基于GPU的复杂三角网格模型多分辨率绘制

Multiresolution rendering of complex triangular mesh model based on GPU

针对高度细节三角网格模型实时绘制问题,通常使用LOD技术对模型进行处理,本文提出了一种适用于GPU的针对零亏格的三角网格模型的多分辨率绘制方法。该方法先将三角网格进行球面参数化,然后利用参数化结果构建几何图像,并在其上建立四叉树模型,最后使用GPU着色器进行模型细节层次的选择并对网格进行简化,实现模型的多分辨率绘制。与已有方法相比,利用GPU高度并行运算的特点,减少了简化模型过程的计算时间。实验结果表明,该方法在保证绘制质量的同时达到实时绘制的要求。

In this paper,an approach is proposed for multiresolution rendering of triangular mesh model with zero genus which based on GPU.The triangular mesh parameterized onto a spherical domain,then the geometry images is constructed with the parametric results.The quad-tree is constructed on the geometry image,finally the LOD selection and mesh simplification is performed in the shader.The time of the simplified model process is reduced with use of GPU highly parallel computing features compared with the existing methods.The experimental results show that this approach ensure the quality of rendering and meet the requirements of real-time rendering.