Non-Linear Chaotic Features-Based Human Activity Recognition

Human activity recognition (HAR) has vital applications in human–computer interaction, somatosensory games, and motion monitoring, etc. On the basis of human accelerate sensor data, through a nonlinear analysis time series, novel method for HAR that is based on non-linear chaotic features proposed this paper. First, C-C G-P algorithm are used to, respectively, compute optimal delay embedding dimension. Additionally, Reconstructed Phase Space (RPS) formed while using time-delay accelerometer data. Subsequently, two-dimensional feature matrix constructed, where composed correlation dimension largest Lyapunov exponent (LLE) attractor trajectory RPS. Next, classification algorithms order to classify recognize two different classes, i.e., basic transitional activities. The experimental results show higher accuracy than traditional frequency domain features.