Mobile phone built-in MEMS sensor has the disadvantages of instable zero-bias and fast error divergence. In order to improve the accuracy of pedestrian step detection,three error processing model:Zero-bias correction,low-pass filtering and Kalman filtering are established based on graphical user interface(GUI). Double-features detection method using peak domain and time domain to constrain peak value is adopted. Then,two dynamic experimental scenes of steady and swing-arm progresses are designd to detect pedestrian's steps. The experimental results show that detection accuracy of pedestrian 's steps detected by double-features detection method is above 95％ in different dynamic scenes and the accuracy and robustness of the proposed algorithm are verified,satisfy the requirements of pedestrian's step detection.
关键词
微机电系统/图形用户界面/零偏校正/低通滤波/卡尔曼滤波
Key words
MEMS/graphical user interface(GUI)/zero-bias correction/low-pass filtering/Kalman filtering
维普
万方数据