Abstract
An image processing method is proposed for ensure person's safety and health care. In the house, because illumination conditions change as time and weather. So, an image of the person is detected by the optical flow procedure. Next, six factors are introduced in order to distinguish and represent postures and positions of the person. Those six factors of various motions are analyzed by the principal component analysis. Results are plotted in a three dimensional eigenspace in the scatter chart. Next, ranges of principal component values of each motion are approximated by cylindrical spaces and used them as discriminant standards to recognize motions. Then, some activities of daily living are distinguished from the change of time series of principal component value in the eigenspace.
