摘要
提出了一种基于小波变换的人脸语音动画合成的平滑方法:将小波变换与数学形态学相结合,从相应音素的图像帧中提取出特征区域的边缘,实现特征点的精确定位。在此基础上进一步对相应区域进行三角剖分,通过前后两帧图像之间的变形产生中间帧,最终合成较为平滑、真实的人脸语音动画效果。
This paper provides the method of smoothing based on wavelet transform in face speech animation synthesis system.The method,which extracts the edge of feature regions from correlation image frames of phonemes using the wavelet transform and mathematic morphology,can locate the feature points exactly.Afterwards it divides the feature areas with these points by triangle and interpolates some frames between existing frames by morphing.At last,the animation effect is smooth and reality.
出处
《计算机工程与应用》
CSCD
北大核心
2005年第28期49-51,共3页
Computer Engineering and Applications
基金
安徽省自然科学基金(编号:01042203)
关键词
小波变换
数学形态学
图像变形
语音动画
wavelet transform, mathematic morphology ,image morphing, speech animation
