摘要
核回归法是一种非参数估计方法,其显著优点是:回归模型完全由数据驱动,无需信号具体形式。近些年来,有研究者将该方法应用在图像降噪和图像修补等方面,得到较好效果。在理解该方法原理的基础上,通过实验说明核回归方法在人脸检测技术方面也值得应用,而且效果较好。
The Kernel regression is a nonparametric estimation methods,it is an important advantage that the method relies on the data itself to dictate the structure of the model without knowing the form of signal.In recent years,some researchers taked the method into image denoising and image retrieval,and they obtained the excellent effect.But the method is applied to face detection in a few studies.This paper introduces that the method is extended to face detection,and shows the effect of experiment.
出处
《微处理机》
2012年第2期58-61,共4页
Microprocessors
关键词
核回归
人脸检测
权特征
Kernel regression
Face detection
Weight feature