Automatic face and facial features detection

Abstract

Many researches have been presented to search face region for human recognition and coding schemes. In this paper, the authors propose an automatic face and facial features detection algorithm. There are two main steps: the face region detection from an original complex image; and the facial feature extraction from the detected face region. A genetic algorithm (GA) is used to search for the face region in the background for the first step. For the facial feature, the authors propose two methods; one of them is a geometric method using people's face information and the other is a method using Gaussian derivative filters. They obtain good experimental results for face detection, irrespective of the image size, wearing glasses and moustaches. For the two methods of detecting facial feature, the method using geometric property is faster than a Gaussian derivative filter, but the former is less exact than the latter.