摘要
对于任意给定的人脸图像,提出了一个推演任意年龄段人脸图像的个性化模拟方法.利用人脸左右对称的特点设计了一个光照补偿算法;通过几何预处理和灰度预处理,建立归一化的人脸数据库和人脸轮廓库;给定的输入图像经过预处理、特征提取、轮廓变换和纹理合成4个步骤之后得到模拟图像.该方法实现了对任一年龄段、不同生活状况不同胖瘦情况下目标人脸的定制.实验结果表明,该方法对于正面人像有较好的模拟效果.
A personalized facial simulation method was put forward for deducing the face image of any given age group. An illumination compensation algorithm was designed based on the analysis of symmetrical facial features. A normalized face database and a face contour database were established after geometric preprocessing and grayscale pretreatment. The given input image was converted to the destination image after four steps： preprocessing, feature extraction, contour deformation and texture synthesis. This method realized the customization of the destination image of any given age group. An experiment showed that the above-mentioned method had a satisfactory simulation effect in dealing with positive face images.
出处
《西南大学学报（自然科学版）》
CAS
CSCD
北大核心
2014年第2期178-185,共8页
Journal of Southwest University(Natural Science Edition)
基金
国家社科基金重点项目(11AZD057
12AZD073)
关键词
光照补偿
个性化人脸图像
图形变换
纹理合成
illumination compensation
personalized facial image
morph
texture synthesis
