室内空间用户行为的智能手机识别

针对地磁匹配中经常会出现相似点,造成定位偏差较大的问题,该文提出利用智能手机识别用户室内行为方式的方法,为地磁匹配算法提供筛选条件。开发了智能手机传感器数据采集工具,获取用户在室内环境下的行为数据。原始数据首先利用一阶低通滤波和平滑滤波算法进行去噪处理,再经过数据分割和特征提取后,应用于行为识别过程。行为识别模型的建立主要使用两种方法,K最近邻算法和隐式马尔可夫模型,并研究了两种方法的不足以及改进途径。通过针对识别准确度的对比实验,在输入最合适的数据的条件下,隐式马尔可夫模型的准确度略优于K最近邻算法。两种方法的识别准确率均在95%以上,能够有效地提高地磁定位精度。利用室内用户行为数据辅助地磁室内定位,很好地改善了地磁数据单一、定位精度较低的问题。

User behavior recognition of indoor space based on smart phone

Aiming at the problem that the geomagnetism similarities often occurs in the geomagnetism matching,the paper put forward the method of identifying the indoor behavior mode of the user by the smart phone which provided the screening condition for the geomagnetic matching algorithm.A smart phone sensor data acquisition tool was developed to obtain the user's behavior data in the indoor environment.The original data was processed by first order low pass filter and smoothing filter,and then it was used in the process of behavior recognition after data segmentation and feature extraction.In this paper,the establishment of identification model mainly used two methods,K nearest neighbor algorithm and hidden Markov model,and shortcomings of the two methods and the ways were studied to improve it.Finally,a comparative experiment was designed to verify the accuracy of the two algorithms,and the accuracy of the hidden Markov model was slightly better than the K nearest neighbor algorithm when the input data was the most suitable.The recognition accuracy of the two methods is more than 95 ％,which could effectively improve the accuracy of geomagnetic location.The problem that geomagnetic data is single and positioning accuracy is relatively low was improved by using indoor user behavior data to assist the geomagnetic indoor location.