A Multi-Source Remote Sensing Spatial Mapping Method for Large-Scale Resident Well-being

Abstract

The present invention discloses a multi-source remote sensing spatial mapping method for large-scale resident well-being. Based on multi-source remote sensing data, remote sensing information mining methods are used to extract remote sensing spatial features such as night light radiation intensity, impervious surface density, vegetation coverage index, road density, etc. At the same time, remote sensing implicit features related to resident well-being are extracted to construct a remote sensing feature set for resident well-being; the remote sensing features of resident well-being are combined with the resident well-being values calculated from the county-scale social statistical yearbook data to construct a spatial simulation model of resident well-being and remote sensing feature indicators. Further, a downscaling method is adopted to achieve continuous spatial simulation of resident well-being, and an improved spatial autocorrelation analysis method is used to realize the evolution spatial mapping of large-scale resident well-being.

Description

Technical Field

[0001] The present invention belongs to the technical field of remote sensing intelligent assessment of residents' well - being, and particularly relates to a method for multi - remote - sensing spatial mapping of large - scale residents' well - being. Background Art

[0002] Residents' well - being is an important indicator to measure social development and people's living quality. It is difficult to conduct large - scale well - being surveys and assessments. And for large - scale mapping of residents' well - being assessment, a large amount of multi - source data such as statistical data and survey data needs to be integrated, and technical methods such as geographic information systems and spatial analysis are used to draw the spatial distribution map of residents' well - being. However, challenges still exist in aspects such as data accuracy, mapping model construction, and large - scale comparability.

[0003] Affected by various factors such as geographical environment, economic development level, and social culture, residents' well - being shows significant spatial heterogeneity. Large - scale assessment of residents' well - being involves a wide range, and data sources are scattered. The data standards and formats of different departments and regions are inconsistent, resulting in difficulties in data acquisition and integration. At the same time, residents' well - being is a concept of dynamic change. Most of the current assessment methods focus on static analysis and lack the ability to effectively monitor and predict the dynamic change of residents' well - being. It is difficult to timely reflect the spatio - temporal evolution trend of well - being levels, which is not conducive to the timely adjustment and optimization of policies. Summary of the Invention

[0004] Aiming at the above deficiencies in the prior art, the multi - remote - sensing spatial mapping method for large - scale residents' well - being provided by the present invention solves the problem that it is difficult for existing related methods to achieve rapid spatial assessment and mapping of large - scale residents' well - being.

[0005] In order to achieve the above - mentioned invention purpose, the technical solution adopted by the present invention is: a multi - source remote - sensing spatial mapping method for large - scale residents' well - being, including the following steps:

[0006] S1. Collect multi - source data of the spatial well - being of residents at the county - scale and pre - process it;

[0007] S2. Extract remote - sensing features related to residents' well - being according to the pre - processed data;

[0008] The remote - sensing features include night - time light intensity features, impervious surface index features, vegetation index features, road density index features, point - of - interest data features, and remote - sensing implicit features;

[0009] S3. Collect data from the national statistical yearbook at the county - scale, conduct weighted statistics on it, and quantitatively determine the residents' well - being value;

[0010] S4. Combine the remote sensing features and the well-being values of residents to construct a random forest model for predicting the well-being values of residents. Map the prediction results of the well-being values of residents from the county scale to the pixel scale through a downscaling function, and combine the remote sensing feature sequences of different periods to obtain the evaluation results of the well-being values of residents at different periods on a large scale;

[0011] S5. Based on the evaluation results of the well-being values of residents, calculate the spatial autocorrelation index of different regions, and draw a spatial distribution map of residents' well-being according to it to show the aggregation and association patterns of residents' well-being in space.

[0012] Furthermore, in the step S1, the multi-source data includes multi-source optical remote sensing data, night light data, point of interest data, and statistical yearbook data;

[0013] The preprocessing includes:

[0014] The preprocessing of multi-source optical remote sensing data includes radiometric calibration, atmospheric correction, geometric correction, band registration, multi-band image synthesis, and image fusion of multi-source optical remote sensing images;

[0015] The preprocessing of night light data includes radiometric calibration, atmospheric correction, geometric correction, and data clipping and splicing;

[0016] The preprocessing of point of interest data includes duplicate data removal, data cleaning, coordinate system unification, and data classification.

[0017] Furthermore, in the step S2, the night light intensity features include the total value of night light intensity, the average value of night light intensity, and the standard deviation of night light intensity;

[0018] The impervious surface index feature is the normalized difference impervious surface index obtained by operating on the reflectance of the short-wave infrared band, the near-infrared band, the green band, and the mid-infrared band;

[0019] The vegetation index features include the normalized vegetation index and the enhanced vegetation index;

[0020] The road density index feature is the proportion of the length of roads per unit area;

[0021] The point of interest data feature is the proportion of the number of various types of points of interest in the region;

[0022] The remote sensing implicit feature is the feature information representing the internal attributes, physical processes, or human activities related to the well-being of residents extracted from the remote sensing images.

[0023] Furthermore, the method for extracting the remote sensing implicit remote sensing features is specifically as follows:

[0024] Clip and normalize the remote sensing images in the study area, and input the obtained input feature map into the trained convolutional neural network to obtain remote sensing latent features;

[0025] Among them, the convolutional neural network includes an input layer, a number of convolutional layers, a pooling layer, and an output layer connected in sequence;

[0026] The output feature map O l (i, j, c) of the l-th convolutional layer is:

[0027]

[0028] In the formula, w l (m, n, d, c) represents the weight of the c-th convolutional kernel at the position (m, n) corresponding to the d-th input channel in the l-th convolutional layer, and b l (c) represents the bias of the c-th convolutional kernel in the l-th convolutional layer, D l represents the number of channels of the input feature map I l , i, j are the coordinates of the output feature map, s l represents the convolution step, k l ×k l represents the size of the convolutional kernel in the l-th convolutional layer, D l represents the number of input channels;

[0029] The output feature map p l (i, j, c) of the pooling layer is:

[0030]

[0031] In the formula, p l ×p l represents the pooling window, represents the pooling step;

[0032] The output layer is a fully connected layer, and the n remote sensing latent features C n output by it are expressed as:

[0033] C n = W fc ×vec(F) + b fc

[0034] In the formula, W fc represents the weight matrix of the fully connected layer, b fc represents the bias vector of the fully connected layer, and vec(F) represents the vector obtained by flattening the output feature map F of the pooling layer, and its dimension is hwd, where h, w, and d respectively represent the height, width, and depth of the output feature map.

[0035] Furthermore, the step S3 includes the following sub-steps:

[0036] S31. Collect statistical yearbook data at the county level and select statistical indicators related to residents' well-being;

[0037] S32. Standardize the selected statistical indicators by using the range standardization method;

[0038] S33. Determine the weight of each statistical indicator after standardization by entropy weight method;

[0039] S34. Quantitatively calculate the resident welfare value of the study area based on the weights of each statistical indicator and statistical yearbook data.

[0040] Furthermore, the step S4 includes the following sub-steps:

[0041] S41. Calculate the mean and variance of various remote sensing characteristics in the county as the remote sensing characteristic values corresponding to the resident welfare value;

[0042] S42. Based on the remote sensing feature values and resident well-being values at the county scale, a random forest model is constructed, and a downscaling function is used to map the resident well-being values at the county scale to the pixel scale.

[0043] S43. Based on the predicted values of resident welfare values at the pixel scale, remote sensing feature sequences of different periods are used to explore the dynamic evolution of resident welfare values over time in a large-scale range by setting quantitative indicators, and the evaluation results of resident welfare values in different periods at a large scale are obtained.

[0044] Furthermore, the step S42 includes the following sub-steps:

[0045] S42-1. Construct dataset D based on the resident welfare values and remote sensing feature values at the county level;

[0046] The data set D contains several subsets (D1, D2, ..., D t ), each subset represents the remote sensing feature values and resident welfare values corresponding to the resident welfare values of different counties, D t ={Y t ,X t1 ,X t2 ,…,X ti ,…,X tn}; where Y t represents the welfare value of residents in the tth subset, X t1 ,X t2 ,…,X ti ,…,X tn Represents Y t The corresponding 1st to nth remote sensing feature values;

[0047] S42-2. For each subset, construct a corresponding decision tree;

[0048] S42-3. Combine the k constructed decision trees into a random forest model, input the remote sensing feature values in the area to be simulated into the random forest model, and perform regression on the prediction results corresponding to each decision tree to obtain the predicted value of the resident well-being value;

[0049] S42-4. Based on the association between the remote sensing feature values at the county scale and the pixel scale, construct a downscaling function f(x);

[0050] S42-5. Use the downscaling function f(x) to map the predicted value of the resident well-being at the county scale to the pixel scale.

[0051] Furthermore, in step S42-2, during the construction of the decision tree, at each node, randomly select p remote sensing feature values from the n remote sensing feature values of the subset, and the corresponding information gain Gain(D,A) is:

[0052]

[0053] In the formula, the information gain Gain(D,A) is used to measure the contribution degree of the remote sensing feature value to the data set D, Entropy(D v ) represents the entropy of the data set, which is used to measure the degree of chaos of the data set, A represents the selected remote sensing feature value, values(A) represents the set of all possible values of the remote sensing feature A, D v represents the sample subset in the data set where the remote sensing feature A takes the value v, represents the proportion of the sample subset where the remote sensing feature A takes the value v in the data set D in the data set D;

[0054] In step S42-5, the predicted value Y pixel of the resident well-being at the pixel scale for the j-th pixel is expressed as:

[0055] Y pixel (j) = f(Y countt ,x 1j ,x 2j ,…,x ij ,…x nj )

[0056]

[0057] In the formula, f(·) represents the downscaling function, x 1j ,x 2j ,……x nj represents the remote sensing feature value of the j-th pixel, Y countyrepresents the well-being value at the county scale, β0(j) represents the regression coefficient related to pixel j, and x ij represents the value of the i-th remote sensing feature corresponding to the j-th pixel position.

[0058] Furthermore, the step S5 includes the following sub-steps:

[0059] S51. Set the distance-based spatial weight matrix W d and the weight matrix W based on the economic connection strength e ;

[0060] S52. Combine the matrices W d and W e to obtain the combined spatial weight matrix W;

[0061] S53. According to the combined spatial weight matrix W, calculate the spatial autocorrelation index I between the well-being values of different regions i mprove d ;

[0062] S54. Divide the research area into grids according to geographical coordinates, and perform hierarchical rendering on the grids according to the magnitude of the well-being value;

[0063] S55. According to the spatial autocorrelation index I improved , mark / connect the regions with significant positive / negative correlations, display the aggregation and association patterns of well-being in space, and obtain the spatial mapping result of well-being.

[0064] Furthermore, in the step S51, the element d in the distance-based spatial weight matrix W represents the degree of association based on the geographical distance between region i and region j, which is expressed as:

[0065]

[0066] In the formula, θ represents the distance decay parameter, and d ij represents the geographical distance between region i and region j;

[0067] The weight matrix W based on the economic connection strength d is expressed as:

[0068]

[0069] In the formula, represents the trade volume index between region i and region j, represents the investment flow index between region i and region j, It represents the industrial structure similarity index between region i and region j, and α, β, and γ represent the weight coefficients of each index;

[0070] In step S53, the spatial autocorrelation index I improved is expressed as:

[0071]

[0072] In the formula, x i and x j respectively represent the values of the residents' well-being indicators in region i and region j, represents the mean value of the residents' well-being indicators, and W ij is an element in the composite spatial weight matrix W;

[0073] Among them, I improved > 0 indicates that the residents' well-being shows positive spatial correlation, that is, high-value regions tend to be adjacent to high-value regions, and low-value regions tend to be adjacent to low-value regions. When I improved < 0 indicates negative spatial correlation, that is, high-value regions are adjacent to low-value regions. When I improved = 0 indicates spatial random distribution.

[0074] The beneficial effects of the present invention are as follows:

[0075] (1) In the present invention, a convolutional neural network is used to mine the implicit feature factors of remote sensing of residents' well-being and realize factor optimization, breaking through the technical problem that it is difficult to quantitatively extract the original remote sensing features of residents' well-being;

[0076] (2) The present invention uses statistical data to construct multi-index residents' well-being data, and uses the entropy weight analysis method to determine the weights of each index to realize the quantitative calculation of residents' well-being;

[0077] (3) In the present invention, a random forest model is used to construct a spatial simulation model of residents' well-being and remote sensing feature indicators, and a downscaling method is further used to realize the spatial simulation and mapping of residents' well-being.

[0078] (4) In the present invention, an improved spatial autocorrelation analysis method is used to explore the evolution characteristics and spatial mapping of large-scale residents' well-being. This method comprehensively considers the influence of geographical spatial proximity and economic connection on the spatial distribution of residents' well-being. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1 It is a flow chart of the multi-remote sensing spatial mapping method for large-scale residents' well-being provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0080] The following describes the specific embodiments of the present invention to facilitate those skilled in the art of this technology to understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art of this technology, as long as various changes are within the spirit and scope of the present invention defined and determined by the appended claims, these changes are obvious, and all inventions created using the concept of the present invention are within the scope of protection.

[0081] An embodiment of the present invention provides a multi-source remote sensing spatial mapping method for large-scale resident well-being, as Figure 1 shown, including the following steps:

[0082] S1. Collect multi-source data of the resident well-being space at the county scale and preprocess it;

[0083] S2. Extract remote sensing features related to resident well-being according to the preprocessed data;

[0084] The remote sensing features include night light intensity features, impervious surface index features, vegetation index features, road density index features, point of interest data features, and remote sensing implicit features;

[0085] S3. Collect national statistical yearbook data at the county scale, conduct weighted statistics on it, and quantitatively determine the resident well-being value;

[0086] S4. Combine the remote sensing features and the resident well-being value to construct a random forest model for predicting the resident well-being value, map the prediction result of the resident well-being value from the county scale to the pixel scale through a downscaling function, and combine the remote sensing feature sequences of different periods to obtain the evaluation results of the resident well-being value at different times on a large scale;

[0087] S5. Based on the evaluation results of the resident well-being value, calculate the spatial autocorrelation index of different regions, and draw a spatial distribution map of the resident well-being according to it to show the aggregation and association patterns of the resident well-being in space.

[0088] In step S1 of the embodiment of the present invention, the multi-source data includes multi-source optical remote sensing data, night light data, located point of interest data, and statistical yearbook data;

[0089] The preprocessing includes:

[0090] The preprocessing of the multi-source optical remote sensing data includes radiometric calibration, atmospheric correction, geometric correction, band registration, multi-band image synthesis, and image fusion of the multi-source optical remote sensing images;

[0091] The preprocessing of the night light data includes radiometric calibration, atmospheric correction, geometric correction, and data clipping and splicing;

[0092] The preprocessing of the located point-of-interest data includes duplicate data removal, data cleaning, coordinate system unification, and data classification.

[0093] Specifically, in this embodiment, for multi-source optical remote sensing data, optical remote sensing data at the county scale such as Landsat 8, Gaofen-1, Gaofen-6, and Sentinel-2 are acquired. The optical remote sensing data selected are cloud-free and snow-ice-free data. Landsat 8, Gaofen-1, Gaofen-6, and Sentinel-2 and other optical remote sensing data include 15-meter multispectral images, 2-meter panchromatic / 8-meter multispectral, 10-meter, 20-meter, 60-meter and other multispectral images; after the images are acquired, preprocessing such as radiometric calibration, atmospheric correction, geometric correction, band registration, multi-band image synthesis, and image fusion is performed.

[0094] For night-time light data, night-time light data at the county scale are acquired. SDGSAT-1 carries a low-light and multispectral imager, with 1 panchromatic band and 3 color bands, including night-time light images with a resolution of 30 meters; after the images are acquired, preprocessing such as radiometric calibration, atmospheric correction, geometric correction, data cropping and stitching is performed.

[0095] For the located point-of-interest data, POI (point-of-interest) data are collected through an online map data platform and mobile device positioning, including data information on various commercial facilities, public service facilities, industrial areas, and township service points within the county. The acquired data are preprocessed by removing duplicate data, data cleaning, coordinate system unification, data classification and statistics, etc.

[0096] For the statistical yearbook data, the China County Statistical Yearbook (county and city volume) is selected, which includes collecting social statistical yearbook data at the county scale, and information such as household registered population, gross regional product, local public budget revenue, household savings deposits, number of above-scale enterprises, and number of primary and secondary school students are used, and counties and cities with missing data are screened and removed.

[0097] In step S2 of the embodiment of the present invention, among the remote sensing features related to residents' well-being:

[0098] The night-time light intensity features in this embodiment include the total night-time light intensity, the average night-time light intensity, and the standard deviation of the night-time light intensity;

[0099] Among them, the night-time light intensity values of all valid pixel points in the preprocessed data of the SDGSAT-1 night-time light remote sensing of the research county are accumulated to obtain the total night-time light intensity (TNL):

[0100]

[0101] In the formula, DN i is the i-th gray level, C iis the number of pixels corresponding to this gray level, and n is the maximum gray level.

[0102] Divide the total value of the county - level night - light intensity by the total number of effective pixels to obtain the average night - light intensity (NLM):

[0103]

[0104] In the formula, C i is the number of pixels corresponding to this gray level, and n is the maximum gray level.

[0105] The standard deviation of night - light intensity NTSD represents the difference in the spatial distribution of night - lights. Calculate the square of the difference between the night - light intensity value of each pixel and the average night - light intensity. Accumulate the squared differences of all pixels, then divide by the total number of effective pixels to obtain the variance, and take the square root of the variance to get the standard deviation of night - light intensity (NTSD):

[0106]

[0107] In the formula, DN i is the i - th gray level, and NTM is the average night - light intensity.

[0108] The impervious surface index feature in this embodiment is the normalized difference impervious surface index obtained by operating on the reflectances of the short - wave infrared band, near - infrared band, green band, and mid - infrared band. Specifically, based on the pre - processed high - resolution satellite images containing visible light, near - infrared, etc. and Landsat series satellite data obtained above, extract the impervious surface index feature at the county - level scale. Use the normalized difference impervious surface index (NDISI), and by operating on the reflectances of the short - wave infrared band (SWIR), near - infrared band (NIR), green band (Green), and mid - infrared band 1 (MIR1), make good use of the reflection differences of different bands for different ground objects to highlight the characteristics of the impervious surface, so as to effectively extract the impervious surface, which is expressed as:

[0109]

[0110] Among them, SWIR is the reflectance of the short - wave infrared band, NIR is the reflectance of the near - infrared band, Green is the reflectance of the green band, and MIR1 is the reflectance of the mid - infrared band 1.

[0111] In this embodiment, the vegetation index features include the normalized vegetation index and the enhanced vegetation index. Specifically, based on the Landsat series satellite data and Sentinel - 2 satellite data containing visible light, near - infrared, etc. obtained after pre - processing above, perform county - level scale feature extraction, mainly including the normalized vegetation index (NDVI) and the enhanced vegetation index (EVI).

[0112] Among them, the Normalized Difference Vegetation Index (NDVI) is calculated by dividing the difference between the reflectance in the near-infrared band (NIR) and the reflectance in the red band (R) by the sum of the two. Utilizing the characteristics of vegetation with high reflectance in the near-infrared band and strong absorption in the red band, the difference in vegetation between these two bands is highlighted, thus being able to effectively reflect the growth status and coverage degree of vegetation, and it is expressed as:

[0113] NVDI = (NIR - R) / (NIR + R)

[0114] In the formula, NIR represents the reflectance in the near-infrared band, and R represents the reflectance in the red band.

[0115] In addition to considering the near-infrared and red bands, the Enhanced Vegetation Index (EVI) also introduces the reflectance in the blue band (B). The numerator part amplifies the difference in vegetation between the near-infrared and red bands by a factor of 2.5, making the vegetation signal more obvious. The 6R term emphasizes the sensitivity of the red band to the absorption of vegetation chlorophyll, and the -7.5B term is used to correct the influence of the blue band to reduce interference from factors such as atmospheric scattering. Finally, adding 1 avoids the denominator being zero. EVI has a larger dynamic range in high-vegetation-coverage areas and low-vegetation-coverage areas and can better reflect the subtle changes in vegetation. The calculation formula is as follows:

[0116] EVI = 2.5×(NIR - R) / (NIR + 6R - 7.5B + 1)

[0117] Among them, NIR represents the reflectance in the near-infrared band, R represents the reflectance in the red band, and B represents the reflectance in the blue band.

[0118] The road density index in this embodiment is characterized by the proportion of the road length per unit area; specifically, after collecting high-resolution remote sensing images and related geographic information data and preprocessing them, the road density index (RDI) uses methods such as image classification, edge detection, or object-oriented methods to extract road information, and obtains the proportion of the road length per unit area, which can intuitively display the density of roads within a county; it is expressed as:

[0119] RDI = total road length / research area

[0120] The POI data in this embodiment is characterized by the proportion of the number of each type of POI in the region; specifically, at the county scale, the number of POIs such as companies, tourist attractions, hotels, restaurants, etc. is counted, and its distribution uniformity is calculated, which is expressed as:

[0121] POI = statistical quantity / research area

[0122] The remote sensing latent features in this embodiment are the feature information representing the inherent attributes, physical processes or human activities related to the well-being of residents extracted from remote sensing images. Among them, the feature information is extracted from remote sensing images through advanced data processing technologies (such as machine learning, deep learning or statistical analysis).

[0123] In this embodiment, the method for extracting the remote sensing latent features is specifically as follows:

[0124] The remote sensing images in the study area are cropped and normalized, and the processed input feature map is input into the trained convolutional neural network to obtain the remote sensing latent features.

[0125] Specifically, operations such as cropping and normalization are performed on the county-scale remote sensing image data to make it suitable for input into the convolutional neural network. Let the input remote sensing image be I(x,y), where x and y are the pixel coordinates of the image. The normalization operation can be expressed as:

[0126]

[0127] In the formula, μ is the mean of all pixel values of the image, and σ is the standard deviation.

[0128] The convolutional neural network includes an input layer, several convolutional layers, a pooling layer and an output layer connected in sequence;

[0129] Among them, the output feature map O l (i,j,c) of the l-th convolutional layer is:

[0130]

[0131] In the formula, w l (m,n,d,c) represents the weight of the c-th convolutional kernel at the position (m,n) in the l-th convolutional layer corresponding to the d-th input channel, and b l (c) represents the bias of the c-th convolutional kernel in the l-th convolutional layer, D l represents the number of channels of the input feature map I l , i and j are the coordinates of the output feature map, s l represents the convolutional step size, k l ×k l represents the size of the convolutional kernel in the l-th convolutional layer, and D l represents the number of input channels;

[0132] The output feature map p l (i,j,c) of the pooling layer is:

[0133]

[0134] In the formula, p l ×pl Denote the pooling window, Denote the pooling stride;

[0135] After a series of convolutional layers and pooling layers, a feature map tensor F is obtained, with dimensions h×w×d (height h, width w, and depth d). Flatten it into a vector vec(F) with dimensions hwd; design a fully connected layer as the output layer, with a weight matrix W fc , with dimensions n×hwd (n is the number of implicit socio-economic features to be extracted), to obtain n remote sensing implicit features C output by the output layer n Denoted as:

[0136] C n = W fc ×vec(F) + b fc

[0137] In the formula, W fc Denotes the weight matrix of the fully connected layer, b fc Denotes the bias vector of the fully connected layer, vec(F) denotes the vector obtained by flattening the output feature map F of the pooling layer, with dimensions hwd, and h, w, and d respectively represent the height, width, and depth of the output feature map

[0138] Step S3 in the embodiment of the present invention includes the following sub-steps:

[0139] S31. Collect the statistical yearbook data at the county scale and select the statistical indicators related to residents' well-being;

[0140] S32. Standardize the selected statistical indicators by the range normalization method;

[0141] S33. Determine the weights of each statistical indicator after standardization by the entropy weight method;

[0142] S34. Quantitatively calculate the residents' well-being value of the research area according to the weights of each statistical indicator and the statistical yearbook data

[0143] In step S31 of this embodiment, the selected statistical indicators related to residents' well-being include statistical indicators such as the registered population, gross regional product, number of enterprises above designated size, and number of students in primary and secondary schools

[0144] In step S32 of this embodiment, the above statistical indicators are standardized. The jth indicator of the ith county is denoted as x ij , and the standardization formula is:

[0145]

[0146] In the formula, y ij Is the indicator value after the range normalization method

[0147] In step S33 of this embodiment, the entropy weight method is an objective weighting method. It determines the weight according to the information entropy of the index, and calculates the entropy value e based on the index value of the range normalization method j , and then calculates the difference coefficient g j , and determines the weight w of each index j . The smaller the information entropy, the greater the amount of information provided by the index, the greater the difference coefficient, and the greater the weight.

[0148] Among them, the entropy value is:

[0149]

[0150] e j =-k∑p ij ln(p ij )

[0151]

[0152] Among them, y ij is the index value after the range normalization method, n is the number of counties for the evaluation of residents' well-being, and k is a constant related to the number of counties n.

[0153] The difference coefficient is:

[0154] g j =1 - e j

[0155] Calculate the weight of each index:

[0156]

[0157] In the formula, m is the number of indexes.

[0158] In step S34 of this embodiment, the residents' well-being value at the county scale is calculated as:

[0159] WB i =x i,1 ×ω1 + x i,2 ×ω2……x i,j ×ω j

[0160] In the formula, WB i is the residents' well-being value of the i-th county, x ij is the j-th index value of the i-th county, and ω j is the weight value of the j-th index.

[0161] Step S4 of the embodiment of the present invention includes the following sub-steps:

[0162] S41. Calculate the mean and variance of various remote sensing characteristics in the county as the remote sensing characteristic values corresponding to the resident welfare value;

[0163] S42. Based on the remote sensing feature values and resident well-being values at the county scale, a random forest model is constructed, and a downscaling function is used to map the resident well-being values at the county scale to the pixel scale.

[0164] S43. Based on the predicted values of resident welfare values at the pixel scale, remote sensing feature sequences of different periods are used to explore the dynamic evolution of resident welfare values over time in a large-scale range by setting quantitative indicators, and the evaluation results of resident welfare values in different periods at a large scale are obtained.

[0165] In step S41 of this embodiment, the mean of each remote sensing feature is calculated. and variance s i 2 for:

[0166]

[0167] Where x ij It represents the value of the i-th remote sensing characteristic index at the j-th pixel, and m is the number of pixels in the county.

[0168] Step S42 of this embodiment includes the following sub-steps:

[0169] S42-1. Construct dataset D based on the resident welfare values and remote sensing feature values at the county level;

[0170] Among them, the data set D contains several subsets (D1, D2, ..., D t ), each subset represents the remote sensing feature values and resident welfare values corresponding to the resident welfare values of different counties, D t ={Y t ,X t1 ,X t2 ,…,X ti ,…,X tn}; where Y t represents the welfare value of residents in the tth subset, X t1 ,X t2 ,…,X ti ,…,X tn Represents Y t The corresponding 1st to nth remote sensing feature values;

[0171] S42-2. For each subset, construct a corresponding decision tree;

[0172] In the process of decision tree construction, p remote sensing feature values are randomly selected from the n remote sensing feature values in the subset at each node, and the corresponding information gain Gain (D, A) is:

[0173]

[0174] In the formula, the information gain Gain(D, A) is used to measure the contribution degree of the remote sensing feature value to the data set D, and Entropy(D v ) represents the entropy of the data set, which is used to measure the degree of chaos of the data set. A represents the selected remote sensing feature value, values(A) represents the set of all possible values of the remote sensing feature A, and D v represents the sample subset in the data set where the remote sensing feature A takes the value v. represents the proportion of the sample subset where the remote sensing feature A takes the value v in the data set D in the data set D;

[0175] Among them;

[0176]

[0177] In the formula, C is the set of sample categories, and p c represents the proportion of each type of sample in C;

[0178] S42-3. Combine the constructed k decision trees into a random forest model, input the remote sensing feature values in the area to be simulated into the random forest model, and perform regression on the prediction results corresponding to each decision tree to obtain the predicted value of the resident well-being value;

[0179] After the decision tree is constructed, combine the k decision trees into a random forest model. For the new data point x, its predicted value is determined comprehensively by the prediction results of each decision tree in the random forest. For regression problems, use:

[0180]

[0181] Among them, k represents the number of sample subsets, that is, the number of decision trees; y t is the prediction result of the t-th decision tree for the new data point. S42-4. Based on the association between the remote sensing feature values at the county scale and the pixel scale, construct a downscaling function f(x);

[0182] Among them, the construction of the downscaling function fully considers the internal association between the county-scale results and the remote sensing feature index values at the pixel scale, and can be determined by means of advanced technical means such as statistical relationships and spatial interpolation. For example, the downscaling method based on geographically weighted regression can achieve accurate conversion from the county scale to the pixel scale by estimating the location-related regression coefficients, so as to achieve high-resolution evaluation of the resident well-being space at the pixel scale.

[0183] S42-5. Use the downscaling function f(x) to map the predicted value of the resident well-being at the county scale to the pixel scale;

[0184] Among them, the predicted value Y of the well-being of residents at the pixel scale of the j-th pixel pixel is expressed as:

[0185] Y pixel (j) = f(Y county , x 1j , x 2j , …, x ij , …x nj ))

[0186]

[0187] In the formula, f(·) represents the downscaling function, x 1j , x 2j , ……x nj represents the remote sensing feature value of the j-th pixel, Y county represents the well-being value of residents at the county scale, β0(j) represents the regression coefficient related to pixel j, x ij represents the i-th remote sensing feature value corresponding to the j-th pixel position.

[0188] In step S43 of this embodiment, for the remote sensing data sequences of different periods, the well-being assessment results corresponding to each period are obtained according to the random forest model, and the dynamic evolution law of the well-being of residents over time is further deeply explored. By calculating quantitative indicators such as the change rate, the temporal change trend of the well-being of residents within a large scale is visually presented, providing scientific data support and decision-making basis for long-term planning and policy formulation. For the remote sensing data in the t-th period, the above steps are repeated to obtain the well-being assessment results y t of different periods, and the change rate r t is calculated:

[0189]

[0190] In step S5 of the embodiment of the present invention, an improved spatial autocorrelation analysis method is adopted to explore the evolution characteristics and spatial mapping of the well-being of residents on a large scale. This method comprehensively considers the influence of geographical spatial proximity and economic connection on the spatial distribution of the well-being of residents; considering the complexity and multi-source nature of the well-being of residents data, not only the classical distance-based spatial weight matrix is introduced, but also the weight matrix based on the economic connection strength is combined.

[0191] Based on this, step S5 of the embodiment of the present invention includes the following sub-steps:

[0192] S51. Set the distance-based spatial weight matrix W d and the economic connection strength-based weight matrix W e ;

[0193] Among them, the distance-based spatial weight matrix W d Elements in It represents the degree of association between region i and region j based on geographical distance, which is expressed as:

[0194]

[0195] Where θ represents the distance attenuation parameter, d ij represents the geographical distance between region i and region j;

[0196] The weight matrix W based on the strength of economic ties d Expressed as:

[0197]

[0198] Where, represents the trade volume indicator between region i and region j, represents the investment flow indicator between region i and region j, represents the similarity index of industrial structure between region i and region j, α, β, γ represent the weight coefficients of each index;

[0199] S52, the matrix W d and W e Compound, get the composite spatial weight matrix W = W d +W e ;

[0200] S53. Calculate the spatial autocorrelation index I between the welfare values of residents in different regions based on the composite spatial weight matrix W. i mprove d ;

[0201] Among them, the spatial autocorrelation index I improved Expressed as:

[0202]

[0203] Where x i and x j represent the resident welfare index values of region i and region j respectively, represents the mean of the resident welfare index, W ij is the element in the composite spatial weight matrix W;

[0204] Among them, I improved >0 indicates that residents’ welfare is spatially positively correlated, that is, high-value areas tend to be adjacent to high-value areas, and low-value areas tend to be adjacent to low-value areas. improved <0 indicates negative spatial correlation, that is, high-value areas are adjacent to low-value areas. improved= 0 represents a spatial random distribution;

[0205] S54. Divide the research area into grids according to geographical coordinates, and perform hierarchical rendering on the grids according to the magnitude of the resident well-being value;

[0206] S55. According to the spatial autocorrelation index I improved , mark / connect the regions with significant positive / negative correlations, display the aggregation and correlation patterns of resident well-being in space, and obtain the spatial mapping result of resident well-being;

[0207] Specifically, perform hierarchical rendering on the grids according to the magnitude of the resident well-being index value. For example, divide the resident well-being index value into three levels: high, medium, and low, and display them with different colors (such as red for high value, yellow for medium value, and blue for low value); use the results of the spatial autocorrelation index to specially mark or connect the regions with significant positive or negative correlations to visually display the aggregation and correlation patterns of resident well-being in space.

[0208] In this invention, specific embodiments are used to elaborate on the principle and implementation manner of the invention. The description of the above embodiments is only used to help understand the method and its core idea of the invention; at the same time, for those of ordinary skill in the art, based on the idea of the invention, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to the invention.

[0209] Those of ordinary skill in the art will realize that the embodiments described here are for helping the reader understand the principle of the invention, and it should be understood that the protection scope of the invention is not limited to such specific statements and embodiments. Those of ordinary skill in the art can make various other specific deformations and combinations that do not depart from the essence of the invention based on the technical revelations disclosed in this invention, and these deformations and combinations are still within the protection scope of the invention.

Claims

1. A large-scale multi-source remote sensing spatial mapping method for resident well-being, characterized by: The following steps are involved: S1. Collect and pre-process multi-source data on residents’ well-being at the county level; S2. Extract remote sensing features related to residents’ well-being based on the preprocessed data; The remote sensing features include nighttime light intensity features, impervious surface index features, vegetation index features, road density index features, point of interest data features, and remote sensing latent features. The method for extracting the remote sensing latent features is specifically as follows: The remote sensing images in the study area are cropped and normalized, and the input feature maps obtained are input into the trained convolutional neural network to obtain the remote sensing latent features; S3. Collect national statistical yearbook data at the county level, perform weighted statistics on it, and quantitatively determine the resident welfare value, including the following steps: S31. Collect statistical yearbook data at the county level and select statistical indicators related to residents' well-being; S32. Standardize the selected statistical indicators by using the range standardization method; S33. Determine the weight of each statistical indicator after standardization by entropy weight method; S34. Quantitatively calculate the resident welfare value of the study area based on the weights of each statistical indicator and statistical yearbook data; S4. Combining remote sensing features and resident well-being values, a random forest model was constructed to predict resident well-being values. The predicted resident well-being values were mapped from the county scale to the pixel scale using a downscaling function. Combined with remote sensing feature sequences from different periods, the resident well-being values at different times were assessed at a large scale. S5. Based on the results of the resident well-being assessment, calculate the spatial autocorrelation index of different regions and draw a spatial distribution map of resident well-being based on it to show the spatial aggregation and correlation patterns of resident well-being, including the following steps: S51. Set the distance-based spatial weight matrix and a weight matrix based on the strength of economic ties ; S52, the matrix and Compound to obtain the composite spatial weight matrix ; S53, according to the composite spatial weight matrix , calculate the spatial autocorrelation index between the welfare values of residents in different regions ; S54. Divide the study area into grids according to geographic coordinates, and render the grids in different levels according to the residents' well-being values; S55, according to the spatial autocorrelation index , mark / connect areas with significant positive / negative correlation, show the spatial aggregation and correlation patterns of residents' well-being, and obtain the results of resident well-being spatial mapping.

2. The large-scale multi-source remote sensing spatial mapping method for resident well-being according to claim 1 is characterized in that: In step S1, the multi-source data includes multi-source optical remote sensing data, nighttime light data, point of interest data, and statistical yearbook data; The pretreatment includes: The preprocessing of multi-source optical remote sensing data includes radiometric calibration, atmospheric correction, geometric correction, band registration, multi-band image synthesis and image fusion. The preprocessing of night light data includes radiometric calibration, atmospheric correction, geometric correction, and data clipping and splicing; The preprocessing of POI data includes duplicate data removal, data cleaning, coordinate system unification and data classification.

3. The large-scale multi-source remote sensing spatial mapping method for resident well-being according to claim 1 is characterized in that: In step S2, the nighttime light intensity characteristics include the total nighttime light intensity, the mean nighttime light intensity, and the standard deviation of the nighttime light intensity; The impervious surface index characteristic is a normalized difference impervious surface index obtained by calculating the reflectivity of the short-wave infrared band, the near-infrared band, the green band, and the mid-infrared band; The vegetation index features include normalized vegetation index and enhanced vegetation index; The road density index characteristic is the ratio of road length per unit area; The point of interest data feature is the proportion of each type of point of interest in the area; The remote sensing latent features are characteristic information extracted from remote sensing images that represent intrinsic properties related to residents' well-being, physical processes or human activities.

4. The large-scale multi-source remote sensing spatial mapping method for resident well-being according to claim 3 is characterized in that: The convolutional neural network includes an input layer, several convolutional layers, a pooling layer and an output layer connected in sequence; No. l The output feature map of the convolutional layer for: Where, Indicates the In the convolutional layer The convolution kernel is at position Corresponding to The weights of the input channels, Indicates the In the convolutional layer The bias of the convolution kernel, Represents the input feature map The number of channels, are the coordinates of the output feature map, represents the convolution step size, Indicates the The convolution kernel size in the convolution layer, Indicates the number of input channels; The output feature map of the pooling layer for: Where, represents the pooling window, represents the pooling step size; The output layer is a fully connected layer, and its output n Remote sensing recessive features Expressed as: Where, represents the weight matrix of the fully connected layer, represents the bias vector of the fully connected layer, Indicates that the output feature map of the pooling layer The vector obtained by flattening has the dimension , 、 and Represent the height, width and depth of the output feature map respectively.

5. The large-scale multi-source remote sensing spatial mapping method for resident well-being according to claim 1 is characterized in that: The step S4 comprises the following sub-steps: S41. Calculate the mean and variance of various remote sensing characteristics in the county as the remote sensing characteristic values corresponding to the resident welfare value; S42. Based on the remote sensing feature values and resident well-being values at the county scale, a random forest model is constructed, and a downscaling function is used to map the resident well-being values at the county scale to the pixel scale. S43. Based on the predicted values of resident welfare values at the pixel scale, remote sensing feature sequences of different periods are used to explore the dynamic evolution of resident welfare values over time in a large-scale range by setting quantitative indicators, and the evaluation results of resident welfare values in different periods at a large scale are obtained.

6. The large-scale multi-source remote sensing spatial mapping method for resident well-being according to claim 5 is characterized in that: The step S42 includes the following sub-steps: S42-1. Constructing a dataset based on county-level resident welfare values and remote sensing feature values ; The dataset D Contains several subsets ( ), each subset represents the remote sensing feature values and resident welfare values corresponding to the resident welfare values of different counties, ;in, Indicates the t The welfare value of residents in the subset, express Corresponding to 1st~ n remote sensing feature values; S42-2. For each subset, construct a corresponding decision tree; S42-3, will be constructed k A random forest model is composed of decision trees. The remote sensing feature values in the simulated area are input into the random forest model, and the prediction results corresponding to each decision tree are regressed to obtain the predicted value of the resident welfare value. S42-4. Construct a downscaling function based on the correlation between remote sensing feature values at the county scale and pixel scale ; S42-5. Using downscaling function , mapping the predicted values of resident welfare at the county scale to the pixel scale.

7. The large-scale multi-source remote sensing spatial mapping method for resident well-being according to claim 6 is characterized in that: In the step S42-2, during the decision tree construction process, at each node, n Randomly select from the remote sensing feature values p Remote sensing feature values, and their corresponding information gain for: In the formula, information gain A dataset for measuring remote sensing feature value pairs D The contribution of Represents the entropy of a data set, which is used to measure the degree of disorder of the data set. Indicates the selected remote sensing feature value, Representing remote sensing features The set of all possible values of Represents remote sensing features in the dataset A The value is v A subset of samples, Indicates in the dataset D Remote sensing features A The value is v The sample subset accounts for the dataset D proportion; In the step S42-5, Pixel-by-pixel prediction of resident welfare Expressed as: Where, represents the downscaling function, Indicates the The remote sensing feature value of the pixel, represents the well-being value of residents at the county level, Representation and Pixels The relevant regression coefficients, Indicates the representative The pixel position corresponding to the remote sensing feature values.

8. The large-scale multi-source remote sensing spatial mapping method for resident well-being according to claim 1 is characterized in that: In step S51, the spatial weight matrix based on the distance Elements in Indicates the display area i and region j The degree of association between them based on geographical distance is expressed as: Where, represents the distance attenuation parameter, Indicates area i and region j geographical distance between them; Weight matrix based on economic connection strength Expressed as: Where, Indicates area i and region j The trade volume indicator between Indicates area i and region j The investment flow indicator between Indicates area i and region j The similarity index of industrial structure between Represents the weight coefficient of each indicator; In step S53, the spatial autocorrelation index Expressed as: Where, and Represents the area and region The resident welfare index value, represents the mean value of the resident welfare index, is the composite spatial weight matrix Elements in in, Indicates that residents' welfare is spatially positively correlated, that is, high-value areas tend to be adjacent to high-value areas, and low-value areas tend to be adjacent to low-value areas. Indicates spatial negative correlation, that is, high-value areas are adjacent to low-value areas. Indicates random distribution in space.

Images