在无约束条件下,人脸表情、姿态、光照以及背景等复杂因素可能导致人脸图像的类内变化大于类间变化.针对如何降低较大的类内变化对人脸验证研究的影响,本文结合加权子空间,提出了一种带先验相似性和先验距离约束的相似度度量学习方法.首...在无约束条件下,人脸表情、姿态、光照以及背景等复杂因素可能导致人脸图像的类内变化大于类间变化.针对如何降低较大的类内变化对人脸验证研究的影响,本文结合加权子空间,提出了一种带先验相似性和先验距离约束的相似度度量学习方法.首先,利用类内人脸对样本,学习带权重的类内协方差矩阵,通过加权子空间的投影,从人脸图像中获得鲁棒性的人脸特征表达;其次,利用样本对的相似性与差异性,建立了带先验相似性和先验距离约束的相似度度量学习模型,优化后的度量矩阵可以有效提高特征向量的类内鲁棒性和类间判别性;最后,利用优化的度量矩阵计算人脸对的相似度.在LFW(Labeled Faces in the Wild)数据集的实验验证了所提模型的有效性,与其它同类相似度度量学习方法相比,优化的度量矩阵更能准确地评估人脸间的相似性,并在受限训练集上取得了91.2%的识别率.展开更多
According to the authoritative data involving social economic indicators and greenhouse gas （GHG） emission from the international universal database, the levels and processes of economic development and GHG emission...According to the authoritative data involving social economic indicators and greenhouse gas （GHG） emission from the international universal database, the levels and processes of economic development and GHG emission in major economic groups, nations and regions of the world are simultaneously analyzed. Obtaining Gross Domestic Product （GDP） and emission per capita from various countries and regions in the past 40 years as the standard, countries and regions in the world are divided into six groups： countries with low emission per capita and low economic level （IA）, countries with low emission per capita and medium economic level （IIA）, countries with low emission per capita and high economic level （IIIA）, countries with high emission per capita and medium economic level （liB）, countries with high emission per capita and high economic level （IIIB）, countries with high emission per capita and low economic level （IB）. Countries belong to IB are quite rare in the study period, while the first five groups correspond to the poor regions, main developing countries, economically transitional countries with rapid economic development, rich islands and developed North America and Europe respectively. Data analysis shows that there is a close relationship between emission and economic development of different countries and regions. The composition relationship between economic development of different countries and regions is relatively stable over a long period of time. From 1970 to 2005, rising trends existed in the economic development of most countries and regions. However, the emission had a significant increase in a small part of countries and regions. In other words, for those with high emission, the emission level is always high. But for those with low GHG emission, the emission does not increase too much. The main processes of the change of countries pattern from IA to IIA and from II B to IIIB, occurring in the 1970s and from the late 1970s to the 1980s respectively. That result has .a significant enlightening effect in understanding the relationship between emission and eco- nomic development and its historical process of various countries and in choosing the position of our country in the future climate diplomatic negotiations.展开更多
文摘在无约束条件下,人脸表情、姿态、光照以及背景等复杂因素可能导致人脸图像的类内变化大于类间变化.针对如何降低较大的类内变化对人脸验证研究的影响,本文结合加权子空间,提出了一种带先验相似性和先验距离约束的相似度度量学习方法.首先,利用类内人脸对样本,学习带权重的类内协方差矩阵,通过加权子空间的投影,从人脸图像中获得鲁棒性的人脸特征表达;其次,利用样本对的相似性与差异性,建立了带先验相似性和先验距离约束的相似度度量学习模型,优化后的度量矩阵可以有效提高特征向量的类内鲁棒性和类间判别性;最后,利用优化的度量矩阵计算人脸对的相似度.在LFW(Labeled Faces in the Wild)数据集的实验验证了所提模型的有效性,与其它同类相似度度量学习方法相比,优化的度量矩阵更能准确地评估人脸间的相似性,并在受限训练集上取得了91.2%的识别率.
文摘According to the authoritative data involving social economic indicators and greenhouse gas （GHG） emission from the international universal database, the levels and processes of economic development and GHG emission in major economic groups, nations and regions of the world are simultaneously analyzed. Obtaining Gross Domestic Product （GDP） and emission per capita from various countries and regions in the past 40 years as the standard, countries and regions in the world are divided into six groups： countries with low emission per capita and low economic level （IA）, countries with low emission per capita and medium economic level （IIA）, countries with low emission per capita and high economic level （IIIA）, countries with high emission per capita and medium economic level （liB）, countries with high emission per capita and high economic level （IIIB）, countries with high emission per capita and low economic level （IB）. Countries belong to IB are quite rare in the study period, while the first five groups correspond to the poor regions, main developing countries, economically transitional countries with rapid economic development, rich islands and developed North America and Europe respectively. Data analysis shows that there is a close relationship between emission and economic development of different countries and regions. The composition relationship between economic development of different countries and regions is relatively stable over a long period of time. From 1970 to 2005, rising trends existed in the economic development of most countries and regions. However, the emission had a significant increase in a small part of countries and regions. In other words, for those with high emission, the emission level is always high. But for those with low GHG emission, the emission does not increase too much. The main processes of the change of countries pattern from IA to IIA and from II B to IIIB, occurring in the 1970s and from the late 1970s to the 1980s respectively. That result has .a significant enlightening effect in understanding the relationship between emission and eco- nomic development and its historical process of various countries and in choosing the position of our country in the future climate diplomatic negotiations.
