Human behavior recognition based on sitting postures
Songlin Wu, Rongyi Cui · 2010
In this paper, on the basis of detecting human skin area, 8 typical sitting postures were recognized using PCA. Firstly, moving object was detected by background contrast attenuation method. Then, considering the clustered skin area in a fixed region of YCbCr space which has an ellipse-like projection in CbCr plane, the skin area of moving object was extracted. Finally, the behavior recognition was implemented using PCA on the grayscale image of skin, and the face motion was analyzed according to the time-variation of pixel number in facial skin area. Experimental results show that the average recognition rate is 84.92%, and the face motion is analyzed effectively. Meanwhile the proposed algorithm is reasonably robust in shadow and varying luminance environment.