抄録
This paper presents a method for removal of eyeglasses frame in facial images. First, we make a parametric model of general shapes of the eyeglasses frames. Second, we estimate the inertia principal axis of each parts, and extract a part of the facial image which matches the model. Finally, by replacing the gray level of extracted region using the gray level of the surrounding skin region, we get an image of a face without eyeglasses. The results can avoid effect of eyeglasses for region segmentation of facial images.