摘要
人脸检测是人脸识别的一项重要任务。论文提出了一种基于Gabor滤波特征和一类分类器的正面人脸检测方法。算法首先利用了Gabor滤波器的良好的空间位置与方向的选择特性,采用了四种方向的Gabor滤波器提取人脸样本图像特征并用PCA方法对特征降维,然后用已降维的特征训练支持向量机分类器。最后应用一类分类器分类检测人脸。实验结果证明该方法是十分有效。
Face detection is an important task in face recognition. A method of frontal face detection basedon feature extract using Gabor filter features and One-class Classification is proposed in this paper. Taking advantages of the desirable characteristics of spatial locality and orientation selectivity of Gabor filter, the paper designs four filters corresponding to four orientation for extracting features from face images. After extracting the features, a reduced feature subspace is learned by PCA. The feature vectors based on Gabor filter is used as the input of One-class Classification to be trained and classified. A test image is detected by the trained One-Classification. The experiment results show the method can effectively detect face.
