【作者】 罗仁泽； 王汝言； 冉瑞生；

【Author】 Luo Renze~1 Wang Ruyan~2 Ran Ruisheng~3 1 Zhongshan Institute,University of Electronic Science and Technology of China(UESTC),Zhongshan,528402; 2 The Division of Communication and Information Engineering,Chongqing University of Posts and Telecommunication,Chnngqing,400065; 3 Computer Science and EngineerinInstitute,UESTC,Chengdu,610054,P.R.China

【机构】 电子科技大学中山学院； 重庆邮电大学通信与信息工程学院； 电子科技大学计算机科学与工程学院；

【摘要】 2DPCA 是最近提出的一种关于人脸图像特征提取和识别的前沿技术,该方法是在使投影特征样本的总体散度最大的准则下给出的主元选取策略。但到目前为止,并未从理论上说明该主元选取策略使得损失的能量最小。本文在最小均方误差准则下,给出了2DPCA 方法的主元选取策略,同时也从理论上证明了2DPCA 的主元选取策略使得图像损失的能量最小,说明了该方法的合理性。更多还原

【Abstract】 Recently,as a novel technique,2DPCA has been proposed for face image feature extraction and recognition.It is based on the maximum of the total scatter of all the projected feature samples.But so far,it is not proved theoretically that the energy lost is minimal.Based on the rule of minimal mean square error,2DPCA is presented.Meanwhile,in theory,it is proved that face feature extraction of 2DPCA ensures the energy lost is minimal and this method is reasonable.更多还原