多源空间数据融合的城市人居环境监测模型与应用研究

人居环境监测作为城市人居环境建设与管理实践提升的基本,是目前人居环境研究落地的重点。传统的城市人居环境监测在数据更新速度、精度等方面存在不足,难以满足精细化管理需求。提出利用遥感数据与互联网的兴趣点POI(Point of interest)数据结合,建立人居环境监测模型。模型主要有两个关键环节,一是构建自动化提取建筑物算法,该算法通过建立地物特征集,以POI点对应样本为种子,利用全局最优和区域生长算法,自动提取城市建筑物,再利用全局最优算法确定其他地类的阈值;二是人居环境指标计算,将建筑物、绿地、水体信息提取结果与POI数据结合,利用密度类与距离类空间分析算法,分别计算自然、社会经济类指标。基于上述模型,利用2018年4月的北京二号遥感影像和POI点数据在北京市回龙观社区进行实验验证,结果显示:信息提取结果中,总体精度超过95%,Kappa系数超过92%,提取效率提高2.3倍,表明信息提取精度高且可信,适合工程化应用。计算回龙观社区人居环境监测指标,分析结果认为,社区内自然类指标差异不大,但缺乏水体生态系统,生物多样性不够丰富,社区内的商业比较繁华,但是学校和医疗不充足,尤其是缺乏大型公立医院。综上,通过人居环境监测模型研究和应用分析,将遥感数据和互联网数据结合应用于人居环境质量监测有效提高了精度和速度,有利于业务化,服务政府管理。

Urban human settlements monitoring model and its application based on multi-source spatial data fusion

Monitoring of human settlement environment, a basic practice in urban human settlement environment construction and management, is the focus of human settlement environment research. The traditional urban human settlement environmental data have shortcomings in terms of renewal speed and accuracy. This paper proposes to use remote sensing images and point of interest (POI) from the internet to build a human settlement environment model. The model is composed of two key parts. The first part involves constructing an automatic building extraction algorithm, which are the global optimization and region growth algorithms, based on urban land coverage feature sets and taking corresponding POI point samples as seeds. The second part involves the calculation of the human settlement environment index using density and distance spatial analysis algorithms, which use the land cover extraction results with POI data as input data. Based on the above model, Beijing-2 remote sensing images and POI data in April 2018 were used to verify the results in the Huilongguan community of Beijing. The results show that the overall accuracy of the information extraction results is more than 95%, the Kappa coefficient is more than 92%, and the extraction efficiency is improved by 2.3-fold. On applying the model to monitor human settlements in Huilongguan community, it was found that there is little difference in natural indicators, but due to the overall lack of water ecosystem, biodiversity is not rich enough. Socioeconomic indicators mainly account for the impact of business, school, and medical care. On the whole, businesses in the community are prosperous, and schools and medical care are adequate, the latter is the case especially in large hospitals. Through the research and application analysis of the human settlements monitoring model, the application of remote sensing data and internet data fusion in human settlement environmental quality monitoring effectively improves the accuracy and speed and is conducive to business applications and government management.