摘要
Haar小波具有算法简单复杂度底的特点,Gabor小波具有能够对目标的局部区域灰度变化进行多方向的有效描述,和进行多尺度的描述的能力。结合两种小波变换在特征提取上的各自优势,提出一种改进的基于复合特征提取方法的Adaboost分类器,在人脸检测实验中取得较好的效果。
Haar wavelet vectors are low-dimension and low computational complexity,and Gabor wavelet has ability to describe the local intensity variation at different orientations of different scales.A modified Adaboost classifier is proposed to detect the face based on the extraction method of compound features.Experiments results show that it has efficiency in facial detection by combining Haarwavelet and Gaborwavelet's respective advantages in feature extraction..
出处
《三明学院学报》
2010年第4期326-330,共5页
Journal of Sanming University
基金
华侨大学科研基金项目(09HZR11)