Scale-space analysis for point sets on the sphere and applications to computer vision

尺度空间分析在欧几里得空间中得到了典型的讨论。本研究将分析扩展到球面空间，以揭示空间的理论基础，并为计算机视觉问题提供一些实用的方法。将球面聚类和球面光流计算两种方法应用于大视场的全向相机图像，对其进行了评价。结果表明，所提出的方法使我们能够高效、鲁棒地构建各种基于视觉问题的系统。

Scale-space analysis has been typically discussed on the Euclidean space. This study extends the analysis to the spherical space to reveal theoretical foundations on the space and to develop some practical methods for computer vision problems.Two methods, spherical clustering and spherical optical flow computation, were evaluated by applying to images captured by omnidirectional camera, which has a large field of view. The results show that the proposed methods enable us to build the various systems for vision-based problems efficiently and robustly.