【作者】 赵向阳； 杜利民；

【Author】 Zhao Xiangyang~(1,2) Du Limin~1 1 Speech Interactive technology Research, Institute of Acoustics, CAS, Beijing 100080 2 Graduate School of Chinese Academy of Sciences, Beijing 100039

【机构】 中国科学院声学研究所语音交互室；

【摘要】 本文利用薄平面样条函数改善了人头模型调整,提出了基于模型分块的视角相关纹理映射新方法,该方法将模型分块并用每块对应的最清晰、非冗余的纹理来做映射,在块与块之间形成了规则过渡带非常易于纹理融合,整个模型逐块地一次渲染完成,并在模型中嵌入了抽象的人脸肌肉,合成出逼真的人脸表情动画。实验表明整个算法快速、简捷,合成的人脸图像细节清晰、逼真。更多还原

【Abstract】 In this paper, the smoothness of thin-plate splines is applied to improve the model adaptation, and a new method of model blocking based view-dependent texture mapping is proposed. In this method, the model is divided into parts and each part is mapped with the clearest and non- redundant texture, which makes the transition band between blocks regular very much and do texture blending easily. The model is rendered one time block by block. Then, many abstract muscles are embedded into the model and realistic face expressions are synthesized. Experiments indicate the algorithm is fast and compact, and the synthesized face is very realistic, and the face details are clear very much.更多还原