The purpose of this thesis is to implement an automatic person identification system based on face profiles. Each person's face profile can be quite unique within a small sample population and therefore it can be used as the basis of an automatic person identification system. To quantify human face profiles for use in the recognition system, Fourier descriptors are used to describe the open curve extracted from a face profile. Fourier descriptors in the low-frequency range are shown to be useful for human face profile recognition. By using 16 Fourier coefficients, a correct recognition rate of 92% for 60 subjects was achieved.
Library of Congress Subject Headings
Optical pattern recognition; Face perception--Data processing; Identification--Data processing
Department, Program, or Center
Electrical Engineering (KGCOE)
Wong, Vincent, "Human face profile recognition" (1994). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus