An Algorithm for Automatic Side Face Portrait Recognition Based on Fourier Descriptor

The aim of this study was to establish a method of automatic side face portrait recognition based on Fourier descriptor. The method consisted of three steps: face outline extraction, Fourier descriptor feature extraction and feature comparison. In the step of face outline extraction, the background was separated from the face first so as to get the overall face contour line, and then, determined two feature points of forehead and lower jaw, next input the contour line between the two points to extract the Fourier descriptor. This algorithm selected the Fourier descriptor which was invariable to the variability of rotation, zooming, and the invariable displacement to eliminate the influence caused by shooting anglepsilas deviation in the process of portrait photograph. The characteristic contrast adopted a simple method of distinction based on the statistical probability of characteristic vector space. At last the article discussed the test results of the algorithm.