A Meta-heuristic Model for Human Micro Movements Recognition Based on Inertial Sensors

Current developments in inertial sensors based on Microelectromechanical Systems technology allows us to design wearable devices for human movements automatic detection. Detection of human movements in natural environments needs detailed inertial signals processing. These activities are detected using human short movements detection. In this paper a method is proposed for continuous human tiny movements based on inertial signal processing. In the proposed method, at first linear acceleration and earth gravity signals are calculated using inertial navigation algorithms. Then discriminant features are extracted using a combination of these to signals. Innovation of this paper is introducing a new classification algorithm for continuous tiny movements recognition named Conditional Models. Each model belongs to a class of micro movement which performs some operations for generating a logical expression set and classifying the samples in that class. The proposed method uses the Particle Swarm Optimization to finding the optimized logical expression for each model. In order to evaluating, this method is tested on prayer micro movements recognition. Running the algorithm on the population of prayers and comparing with well-known micro movements classification models demonstrates the accuracy and high recognition of the proposed method.