基于Candide-3算法的图像中面部替换技术
    点此下载全文
引用本文:黄 诚.基于Candide-3算法的图像中面部替换技术[J].计算技术与自动化,2018,(2):100-104
摘要点击次数: 1572
全文下载次数: 0
作者单位
黄 诚 (上海工艺美术职业学院上海 201808) 
中文摘要:提出了一种基于CANDIDE-3算法的人脸替换算法,该算法使用CLM作为面部特征点定位算法,将2维的面部特征点和CANDIDE中3维的顶点相对应,建立相应的源人脸算法。根据面部特征点估计出头部姿态和表情相关参数,对源人脸进行方向和角度调整,通过颜色转移算法,将源人脸的色彩转换为目标人脸的色彩,然后利用图像融合算法进行融合。实验结果表明,本文的方法能够在图像和视频中进行有效的进行人脸替换。
中文关键词:图像处理  人脸替换  面部特征点检测  人脸建模  图像融合
 
Facial Replacement Technique in Image Based on Candide-3 Algorithm
Abstract:This paper presents a face replacement algorithm based on CANDIDE-3 algorithm. The algorithm uses CLM as the facial feature point localization algorithm, which combines the two-dimensional facial feature points with the three-dimensional vertices of CANDIDE to establish the corresponding source Face algorithm. According to the facial feature points,the head attitude and expression related parameters are estimated, and the direction and angle of the source face are adjusted. The color of the source face is converted into the color of the target face by the color transfer algorithm, and then the image fusion algorithm is used Fusion. The experimental results show that the proposed method can effectively perform face replacement in images and video.
keywords:image processing  face replacement  facial feature point detection  face modeling  image fusion
查看全文   查看/发表评论   下载pdf阅读器