基于几何聚类指纹库的约束KNN室内定位模型

在室内空间提供高精度的实时定位服务非常重要。室内数字地图是室内实时定位的基础和信息载体。研究了面向实时定位的室内空间理论，包括室内空间的定义、室内空间及室内实时定位的特点。分析了地图可视化和基于地图约束的定位精度改进等室内实时定位应用对室内地图的要求，明确了需要表达的室内空间结构为建筑体和室内家具。设计了一种轻量室内地图数据结构，探索了基于该数据结构的室内2D和2.5D地图的制图方法。最后以工程实验楼室内空间为例开展了制图实践。

A Constrained KNN Indoor Positioning Model Based on a Geometric Clustering Fingerprinting Technique

People generally spend around 90% of their time in indoor space every day. And they hope to have a high-precision real-time location based system for indoor navigation. Indoor digital maps are the basis and information carrier of indoor real-time positioning. In this paper, we discuss the theory of real-time location, including the definition of indoor place, features of indoor space and indoor space real-time location characteristics. The demands for visualization and map-filtering-based location accuracy improvement are analysed. In order to meet the needs of real-time indoor location, the building structures such as rooms, walls, doors, and corridors as well as furniture and other obstacles must be expressed. We also detail a light data structure for indoor maps and explore mapping methods for 2D floor plans and 2.5D maps of indoor space. We map the indoor space of a Engineering Laboratory building as an experimental case.