随着我国社会经济的快速发展，在乡村振兴的战略背景下，农村居民地范围和人口变化信息是重要的基础信息资源。与传统的普查等调查方式相比，遥感调查方式是获取农村居民地和人口相关信息更加经济、快捷的技术手段。农村居民地存在空间分布不均衡、区域内分散分布、居民地规模小的特点，传统的中高分辨率影像提取农村居民地范围存在效率低下的问题。随着城市化的发展，我国农村人口在逐步减少，造成大量农村人口人户分离，农村户籍人口已不能客观反映农村常住人口的数量。
论文提出了协同夜间灯光数据和中等空间分辨率遥感数据，用夜间灯光遥感数据提取包含农村居民地的靶区、中等分辨率提取靶区内农村居民地的方法，并分别以伊春、中卫、龙岩市为实验区进行了靶区与农村居民地的精度评价。论文通过构建时间序列一致的夜间灯光时间序列数据，利用时间序列变化分析方法探测县级行政单元尺度上农村人口开始减少的时间和减少速率，分析了我国农村人口数量开始减少时间的时空特征，并利用地理探测器分析了农村人口减少的主要影响因素。本研究的主要成果和创新点如下：
（1）提出了多源遥感数据协同的农村居民地高效提取方法。针对现有农村居民地遥感提取空间分辨率低和效率低下的问题，利用粗分辨率的NPP/VIIRS夜间灯光数据时间序列与Sentinel-2 MSI (MultiSpectral Instrument)数据协同，利用NPP/VIIRS夜间灯光数据提取农村居民地靶区，利用Sentinel-2 MSI数据提取多种特征参数，利用面向对象的方法识别农村居民地，在三个不同的实验区提取农村居民地精度均在90%以上。
（2）发展了基于夜间灯光遥感数据的县级农村人口变化分析方法。农村夜间灯光相对亮度较低，农村人口分散，建立以县级尺度的为最小研究单元的分析是有意义的，本研究以DMSP/OLS、NPP/VIIRS两种夜间灯光遥感数据建立1992年以来的灯光数据时间序列，确立灯光数据和农村人口的关系，并在时间序列上进行统计学上的验证，在县级尺度上得到可以代表农村人口的农村夜间灯光时间序列。
（3）分析了1992年以来中国农村人口减少的时空特征及其影响因素。利用空间分析和带有空间约束的K-means算法分析中国农村人口流动的时空格局，结合地形数据和社会经济数据等辅助数据，利用地理探测器分析中国不同地区发生农村人口减少的主要影响因素及其影响因素之间的交互探测作用，发现人口、经济、地形要素均对农村人口变化产生影响，当探测因子发生交互作用时，能够解释80%以上的农村人口变化。

With the rapid development of society and economy in China, rural settlements and population change is an important basic information resource under the strategic background of rural revitalization. Compared with the traditional survey methods such as census, remote sensing survey is a more economical and fast technical means to obtain the information related to rural settlements and population. Due to the characteristics of unbalanced spatial distribution, scattered distribution within the region and small scale of rural settlements, the traditional medium and high spatial resolution image extraction of rural settlements has the problem of low efficiency. With the development of urbanization, the rural population in China is gradually decreasing, resulting in the separation of a large number of rural population. Rural registered population can not objectively reflect the number of rural resident population.
This study presents a collaborative nighttime light and moderate spatial resolution remote sensing data, nighttime light remote sensing data was used to extract contains rural residents to the target area, moderate spatial resolution data to extract the target of rural residents in the area. In this paper, Yichun, Zhongwei and Longyan were taken as experimental areas to evaluate the accuracy of target area and rural residential area. This study consistent time series is built with nighttime light series data, using the time sequence analysis method for change detection at the county level administrative unit scale time and decline rate of the rural population began to decline, has analyzed our country rural population began to decrease time characteristics of space and time, and making use of the geographical probe analysis the main factors influencing rural population decline. The main achievements and innovations of this study are as follows:
(1) Develop an efficient extraction method of rural residential land based on multi-source remote sensing datserts. Aiming at the deficiencies of the existing rural residential land survey methods, this paper designs a new rural settlements extraction algorithm. In this algorithm, the time series of NPP/VIIRS night light data and Sentinel-2 MSI (MultiSpectral Instrument) multispectral data are coordinated. The rural target areas are extracted from NPP/VIIRS night light data, and a variety of characteristic parameters are extracted from Sentinel-2 MSI data. The rural settlements are identified by object-oriented method. Finally, this study conducted statistical verification on the time series and obtained the rural night light time series that could represent the rural population at the county level.
(2) An analysis method of rural population change at county level based on night light remote sensing data was designed. Since the relative brightness of rural light is low and the rural population is scattered, it is meaningful to establish the analysis with the county scale as the minimum research unit. In this part, DMSP/OLS and NPP/VIIRS night light remote sensing data were used to establish the time series of light data since 1992, and the relationship between light data and rural population was established. This study conducted statistical verification on the time series and obtained the rural night light time series that could represent the rural population at the county level.
(3) Analyze the spatial pattern and influencing factors of rural population mobility in China since 1992. Spatial analysis and K-means algorithm with spatial constraints were used to analyze the spatial and temporal pattern of rural population mobility in China. Combined with topographic data and socio-economic data, this paper uses GeoDetector to analyze the main influencing factors of rural population migration in different regions of China and the interaction between the influencing factors. It is found that population, economy and topography all influence the change of rural population. When the detection factors interact with each other, they can explain more than 80% of the change of rural population.