基于多视点图像的人脸表面光场重构

    Surface Light Field Rendering of Face Based on Multi-images

    • 摘要: 为了实现任意新视点下三维人脸模型的绘制,针对三维人脸模型数据的特殊性,提出一种借助一般三维人脸模型基于稀疏多视点图像重建特定人脸表面光场的方法,该方法改进了Laplacian网格变形技术,实现了一般人脸三维模型到多视点图像中特定人脸三维模型的变形,避免了直接扫描人脸模型的操作复杂、模型粗糙、数据量大等缺点. 最后,通过任意视点下人脸模型绘制验证了基于多视点图像重构获得的人脸表面光场数据的有效性.

       

      Abstract: For rendering the 3D human face model at new views, according to the specificity of 3D face model, this paper suggests human face surface light field reconstruction based on sparse multi-view images with the help of general face model. The method improved Laplacian mesh deformation technology to realize deformation from a general human face model to a specific face model which is fitted for the face defined in the multi-view images. The method avoided some disadvantages such as complex operation, coarse model and large amount data. Finally, rendering the human face model at new view was operated to validate the effectiveness of surface light field reconstruction.

       

    /

    返回文章
    返回