Surface Fairing and Surface Compression for 3D Ship Modeling

本研究的目的是开发一种用于三维船体设计的自动表面整流罩方法和一种用于三维数据交换的表面压缩方法。试验结果如下：1。表面整流罩由传统整流罩方法产生的表面通常具有不需要的特征，如局部扰动和全局波。本研究开发了一种新的整流罩方法，避免了这些特征。该方法包括以下三个步骤。(1)重新排列线-曲线控制点，使其弧长和曲率的加权和均匀分布。(2)反复应用低通滤波器，沿直线曲线重新布置控制点。(3)为了消除局部扰动，采用薄板能量最小化技术，将每个表面视为薄板。我们的方法应用于船体形状表面的例子，可以有效地消除整体波浪和局部扰动，而形状的变化有限。表面压缩表面压缩是存储和分布三维CAD数据的有效方法，但现有的压缩方法不能保持表面边界的平滑性。在我们的方法中，NURBS曲面使用其边界和附加数据进行编码。首先，利用原曲面的边界曲线计算插值曲面。然后，计算原始曲面与插值曲面之间的差值。差异数据可选地使用，它们以两种类型表示：仅距离和距离和方向。根据每个表面的公差选择一种类型。最后，利用离散余弦变换（DCT）对边界曲线和差分数据进行压缩。结果表明，该方法可以很容易地控制压缩比和精度之间的权衡，并获得良好的压缩率。(291字)

The purposes of this research are the development of an automatic surface fairing method for 3D ship-hull design and the development of a surface compression method for exchanging 3D data. The results of them are described as follows :1. Surface FairingSurfaces generated by conventional fairing methods often have unwanted features such as local perturbations and global waves. This research developed a new fairing method that avoids these features. The method consists of the following three steps. (1) Rearranging control points of lines-curves so that the weighted sum of their arc lengths and curvatures are equally distributed. (2) Rearranging control points along lines-curves by applying low-pass filters repeatedly. (3) For eliminating the local perturbations, an energy minimization technique for thin-plates is used by regarding each surface as a thin-plate. Our method was applied to examples of hull-form surfaces, and could effectively remove the both of global waves and local perturbations with limited changes of the shape.2. Surface CompressionSurface compression is effective for storing and distributing 3D CAD data, but existing compression methods failed to maintain the smoothness at surface boundaries. In our method, a NURBS surface is encoded using its boundaries and additional data. At first, an interpolating surface is calculated using the boundary curves of the original surface. Then, the differences between the original surface and the interpolating one are calculated. Difference data are optionally used, and they are represented in two types : distances only, and distances and orientations. One type is selected according to the tolerances of each surface. Finally, the boundary curves and the difference data are compressed using the discrete cosine trans-form (DCT). The results show our method can easily control the trade-off between the compression ratio and accuracy, and achieve a good compression rates. (291 words)