Mathematical representation and image generation of human faces by metamorphosis



This study aims at emotional information processing and extracts the functions which generate several specified human facial expressions from a portrait. No satisfactory answer has been given to the problem as to how humans obtain information from facial expressions to identify another individual or to analyze an emotion. The major issue in studying the identification of an individual or the recognition of a facial expression is how and by what method the facial features should be described.

This paper presents an approach to this problem by tracing the change of the facial expression on a planar portrait using the metamorphosis by a thin-plate spline and extracts the change quantitatively by the form of the facial function. Using the facial function, six kinds of facial images representing typical human emotions are generated from a portrait. The facial function will be used for the security check system. © 1997 Scripta Technica, Inc. Electron Comm Jpn Pt 3, 80(1): 36–46, 1997