A Human Action Recognition Method Based on Tchebichef Moment Invariants and Temporal Templates

In this paper, a new human action recognition method based on Tchebichef moment invariants and temporal templates is presented. We use the motion energy image (MEI) and motion history image (MHI) as the feature representation of the human action at first. Then the Tchebichef moment invariants extract the feature vectors of MEI and MHI. Tchebichef moment invariants perform better than Hu moment invariants and Zernike moment invariants. Finally cluster the actions and use the nearest neighbor algorithm to recognize each human action. The result of these experiments suggests that this method has a high recognition rate in in both noise-free and noisy condition. Therefore, the algorithm has a good robustness.