학술논문
A new method for indoor non-rhythmic human motions classification using ultra-wideband radar
Document Type
Conference
Author
Source
2017 International Applied Computational Electromagnetics Society Symposium (ACES) Applied Computational Electromagnetics Society Symposium (ACES), 2017 International. :1-2 Aug, 2017
Subject
Language
Abstract
In order to classify non-rhythmic human motions by ultra-wideband (UWB) radar, a weighted rang-time-frequency transform (WRTFT) method is proposed to extract both Doppler and range features. And the bootstrap-aggregated decision trees (Bagged Trees) classifier is utilized to recognize each non-rhythmic human motion. Experiments show that the proposed method can achieve 92.9% classification accuracy for recognizing seven typical non-rhythmic activities.