基于高分遥感影像和POI的国土空间规划现状细化调查

当前国土空间规划现状细化调查尚处于外业核查与内业人工目视判读的方法探索阶段，缺乏较高效率、较高精准度的技术方法。因此，文章提出一种融合高分遥感影像图像信息和POI类别属性信息的数据特点，利用核密度分析和面向对象的信息提取方法，实现对国土空间规划现状地类的细化调查。通过对城镇中心和城镇边缘2个研究区进行建筑物提取和POI属性挂接，识别出各建筑物的使用功能类别，再与2019年第三次全国国土调查的国土利用现状图层进行空间叠加，完成对研究区的地类细化调查。结果表明，经样本误差矩阵评价建筑物提取结果具有较高分类精度，研究区的地类细化调查结果准确度达到86.2%，且大大提高了工作效率，为国土空间规划现状调查逐步走向精准化、高效化和智能化道路提供了方法和思路借鉴，具有重要的深入研究和应用推广意义。

Investigating Land Use Types for the Spatial Planning of National Land Based on High-Resolution Remote Sensing Imagery and Point of Interest

A detailed investigation of land use status is the basis of spatial planning conducted on National Land and the core of the producing basic map . Research towards creating a detailed investigation method for determining land use status can provide technical support for the converting land use statuses, which is favorable for promoting the scientific, rational, and operational results of conducting spatial planning of National Land. Presently, detailed investigations of land use are still in the stage of manual visual interpretation and data exploration, and there is practically no attempt to apply the method with a higher efficiency and accuracy. Therefore, this study proposes an image information extraction technique based on a combination of highly scoring remote sensing images and class attribute information of point of interest , kernel density analysis and object-oriented methods (for information extraction), and the current detailed method used to investigate land use types. eCognition 9.0 and ArcGIS 10.2 software are used to extract buildings and link POI attributes in two study areas, the town center and town edge, to identify the functional classes of each building. Information on the attributes of the POI is extended from point to area before the land types, based on the results of the Third National Land Survey, are used for a spatial overlay analysis that completes the classification of land in the study area. The results show that the overall accuracy of the training samples extracted by eCognition 9.0, when combined with the POI information-assisted land classification survey results, reaches 100% after the confusion matrix is tested for accuracy; when compared with the visual interpretation method, the method is useful in the study area. The accuracy of the segmented survey results reached 86.2%. In addition, the buildings in the town center are mainly high-rises, and their functions are mainly residential, scientific, educational, cultural, and related to health and commercial services whereas the function of buildings on the town edge are mainly residential, meant for logistics or storage, or are related to industrial and transportation services. In the future research, the central area of the town can be studied using a variety of methods to improve the accuracy with which high-rises are extracted; for the town edge, it is necessary to improve the discrimination and precision of the extraction of exposed land and concrete buildings while simultaneously increasing the extraction of roads and other shape objects to aid the determination of the function of buildings adjacent to them. The application of this method can greatly improve the efficiency of work and provide a new method and way of thinking about the precision, efficiency, and intelligence of basic spatial planning work conducted on National Land. Moreover, it will be of great significance for other applications and further research.