结合多分辨率、网格排布和基于视点的绘制技术,提出一种外存多分辨率构建和绘制算法.采用适应性八叉树对模型的包围盒进行划分,自顶向下构建模型的多分辨率层次结构,较好地保持了原模型的细节分布;并对多分辨率结构中每个节点所包含的三角形片段进行网格排布优化,降低了缓存的平均失效率;在实时绘制时,采用基于视点的细节层次选择策略进行模型的细化;最后通过引入数据预取机制来隐藏磁盘I/O延时,进一步提高绘制性能.实验结果表明,该算法在绘制速度与细节保留上均优于同类MRMM算法.
This paper proposes a novel approach for constructing and rendering out-of-core multiresolution representation for massive polygonal meshes. It keeps the details of the original model by adaptive octree partitioning of the mesh bounding box, and reduces the average cache miss rate through mesh layout optimization for triangles contained in the octree node. At rendering stage, the multiresolution representation structure is refined and rendered with view-dependent .LOD selection. At last, the approach employs data prefetch strategy to hide I/O latency and then further improves the rendering performance. Experimental results show that the proposed approach is superior to similar MRMM algorithms in term of rendering speed and capability of preserving details.