Nonlinear image processing for beautifying facial images

Summary form only given. A nonlinear image processing system is proposed for beautifying face images. The system is composed of a nonlinear filter bank using an e-filter, which is a nonlinear digital filter previously proposed by the author for removing small-amplitude and high-frequency noise added to a large-amplitude signal. By combining e-filters, a filter bank is obtained which separates signals considering not only the frequency, but also the amplitude. By applying this nonlinear filter bank to a face image, signal components in the face image can be obtained separately. Thus, by removing small-amplitude, relatively high-frequency components, which correspond to wrinkles and spots, the skin looks smooth and beautified. Moreover, by combining the filter bank with signal enhancement, a face image becomes sharp while keeping the skin smooth. Computer simulations verify its high performance.