网格参数化与几何图像

将三角网格模型进行参数化并重采样生成几何图像,借助当前成熟、高效的图像处理技术,利用图像代替复杂三角网格进行数据处理,可以有效地解决三维几何数据压缩、传输、渐进表示等问题。作者引入了不同的参数化方法生成三维模型的几何图像,并分析了Tutte重心映射法、离散保角映射法、离散保积参数化方法、Floater均值坐标法和几何拉伸参数化方法五种参数化方法在几何图像方法中对三维模型重构的影响。实验结果表明,采用几何拉伸方法所生成的几何图像具有最好的PSNR值和主观的重构质量。

Mesh Parameterizations and Geometry Images

By using highly efficient image coding technique,geometry image becomes an important tool for compression and progressive representation of geometry data.The geometry images method was designed and implemented based on several classic mesh parameterizations:barycentric mapping parameterization,discrete conformal parameterization,discrete authalic parameterization,mean value coordinate parameterization and geometric-stretch parameterization.Experiments show that the geometric-stretch parameterization provides the best results on both PSNR values and reconstructive quality.