Feature extraction and dimensions reduction using R transform and Principal Component Analysis for abnormal human activity recognition

In this paper the recognition of abnormal human activities: forward fall, backward fall, chest pain, fainting, vomiting, and headache is studied. The proposed system model presents a novel combination of R transform and Principal Component Analysis (PCA) for abnormal activity recognition. The idea is to take advantage of both local and global feature extractions by R transform and PCA methods respectively. R transform reduces 2-D sequence of activities to a set of 1-D signal by focusing on local shape features. PCA applied on the 1-D signal further reduce the dimensions and provide global feature representation. Hidden Markov Model (HMM) is applied on extracted features for training and activity recognition. By testing our system on six different abnormal activities, we have obtained an average recognition rate of 86.5%. The experimental results show that our proposed approach provides improved recognition rate of 6% to 10.5% on average as compared to PCA, Linear Discriminant Analysis (LDA), and PCA, LDA combination.