Probabilistic LMA-based human motion analysis by conjugating frequency and spatial based features

This paper presents an approach to analyse human motions using Laban Movement Analysis (LMA) system. LMA is a known descriptor for analysing human body motion by using different components that it has. We attempt to present how the frequency and spatial-based features can be applied in LMA concepts, and then we present that how the conjugating of these two kinds of different domains features can be fused by a Bayesian Network (BN). The point is that, the human motions based on the type of body parts motion, can be recognized by some aspects or features better than the others. By combining those different aspects (LMA components) or features, more precise estimation for more various human motions can be obtained. The results approve that our idea to recognize human motion is very successful.