Abstract: In this paper, we presented a new method for fatigue recognition based on facial motion and improved locality preserving projections. Facial velocity information,which is determined using Optical flow techniques,was used to characterize fatigue. The improved locality preserving projections,which preserve the local and global structure of the data manifold,was proposed to extract the effective fatigue feature. Weighted K-nearest neighbor (WKNN) was instructed to recognize fatigue. A set of experiments of fatigue recognition were presented, and the experiment results show that the proposed method is effective and attains a satisfactory effect.

Key words: Fatigue recognition,Improved locality preserving projections,WKNN