Shape normalization through compacting
Shape normalization through compacting
复制标题
DOI：
10.1016/0167-8655(89)90095-0
复制
发表时间：
1989-10
期刊：
影响因子：
--
通讯作者：
J. Leu
中科院分区：
文献类型：
--
作者：
J. Leu
当物体的表面法向量不平行于观察者的观察轴时，物体的感知形状是偏斜的。现有的形状识别方法大多在形状歪斜时失效。在本文中，我们提出了一个平面形状规范化的方法，通过线性变换将感知到的形状变成最紧凑的形式，以抵消形状歪斜的影响。对于一个给定的形状，我们首先计算一个色散矩阵，它表征的形状的紧凑性。然后，我们根据色散矩阵的特征向量旋转形状，使x轴位于形状最色散的方向。最后，我们根据色散矩阵的特征值沿两个轴沿着缩放形状，以使形状达到最紧凑的形式。本文提出的技术可以作为一个预处理步骤，大多数现有的形状识别方法。
When the surface normal vector of an object is not parallel to the viewing axis of the viewer, the perceived shape of the object is skewed. Most existing shape recognition methods become ineffective when shapes are skewed. In this paper we present a planar shape normalization method to neutralize the effect of shape skewing by turning a perceived shape into its most compact form through linear transformations. For a given shape, we first compute a dispersion matrix which characterizes the compactness of the shape. Then we rotate the shape according to the eigenvectors of the dispersion matrix so that thex-axis lies in the direction in which the shape is most dispersed. Lastly, we scale the shape along the two axes according to the eigenvalues of the dispersion matrix to bring the shape to its most compact form. The technique suggested in this paper can serve as a pre-processing step for most existing shape recognition methods.