Face Detection Approach Based on Fuzzy Logic and GST

This paper proposes a fast and confident approach to face detection. Initially we detect and segment the regions of image with human skin color. The regions with similar features are merged in order to generate a bigger region. Following the extraction of each region's features, the resulted data is given to the fuzzy system trying to determine regions which are most likely to belong to skin. After performing preprocessing, the regions with highest possibility of containing face are examined and similar regions are merged to generate a large region. After segmenting the face regions of image, the regions are divided into two categories of face and non-face by applying the Generalized Symmetry Transform (GST) algorithm. The results prove that new algorithm not only runs relatively fast but also provides an improved performance.