A Study on Face Similarity Using Knowledge Processing Methods

The human face plays an important role in social interaction, such as communications, identifications, and security. Thus, facial information processing becomes an active subfield of the pattern recognition area. Facial recognition/identification algorithms usually calculate the similarity between a face and faces stored in databases and determine if they match or not. In those algorithms, the measurement of similarity is important. This paper presents knowledge about face similarity acquired by knowledge information processing methods and discusses the results. In the experiments, three kinds of questionnaires are used to obtain training examples, and these examples are processed by decision tree system C4.5 and inductive logic programming system ProgoL.