Activity Recognition Method Based On Weighted Lda Data Fusion

Human Activity Recognition (HAR) has a positive impact on people's well-being and it can help decrease the occurrence of chronic diseases in the senior population. The main purpose of this paper is to present a novel activity recognition method based on missing data processing and multi-sensor data fusion that can be applied to identify Activities of Daily Living (ADLs). Hereinto, missing data processing based on the temporal correlation is presented first to estimate the missing data, which utilizes the neighboring non-missing values to construct a linear spline model. Then, considering that sensors on different body positions may play as experts on different activity classes, a multi-sensor fusion method based on weighted Linear Discriminant Analysis (LDA) to learn activity-specific sensor weights is presented. Successively, an activity recognition method based on missing data processing and weighted LDA data fusion is proposed, which can further enhance data quality and the recognition accuracy. Experimental results show that the proposed method is more effective and robust, and its performance is competitive against other state-of-the-art methods.