摘要
Gabor核函数的幅值反映了图像局部的能量,且在真实边缘附近具有良好的光滑性,适宜于匹配识别;AdaBoost算法用于Gabor特征集中选择最优特征。每个特征对应一个弱非类器,集合所有弱分类器组成一个最终分类器。构建了基于上述特征的人脸定位评估函数。实验表明,其对主动外观模型和主动形状模型的人脸定位有很好的评估功能。
The magnitude of Gabor＇s kernel function reflects the partial energy of image＇s, and has better velvet character around reality brink, so it is fit for matching and recognition. AdaBoost algorithm selects optimized characters from Gabor＇s character database, each character corresponds to an infirm classified implement, and all infirm classified implements form the final classified implement. In this paper it designs an evaluation function of face detection based on characters above. The experiment shows that this evaluation function has better evaluating function in active appearance model and active shape model of face detection.
出处
《计算机应用与软件》
CSCD
北大核心
2008年第10期214-216,共3页
Computer Applications and Software
