A village feature recognition method and device based on multi-source data

Abstract

The application discloses a village feature recognition method and device based on multi-source data, and belongs to the technical field of image recognition and artificial intelligence. The method comprises the following steps: generating a first population outflow characteristic value of a target village based on population census data; generating a second population outflow characteristic value of the target village based on a night light remote sensing image of the target village; generating a third population outflow characteristic value of the target village based on electricity consumption data; calculating a population outflow comprehensive characteristic value by adopting a multi-target linear weighting function method based on the first population outflow characteristic value, the second population outflow characteristic value and the third population outflow characteristic value; and identifying the hollowing degree of the village based on the population outflow comprehensive characteristic value. The application improves the accuracy of the hollowing degree recognition of the village, and the method is simple and fast.

Description

Technical Field

[0001] This invention belongs to the field of image recognition and artificial intelligence technology, specifically relating to a method and apparatus for village feature recognition based on multi-source data, used to identify the degree of hollowing out in rural areas. Background Technology

[0002] The key to identifying "hollow villages" is quantitatively determining the degree of population outflow from rural areas. Due to the difficulty in tracking the dynamics of the permanent resident population, the lack of timely and efficient statistical methods in rural areas, and the incompleteness of statistical data, identifying "hollow villages" has always been a challenging problem.

[0003] With the rapid development of "3S" technologies (sun, water, and air), especially the emergence of nighttime light data, important references have been provided for detecting population changes. Numerous studies have applied remote sensing nighttime light data to population spatial distribution and movement patterns. Such remote sensing nighttime light data includes DMSP / OLS, Suomi NPP / VIIRS, and the VIIRS-based Huoshi Earth Night Light product. Furthermore, nighttime light data can characterize the intensity of human activity using brightness, detecting different intensities of light generated by cities, small residential areas, and traffic flow, making it an effective data source for monitoring human activity.

[0004] By using big data on electricity consumption to measure the degree of rural hollowing-out, a supporting basis can be provided for the construction of public facilities such as schools and roads, as well as for the optimization of market and industrial allocation. For example, Chinese patent document CN111861257 discloses a method for identifying hollow villages based on electricity data heat maps. This method generates a village heat map based on the target village's electricity data and boundary information, and uses a machine learning intelligent model to process the village heat map to obtain the hollow village identification results.

[0005] However, using single data points to identify hollow villages still has certain limitations. Census data has a specific temporal cross-section effect and a long data update cycle. Compared with urban areas, nighttime light data at the village and town level generally has smaller values, smaller illuminated areas, and is difficult to directly detect. Moreover, village and town-level lights are difficult to distinguish from unlit areas, meaning that nighttime light data at the village and town level has a low signal-to-noise ratio, making it difficult to identify. In addition, for hollow villages, only a small number of villagers remain, or those unable to relocate due to economic difficulties, leaving the original power lines and equipment running almost unused, resulting in serious waste, and limiting the accuracy of power big data. Summary of the Invention

[0006] The purpose of this invention is to provide a village feature identification method and device based on multi-source data, which combines traditional population census data with remote sensing nighttime light data and power big data to more accurately measure population outflow and comprehensively and accurately identify the degree of village hollowing out.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] A first aspect of this invention discloses a village feature identification method based on multi-source data, comprising:

[0009] Generate the first population outflow characteristic value for the target village based on census data;

[0010] Generating a second population outflow feature value for the target village based on nighttime light remote sensing images of the target village;

[0011] Generate a third population outflow characteristic value for the target village based on electricity consumption data;

[0012] The comprehensive characteristic value of population outflow is calculated using a multi-objective linear weighted function method based on the first, second, and third population outflow characteristic values.

[0013] Identifying the degree of village hollowing out based on comprehensive characteristics of population outflow;

[0014] Specifically, the second population outflow characteristic value of the target village is generated based on the remote sensing image of the target village's nighttime lights, including corrections to the nighttime light data:

[0015] Based on the conditional function model, the annual-scale remote sensing light data is filtered to divide the target area into at least a first region and a second region. The first region is the area in the target area where the light radiance of the annual-scale remote sensing light data is greater than or equal to a preset threshold, and the second region is the area in the target area where the light radiance of the annual-scale remote sensing light data is less than a preset threshold.

[0016] Data reconstruction was performed based on the annual-scale remote sensing light data corresponding to the first region and after conditional filtering, to obtain the annual-scale remote sensing light data after dark-eye enhancement processing.

[0017] In other embodiments, generating a second population outflow feature value for the target village based on a nighttime light remote sensing image of the target village further includes:

[0018] Using GIS application software tools, rural settlements were extracted as the boundaries of densely populated areas. Using the boundaries as a mask, the corrected nighttime light data was extracted.

[0019] Based on the annual-scale remote sensing light data reconstructed from the extracted data, the population regression model based on nighttime light data is constructed as a cubic polynomial function expression of population and nighttime light radiance.

[0020] Based on the regression results of rural permanent residents using nighttime light data, a second population outflow characteristic value based on nighttime light data is calculated.

[0021] In other embodiments, the latitude and longitude information of each residential electricity consumption data acquisition station is obtained using an electronic map API. Point data is generated in the GIS application software tool based on the latitude and longitude of each station. Electricity consumption data is input and the point data is matched with village surface data. Electricity consumption data is summarized by village.

[0022] The population regression model based on electricity consumption data is constructed as a cubic polynomial function expression of population and electricity consumption values;

[0023] Based on the regression results of rural permanent residents using electricity consumption data, a third population outflow characteristic value based on electricity consumption data is calculated.

[0024] In other embodiments, the calculation of the comprehensive characteristic value of population outflow using the multi-objective linear weighted function method includes: determining the weights of the first population outflow characteristic, the second population outflow characteristic, and the third population outflow characteristic using the entropy weight method.

[0025] A second aspect of the present invention discloses a village feature identification device based on multi-source data, comprising:

[0026] The first feature extraction module is used to generate the first population outflow feature value of the target village based on census data;

[0027] The second feature extraction module is used to generate a second population outflow feature value of the target village based on the remote sensing image of the target village's nighttime lights.

[0028] The third feature extraction module is used to generate the third population outflow feature value of the target village based on electricity consumption data;

[0029] The feature fusion module is used to calculate the comprehensive feature value of population outflow based on the first, second, and third population outflow feature values ​​using a multi-objective linear weighted function method.

[0030] The identification module is used to identify the degree of hollowing out of villages based on comprehensive characteristics of population outflow;

[0031] The second feature extraction module includes a correction module, which is used to correct the nighttime light data, including:

[0032] Based on the conditional function model, the annual-scale remote sensing light data is filtered to divide the target area into at least a first region and a second region. The first region is the area in the target area where the light radiance of the annual-scale remote sensing light data is greater than or equal to a preset threshold, and the second region is the area in the target area where the light radiance of the annual-scale remote sensing light data is less than a preset threshold.

[0033] Data reconstruction was performed based on the annual-scale remote sensing light data corresponding to the first region and after conditional filtering, to obtain the annual-scale remote sensing light data after dark-eye enhancement processing.

[0034] A third aspect of the present invention also discloses a village feature recognition method based on multi-source data, comprising:

[0035] Generate the first population outflow characteristic value for the target village based on census data;

[0036] Generating a second population outflow feature value for the target village based on nighttime light remote sensing images of the target village;

[0037] Generate a third population outflow characteristic value for the target village based on electricity consumption data;

[0038] The comprehensive characteristic value of population outflow is calculated using a multi-objective linear weighted function method based on the first, second, and third population outflow characteristic values.

[0039] The degree of village hollowing out is identified by using trained machine learning models based on population outflow characteristics, population outflow characteristics based on nighttime light data, population outflow characteristics based on electricity consumption data, and comprehensive population outflow characteristics.

[0040] Specifically, the second population outflow characteristic value of the target village is generated based on the remote sensing image of the target village's nighttime lights, including corrections to the nighttime light data:

[0041] Based on the conditional function model, the annual-scale remote sensing light data is filtered to divide the target area into at least a first region and a second region. The first region is the area in the target area where the light radiance of the annual-scale remote sensing light data is greater than or equal to a preset threshold, and the second region is the area in the target area where the light radiance of the annual-scale remote sensing light data is less than a preset threshold.

[0042] Data reconstruction was performed based on the annual-scale remote sensing light data corresponding to the first region and after conditional filtering, to obtain the annual-scale remote sensing light data after dark-eye enhancement processing.

[0043] In other embodiments, the population regression model based on nighttime light data is constructed as a cubic polynomial function expression of population and nighttime light radiance. Based on the regression results of rural permanent residents based on nighttime light data, a second population outflow characteristic value based on nighttime light data is calculated. The population regression model based on electricity consumption data is constructed as a cubic polynomial function expression of population and electricity consumption values. Based on the regression results of rural permanent residents based on electricity consumption data, a third population outflow characteristic value based on electricity consumption data is calculated.

[0044] The beneficial effects of this invention are: it improves the accuracy of identifying the degree of village hollowing out, and the method is simple and quick.

[0045] Other features and advantages of the invention will become clear from the following detailed description of exemplary embodiments of the invention with reference to the accompanying drawings. Attached Figure Description

[0046] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0047] Figure 1 This is a schematic diagram illustrating the execution flow of a village feature recognition method based on multi-source data according to an embodiment of the present invention.

[0048] Figure 2 This is a schematic diagram comparing unprocessed and processed images of nighttime light data.

[0049] Figure 3 This is a schematic diagram illustrating the execution flow of a village feature recognition method based on multi-source data according to another embodiment of the present invention. Detailed Implementation

[0050] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0051] Figure 1 This is a schematic diagram illustrating the workflow of a village feature identification method based on multi-source data according to an embodiment of the present invention. Figure 1 As shown, the method includes the following steps:

[0052] Step S1: Generate the first population outflow characteristic value based on census data;

[0053] The survey was conducted by village, which was used to identify the survey subjects and obtain population data such as the village's permanent residents, registered residents, and migrant workers.

[0054] As an example, the population outflow index (P) based on census data is calculated using the following formula:

[0055]

[0056] In the formula, I G REP and RHP are population outflow indices calculated based on census data. REP represents the rural emigration population and the rural hukou-registered population, respectively.

[0057] Step S2: Generate a second population outflow feature value based on the remote sensing image of village nighttime lights;

[0058] Because village-level nighttime light data generally exhibits characteristics such as small numerical values, relatively small illuminated areas, and difficulty in distinguishing it from unlit areas, it is difficult to directly use nighttime light data to measure the village's permanent resident population. Therefore, this invention employs the following methods to correct the directly obtained nighttime light data, including:

[0059] (1) Conditional filtering of raw annual-scale remote sensing light data based on conditional function model;

[0060] For example, filtering can be performed using the following function model:

[0061]

[0062] In the formula, x represents the annual nighttime light data, x (1) This refers to the annual-scale remote sensing light data after conditional filtering.

[0063] The aforementioned conditional function model sets the maximum brightness value (illuminance) to 200 and the minimum brightness value (preset threshold) to 0.25. The maximum and minimum brightness values ​​are selected based on empirical thresholds obtained from statistical analysis of areas with perennial darkness. This invention, by conditionally filtering annual-scale remote sensing light data, can reduce the impact of perennially dark areas on annual-scale nighttime light data and the overall computational load while ensuring the accuracy of village-level resident population measurements.

[0064] (2) The target area with a value of 0 in the annual scale remote sensing light data after condition filtering is recorded as the second area and set as the NoDATA area, and the remaining areas are recorded as the first area.

[0065] This invention divides the target area into at least a first region and a second region. The first region consists of areas within the measurement area where the annual-scale remote sensing light data values ​​are greater than or equal to a preset threshold; these are the valid data areas requiring normal processing. The second region, designated as the NoDATA region, consists of areas within the measurement area where the annual-scale remote sensing light data values ​​are less than the preset threshold. By setting the NoDATA region (second region), noise interference from invalid data on the annual-scale data is effectively avoided.

[0066] (3) Data reconstruction is performed based on the annual-scale remote sensing light data corresponding to the first region and after conditional filtering, so as to obtain annual-scale remote sensing light data with enhanced dark-eye processing.

[0067] For example, the calculation formula corresponding to data reconstruction is:

[0068]

[0069] In the formula, x (2) For the reconstructed annual-scale remote sensing light data, x max The annual-scale remote sensing light data x after condition filtering (1) The maximum value in.

[0070] To validate the method for enhancing dark targets in nighttime light data, such as... Figure 2 As shown, the annual-scale remote sensing light data enhanced with dark targets is compared with the original data. Figure 2 (c) Figure 2 (d) is Figure 2 (a) Figure 2 (b) is an enlarged view of the area within the white box. The annual data obtained after the above processing can clearly show the dark targets at the village and town level.

[0071] Since rural populations are concentrated in rural settlements, this invention utilizes GIS application software tools, such as ArcGIS software extraction tools, to extract rural settlements as the boundaries of population-concentrated areas. Using these boundaries as masks, the corrected nighttime light data is extracted.

[0072] As an example, based on the annual-scale remote sensing light data reconstructed from the extracted data, this invention constructs a population regression model based on nighttime light data as a functional expression of population and nighttime light radiance:

[0073]

[0074] in, For the regression results of rural residents based on nighttime light data, x (2)The data represents the reconstructed annual-scale remote sensing light data, where x represents the nighttime light radiance, and a, b, c, and d are the coefficients of a cubic polynomial.

[0075] The goodness of prediction (G) can be used as a measure of prediction accuracy to test the model's accuracy. The formula for calculating G is as follows:

[0076]

[0077] In the formula, z(i) is the true value of the i-th data point. Let i be the predicted value of the i-th data point. This is the average of all the real data. k represents the number of samples used to calculate the prediction fit.

[0078] As can be seen from the formula for calculating the prediction accuracy, the closer the predicted value is to the actual value, the higher the prediction accuracy, and the closer the prediction accuracy is to 100%.

[0079] Based on the regression results of rural resident population using nighttime light data, this invention calculates the second population outflow characteristic value based on nighttime light data using the following formula:

[0080]

[0081] In the formula, I D This is the second population outflow characteristic value calculated based on nighttime light data. Population size for the starting year of a preset time period. The population at the end of the aforementioned preset time period.

[0082] Step S3: Generate third-party population outflow characteristic values ​​based on electricity consumption data;

[0083] In this invention, the statistical period for electricity consumption data is set to one month before and one month after the Spring Festival. Since the current rural outflow population is mainly migrant workers, the permanent resident population is relatively small except for the special time of the Spring Festival when families traditionally reunite. Therefore, the movement of permanent residents before and after the Spring Festival can reflect the outflow of the village population. The more serious the outflow of the village population, the higher the degree of hollowing out of the village.

[0084] Based on the residential electricity consumption data recorded by the power department within the research area, the monthly residential electricity consumption of each station was statistically analyzed, and the urban electricity consumption data was removed.

[0085] As an example, the latitude and longitude information of the electricity consumption data acquisition stations for each resident is obtained using an electronic map API (such as Amap). Point data is generated in ArcGIS software based on the latitude and longitude of each station, while simultaneously inputting the electricity consumption data. Finally, the point data is matched with village area data to summarize the electricity consumption data on a village-by-village basis.

[0086] As an example, this invention constructs a population regression model based on electricity consumption data as a functional expression of population and electricity consumption values:

[0087]

[0088] in, The regression results for rural permanent residents are based on electricity consumption data, where y is the electricity consumption value, and a, b, c, and d are the coefficients of a cubic polynomial.

[0089] Similarly, the prediction accuracy is used as the prediction precision to test the model's accuracy.

[0090] Based on the regression results of rural permanent residents using electricity consumption data, this invention calculates the third population outflow characteristic value based on electricity consumption data using the following formula:

[0091]

[0092] In the formula, I E These are population outflow characteristic values ​​calculated based on electricity consumption data. The population size for the starting month of the preset time period. The population size for the end month of the aforementioned preset time period.

[0093] Step S4: Calculate the comprehensive characteristic value of population outflow using the multi-objective linear weighted function method;

[0094] This invention is based on the population outflow characteristic value (I) of census data. G Population outflow characteristics based on nighttime light data (I) D Population outflow characteristics based on electricity consumption data (I) E ) Calculate the characteristic values ​​of population outflow.

[0095] Since the values ​​of different population outflow characteristics have significantly different dimensions, in order to eliminate the resulting shielding effect, this invention first performs data standardization processing.

[0096] For example, the standardized formula is as follows:

[0097]

[0098]

[0099] In the formula, Y ij I represents the standardized value of the j-th sample for the i-th feature; ij This represents the original data of the j-th sample with the i-th feature; min{I i} represents the minimum value of the i-th feature; max{I i} represents the maximum value of the i-th feature; 0 ≤ Y ij ≤1, and all are dimensionless. Positive features are standardized using formula (1), and negative features are standardized using formula (2).

[0100] This invention employs a multi-objective linear weighted function method to calculate the comprehensive characteristic value of population outflow:

[0101]

[0102] In the formula, C represents the comprehensive characteristic value of population outflow; Y ij For each feature, the standardized score is W. i Let be the weight of feature i, and n be the number of samples.

[0103] The weights of each feature are determined using the entropy weight method, which can objectively determine the weights with high accuracy and avoids the influence of subjective factors. Specifically, this includes:

[0104] (1) Define the normalization matrix;

[0105] The data standardization matrix is ​​defined as:

[0106]

[0107] (2) Calculate information entropy;

[0108] The information entropy is calculated using formula (4):

[0109]

[0110] (3) Calculate the information effect value

[0111] The information effect value is calculated using formula (5):

[0112] g i =1-e i (5)

[0113] (4) Calculate the weights;

[0114] The weights are calculated using formula (6):

[0115]

[0116] In the formula, M ij Refers to the data standardization matrix; ei Information entropy; g i W represents the information effect value. i The feature labels are weights; n is the number of samples; k is a constant.

[0117] Step S5: Identify the degree of village hollowing out based on comprehensive population outflow characteristic values;

[0118] For example, the comprehensive characteristic value of population outflow is visualized and graded in ArcGIS using the quartile method. The comprehensive value of population outflow is mapped one-to-one with the hollowing-out situation, and divided into three categories: mild hollowing-out, moderate hollowing-out, and severe hollowing-out. This enables the identification and analysis of the spatiotemporal distribution and degree classification of hollowing-out phenomena in the target area.

[0119] In another example, such as Figure 3 As shown, a trained machine learning model is used based on population outflow characteristic values ​​(I... G Population outflow characteristics based on nighttime light data (I) D Population outflow characteristics based on electricity consumption data (I) E The model identifies the degree of village hollowing-out based on comprehensive population outflow characteristics, such as mild, moderate, and severe hollowing-out. The model training process is known to those skilled in the art. For example, sample data containing population outflow feature values, population outflow feature values ​​based on nighttime light data, comprehensive population outflow characteristics, and corresponding village hollowing-out degree labels are used to train a neural network-based machine learning model. The specific process will not be elaborated upon in this invention.

[0120] According to another embodiment of the present invention, a village feature recognition device based on multi-source data is also disclosed.

[0121] The first feature extraction module is used to generate the first population outflow feature of the target village based on census data;

[0122] The second feature extraction module is used to generate a second population outflow feature of the target village based on the remote sensing image of the target village's nighttime lights.

[0123] The third feature extraction module is used to generate the third population outflow feature of the target village based on electricity consumption data;

[0124] The feature fusion module is used to calculate the comprehensive characteristics of population outflow based on the first, second, and third population outflow characteristics using a multi-objective linear weighted function method.

[0125] The identification module is used to identify the degree of hollowing out of villages based on comprehensive characteristics of population outflow.

[0126] Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications can still be made to the specific embodiments of the present invention or equivalent substitutions can be made to some technical features without departing from the spirit of the technical solution of the present invention, and all such modifications and substitutions should be covered within the scope of the technical solution claimed in the present invention.

Claims

1. A method for village feature identification based on multi-source data, characterized in that, include: Generate the first population outflow characteristic value for the target village based on census data; Generating a second population outflow characteristic value for a target village based on nighttime light remote sensing images; including: correcting the nighttime light data: conditionally filtering the annual-scale remote sensing light data based on a conditional function model to divide the target area into at least a first region and a second region; wherein, the first region is the area within the target area where the annual-scale remote sensing light data has a light radiance greater than or equal to a preset threshold, and the second region is the area within the target area where the annual-scale remote sensing light data has a light radiance less than a preset threshold; reconstructing the data based on the conditionally filtered annual-scale remote sensing light data corresponding to the first region to obtain dark-eye enhanced annual-scale remote sensing light data; and Using GIS application software tools, rural settlements were extracted as the boundaries of population-concentrated areas. Using the boundaries as masks, the corrected nighttime light data was extracted. Based on the annual-scale remote sensing light data reconstructed from the extracted data, the population regression model based on the nighttime light data was constructed as a cubic polynomial function expression of population and nighttime light radiance. Based on the regression results of the rural resident population based on the nighttime light data, the second population outflow characteristic value based on the nighttime light data was calculated. Generating a third population outflow characteristic value for target villages based on electricity consumption data includes: obtaining the latitude and longitude information of each resident's electricity consumption data acquisition station using an electronic map API; generating point data based on the latitude and longitude of each station in a GIS application software tool; inputting electricity consumption data; matching the point data with village area data; and summarizing electricity consumption data by village; constructing a population regression model based on electricity consumption data as a cubic polynomial function expression of population and electricity consumption values; and calculating the third population outflow characteristic value based on electricity consumption data based on the rural resident population regression results; wherein, the statistical time period for electricity consumption data is set to one month before and one month after the Spring Festival. The comprehensive characteristic value of population outflow is calculated using a multi-objective linear weighted function method based on the first, second, and third population outflow characteristic values. The degree of village hollowing-out is identified based on comprehensive characteristics of population outflow. 2.The village feature recognition method according to claim 1, characterized in that, The calculation of the comprehensive characteristic value of population outflow using the multi-objective linear weighted function method includes: determining the weights of the first, second, and third population outflow characteristics using the entropy weight method.

3. A village feature identification device based on multi-source data, characterized in that, include: The first feature extraction module is used to generate the first population outflow feature value of the target village based on census data; The second feature extraction module is used to generate a second population outflow feature value for the target village based on the remote sensing image of the target village's nighttime lights. This includes: correcting the nighttime light data; filtering the annual-scale remote sensing light data based on a conditional function model to divide the target area into at least a first region and a second region; wherein the first region is the area within the target area where the annual-scale remote sensing light data has a light radiance greater than or equal to a preset threshold, and the second region is the area within the target area where the annual-scale remote sensing light data has a light radiance less than a preset threshold; reconstructing the data based on the conditionally filtered annual-scale remote sensing light data corresponding to the first region to obtain the dark-eye enhanced annual-scale remote sensing light data; and Using GIS application software tools, rural settlements were extracted as the boundaries of population-concentrated areas. Using the boundaries as masks, the corrected nighttime light data was extracted. Based on the annual-scale remote sensing light data reconstructed from the extracted data, the population regression model based on the nighttime light data was constructed as a cubic polynomial function expression of population and nighttime light radiance. Based on the regression results of the rural resident population based on the nighttime light data, the second population outflow characteristic value based on the nighttime light data was calculated. The third feature extraction module is used to generate the third population outflow feature value of the target village based on electricity consumption data. This includes: obtaining the latitude and longitude information of each resident's electricity consumption data acquisition station using an electronic map API; generating point data based on the latitude and longitude of each station in a GIS application software tool; inputting the electricity consumption data; matching the point data with village area data; and summarizing the electricity consumption data by village. The module also constructs a population regression model based on electricity consumption data as a cubic polynomial function expression of population and electricity consumption values. Finally, it calculates the third population outflow feature value based on the rural resident population regression results based on electricity consumption data. The statistical period for electricity consumption data is set to one month before and one month after the Spring Festival. The feature fusion module is used to calculate the comprehensive feature value of population outflow based on the first, second, and third population outflow feature values ​​using a multi-objective linear weighted function method. The identification module is used to identify the degree of hollowing out of villages based on comprehensive characteristics of population outflow.

4. The village feature recognition apparatus according to claim 3, characterized by The feature fusion module uses a multi-objective linear weighted function method to calculate the comprehensive feature value of population outflow, including: using the entropy weight method to determine the weights of the first population outflow feature, the second population outflow feature, and the third population outflow feature.

5. A method for village feature identification based on multi-source data, characterized in that, include: Generate the first population outflow characteristic value for the target village based on census data; Generating a second population outflow characteristic value for a target village based on nighttime light remote sensing images; including: correcting the nighttime light data: conditionally filtering the annual-scale remote sensing light data based on a conditional function model to divide the target area into at least a first region and a second region; wherein, the first region is the area within the target area where the annual-scale remote sensing light data has a light radiance greater than or equal to a preset threshold, and the second region is the area within the target area where the annual-scale remote sensing light data has a light radiance less than a preset threshold; reconstructing the data based on the conditionally filtered annual-scale remote sensing light data corresponding to the first region to obtain dark-eye enhanced annual-scale remote sensing light data; and Using GIS application software tools, rural settlements were extracted as boundaries of population-concentrated areas. These boundaries were then used as masks to extract corrected nighttime light data. Based on the extracted data and reconstructed annual-scale remote sensing light data, a population regression model based on nighttime light data was constructed as a cubic polynomial function expression of population and nighttime light radiance. Based on the regression results of the rural resident population based on nighttime light data, a second population outflow characteristic value was calculated. A third population outflow characteristic value was generated for the target village based on electricity consumption data. This included: obtaining the latitude and longitude information of each residential electricity consumption data acquisition station using an electronic map API, and then applying this information to the GIS application software. The tool generates point data based on the latitude and longitude of each station, inputs electricity consumption data, and matches the point data with village area data to summarize electricity consumption data by village. It constructs a population regression model based on electricity consumption data as a cubic polynomial function expression of population and electricity consumption values. Based on the rural resident population regression results based on electricity consumption data, it calculates a third population outflow characteristic value based on electricity consumption data. The statistical period for electricity consumption data is set to one month before and one month after the Spring Festival. A multi-objective linear weighted function method is used to calculate a comprehensive population outflow characteristic value based on the first, second, and third population outflow characteristic values. The degree of village depopulation is identified by using trained machine learning models based on population outflow characteristics, population outflow characteristics based on nighttime light data, population outflow characteristics based on electricity consumption data, and comprehensive population outflow characteristics.

Images