Abstract

The extraction of human facial features from images in binary form has been shown to be useful in at least two areas of low data-rate image coding. One is in the transmission of moving cartoons over the public switched telephone network [1]-[3]. Another is in model-based coding, where analysis of the camera signal is needed to line up the software model with the exterior world [4]. Binarization of the input image is also used as a first stage in machine recognition of faces [5].

© 1989 Optical Society of America