ij =1; when area i and area j are not adjacent, w ij =0).

[0105] In this study, local Moran's I is used to measure local spatial autocorrelation. In essence, local Moran's I decomposes the global correlation index into the local space, which is a decomposition form of global Moran's I. Anselin calls it LISA, which stands for local indicators of spatial association. For a certain spatial unit i, the Moran's index formula for the local spatial autocorrelation is:

[0106]

[0107] Among them, I i represents the local Moran's I index; the calculation formula of the intermediate variable S3 is:

[0108]

[0109] I i The value range is between positive and negative 1. A value greater than 0 means a high value is surrounded by high values ​​(high to high), or a low value is surrounded by low values ​​(low to low); a value less than 0 means a low value is surrounded by high values ​​(low to high), or a high value is surrounded by low values ​​(high to low); equal to 0 means a homogeneous regional distribution.

[0110] Local spatial autocorrelation can not only measure the degree of spatial association within a region, but also find out where the spatial clustering points are. This study uses Moran scatter plots and LISA cluster plots to visualize the local differences in impervious surface expansion, which are used to identify local regional clustering types and "hot spots" and "cold spots" within regional development. It should be noted that the local Moran's I results for all regions will be displayed in the Moran scatter plot, but only through Z Ii The area of ​​statistical test will be displayed in the LISA cluster diagram. Ii The test formula is the same as the global Moran's I Z Ii The test formula is the same.

[0111] The Moran scatter plot can be divided into four quadrants, corresponding to four different spatial patterns, such as Figure 4 As shown. Upper right quadrant (HH): high-value areas are surrounded by high-value neighbors; upper left quadrant (LH): low-value areas are surrounded by high-value neighbors; lower left quadrant (LL): low-value areas are surrounded by low-value neighbors; lower right quadrant (HL): high-value areas are surrounded by low-value neighbors. HH and LL indicate that the difference between the region and its surrounding areas is small, that is, the areas with higher or lower values ​​are concentrated, while LH and HL indicate that there is a certain degree of difference in the attribute values ​​between the region and its surrounding areas. Among them, H indicates that the variable attribute value is higher than the average value, and L indicates that the variable attribute value is lower than the average value. The LISA cluster map can intuitively display the areas that pass the significance test on the map and identify the "hot spots" and "cold spots" in the regional development process.

[0112] Geographic detectors are spatial statistical methods used to explore the influence of geographical factors on the spatial differentiation of a geographical phenomenon and the reasons for this differentiation and to analyze them. The core idea is based on the assumption that if an independent variable has an important influence on a dependent variable, then the spatial distribution of the independent variable and the dependent variable should be similar. The specific approach is to compare the total variance of the detection index in different category partitions with the total variance of the index in the entire study area, such as Figure 5 As shown, use P D,I Value measurement, the expression is:

[0113]

[0114] D is the driving factor, I is the impervious surface expansion rate, P D,I is the contribution rate of D to I, n and σ 2 are the number of samples and variance of the impervious surface expansion rate under a certain driving factor, m is the number of categories of variable Y or factor X, n D,i is the number of D indicators on class i samples. D,I The value range of is [0,1]. The larger the value, the stronger the explanatory power of the driving factor D for the impervious surface expansion rate.

[0115] The above method is used to enhance the spectral and spatial characteristic components through high-resolution images, making full use of the advantages of images with different resolutions to provide more comprehensive and richer information about the objects. While improving the accuracy and stability of the spectral and spatial characteristic components, it avoids the high consumption of computing resources by high-resolution images and the information interference caused by redundant information. Then the enhanced characteristic components are layered into bright and dark layers. The images of these two layers reflect the spectral and spatial characteristics of different objects respectively. This step helps to better capture the spectral characteristics of the objects and reduce the interference of the same object with different spectra and different objects with the same spectrum. It provides more accurate data support for the subsequent inversion of impervious surfaces. The variable endmember method is used for endmember extraction, and the number of endmembers can be flexibly adjusted according to the image characteristics of different levels, so as to more accurately reflect the spectral characteristics of the objects. The FCLS spectral mixed decomposition can further decompose the mixed pixels and improve the accuracy of the inversion of the impervious surface. In this way, through the processing and analysis of the above steps, a high-precision inversion result of the impervious surface can be obtained. These results can provide strong data support for urban planning, environmental monitoring and other fields, and help to achieve more scientific and accurate management and decision-making.

[0116] Secondly, the present invention also provides an inversion device for urban impervious surfaces, such as Figure 6 As shown, including:

[0117] The acquisition module 601 acquires remote sensing images of the target area at multiple resolutions and constructs a variable end member linear spectral mixture decomposition model DELSMA.

[0118] The feature extraction module 602 is used to input the remote sensing image into DELSMA, extract spectral and spatial feature components of the medium-resolution remote sensing image in the image layering stage of DELSMA; and enhance the extracted spectral and spatial feature components based on the high-resolution remote sensing image.

[0119] The stratification module 603 is used to perform stratification on the enhanced feature components to obtain a bright layer image and a dark layer image.

[0120] The endmember extraction module 604 is used to extract endmembers from the bright layer image and the dark layer image respectively by using the variable endmember method, and to perform FCLS spectral mixing decomposition on the extracted spectral information respectively to obtain the impermeable surface inversion result.

[0121] By using the above device, the spectral and spatial characteristic components are enhanced through high-resolution images, and the advantages of images with different resolutions are fully utilized to provide more comprehensive and richer information about the objects. While improving the accuracy and stability of the spectral and spatial characteristic components, the high consumption of computing resources by high-resolution images and the information interference caused by redundant information are avoided; then the enhanced characteristic components are image-stratified into bright layer and dark layer. The images of these two layers respectively reflect the spectral and spatial characteristics of different objects. This step helps to better capture the spectral characteristics of the objects and reduce the interference of the same object with different spectra and different objects with the same spectrum. It provides more accurate data support for the subsequent inversion of impervious surfaces; the variable endmember method is used for endmember extraction, and the number of endmembers can be flexibly adjusted according to the image characteristics of different levels, so as to more accurately reflect the spectral characteristics of the objects; FCLS spectral mixed decomposition can further decompose mixed pixels and improve the accuracy of impervious surface inversion. In this way, through the processing and analysis of the above steps, a high-precision inversion result of the impervious surface can be obtained. These results can provide strong data support for urban planning, environmental monitoring and other fields, and help to achieve more scientific and accurate management and decision-making.

[0122] The present invention also provides a computer-readable storage medium, which stores a computer program, which can be used to execute the above Figure 2 The steps of the inverse method of urban impervious surface are provided.

[0123] The present invention also provides a computer device. At the hardware level, the computer device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory. Of course, it may also include hardware required for other services. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to achieve the above Figure 2 The steps of the inverse method of urban impervious surface are provided.

[0124] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0125] The present invention is described with reference to flowcharts and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as a combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0126] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0127] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0128] It should be noted that the above specific implementation method can enable those skilled in the art to understand the invention more comprehensively, but does not limit the invention in any way. Therefore, although the present invention has been described in detail in this specification, those skilled in the art should understand that the invention can still be modified or replaced by equivalents; and all technical solutions and improvements that do not deviate from the spirit and scope of the invention are included in the protection scope of the patent for the invention. Any figure mark in the claims should not be regarded as limiting the claims involved.

Claims

1. A method for inverting urban impervious surfaces, characterized in that: The method comprises: Acquire remote sensing images of the target area at multiple resolutions and construct the DELSMA variable endmember linear spectral mixture decomposition model; The remote sensing image is input into DELSMA, and spectral and spatial characteristic components are extracted from the medium-resolution remote sensing image in the image layering stage of DELSMA; and characteristic component enhancement is performed on the extracted spectral and spatial characteristic components based on the high-resolution remote sensing image; The enhanced feature components are layered to obtain a bright layer image and a dark layer image; The endmembers of the bright layer image and the dark layer image are extracted by the variable endmember method, and the extracted spectral information is subjected to FCLS spectral mixing decomposition to obtain the impervious surface inversion result.

2. The inversion method of urban impervious surface according to claim 1, characterized in that: The FCLS spectral mixture decomposition formula is: Among them, R b is the reflectance of the image b band in the pixel. The image b band refers to a preset specific spectral band in the remote sensing image. i，b is the reflectivity of endmember i in band b, that is, the spectral information obtained by endmember extraction, N is the number of endmembers, and f i is the proportion of end member i in the pixel, e b is the standardized residual; Based on R b Calculate the proportion of high reflectivity objects in the bright layer image and the proportion of low reflectivity objects And the proportion of low reflectivity objects in dark layer pixels Then the proportion of impervious surface is calculated, and the specific formula is: ISP DELSMA =ISP 亮 +ISP 暗 ; Among them, ISP 亮 is the proportion of impervious surface in the bright layer image, ISP 暗 is the proportion of impervious surface in dark layer image, ISP DELSMA is the impervious surface ratio of the entire image.

3. The inversion method of urban impervious surface according to claim 1, characterized in that: The method further comprises: Constructing a standard deviation ellipse model, a spatial autocorrelation model and a geographic detector model based on the impervious surface inversion result; The spatial distribution information, spatial correlation and driving factors of impervious surfaces were determined by standard deviation ellipse model, spatial autocorrelation model and geographic detector model. The expansion driving mechanism of urban impervious surfaces is determined through the spatial distribution information, spatial correlation and driving factors.

4. The inversion method of urban impervious surface according to claim 3 is characterized in that: The standard deviation ellipse calculation formula is: Among them, SDE x and SDE y are the coordinates of the weighted center of the impervious surface, i.e., the center of the ellipse; x i and i are the coordinates of the i-th pixel in the impervious surface coverage map, and w is the coordinate of the arithmetic mean center of the impervious surface coverage map; i is the impervious surface coverage of pixel i; n is the total number of pixels; θ is the ellipse azimuth; and is the deviation between the center of the ith pixel and the weighted center of the impervious surface in the x and y directions; δ x and δ y Represents the major and minor semi-axes of the ellipse.

5. The inversion method of urban impervious surface according to claim 3, characterized in that: The spatial autocorrelation model includes global spatial autocorrelation and local spatial autocorrelation; The Moran index formula for the global spatial autocorrelation is: Where I represents the global Moran's I, n is the number of regions involved in the analysis; x i and x j are the attribute values ​​of region i and region j respectively, represents the average value of all regional attribute values; w ij is the spatial weight; For a certain spatial unit i, the Moran index formula of the local spatial autocorrelation is: Among them, I i represents the local Moran's I index; the calculation formula of the intermediate variable S3 is:

6. The inversion method of urban impervious surface according to claim 3, characterized in that: The driving factor contribution rate formula of the geographic detector model is: D is the driving factor, I is the impervious surface expansion rate, P D，I is the contribution rate of D to I, n and σ 2 are the number of samples and variance of the impervious surface expansion rate under a certain driving factor, m is the number of categories of variable Y or factor X, n D，i is the number of D indicators on class i samples.

7. An inversion device for urban impervious surfaces, characterized in that: The device comprises: The acquisition module acquires remote sensing images of the target area with multiple resolutions and constructs the DELSMA model with variable endmember linear spectral mixture decomposition; A feature extraction module is used to input the remote sensing image into DELSMA, extract spectral and spatial feature components of the medium-resolution remote sensing image in the image layering stage of DELSMA; and enhance the extracted spectral and spatial feature components based on the high-resolution remote sensing image; A stratification module is used to stratify the enhanced feature components to obtain a bright layer image and a dark layer image; The endmember extraction module is used to extract endmembers from the bright layer image and the dark layer image respectively by using the variable endmember method, and to perform FCLS spectral mixing decomposition on the extracted spectral information to obtain the impervious surface inversion result.

8. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

9. A computer device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method according to any one of claims 1 to 6 is implemented.

Images