Human face profile recognition by computer

摘要

A human face profile recognition system with learning capability is developed. One key point of the imaging system is the use of back-lighting so that the outline curve can be extracted easily. Cubic B-splines are used to extract the interesting points (turning points) and a total of six interesting points are extracted. From the five curve segments determined by these interesting points 24 features are extracted and used for the recognition. Eighteen people, typically representing most faces here, were tried for the learning and the recognition test. After complete learning the recognition tests were all successful.