Human action recognition based on non-negative matrix factorization

In this paper, we propose a method to recognize human behavior by combining motion history images (MHI) and non-negative matrix factorization (NMF). The MHI preserves the temporal information of a behavior by holding the temporal motion appearance. Then, NMF is applied to extract the middle-level features of the moving object. The experimental results show that the proposed scheme can achieve robust recognition results using a public dataset.