摘要
为了实现对脸部细节的移植,首先根据脸部特征点的对应关系,分别提出了全局和局部的拉普拉斯人脸细节迁移模型,提取样本图像的细节特征,并将其应用到目标图像上;然后给出人脸皮肤颜色迁移模型,并提出了一个用于解决发型替换过程中边界融合的局部影响图方法,使得对整个人脸细节变换的模拟更为全面和真实.实验结果表明:局部拉普拉斯方法效果更为真实自然,同时也证明文中提出的方法是灵活和有效的.
In order to transfer the details between human faces, the face details are first extracted from example image based on the correspondence of the facial feature points, applied to the target face image by a global and a local Laplacian detail transfer model respectively. Then, the skin color transition is further carried out to adjust the target face appearance. Thirdly, a local influence map is presented to blending the boundary during the hair style replacement, which makes the whole simulation of face detail transfer more competitive and natural. Experimental results demonstrate that the local Laplacian detail transfer model can take advantage over the global one to produce more realistic results, and consequently, the proposed method proves to be flexible and powerful.
出处
《计算机辅助设计与图形学学报》
EI
CSCD
北大核心
2008年第1期50-54,共5页
Journal of Computer-Aided Design & Computer Graphics
基金
中国博士后科学基金(20060401040)
微软联合实验室研究项目
关键词
拉普拉斯细节迁移模型
肤色变换
局部影响图
人脸老化
表情映射
expression Laplacian detail transfer model
skin color transition
local influence map
face aging
mapping
