On Scalable Dimensional Reduction and Segmentation for 3D Shape Retrieval

我们已经成功地开发了一种新的降维方法，可以以可扩展的方式处理3D形状数据。我们还开发了一种高精度的3D形状描述符，适用于具有“孔”和“表面粗糙度”的机械零件，并在2013年的国际会议上获得了最佳论文奖。同时，通过应用我们的三维分割技术，我们开发了一种新的三维部分形状检索方法，以及一种从二维图像中搜索三维形状作为查询的方法。为了证明我们的方法的有效性，我们参加了SHREC2013（形状检索大赛），并获得了三维形状搜索的世界最佳精度。

We have succeeded in developing a new method for dimensional reduction that can deal with 3D shape data in a scalable way. We have also developed a highly accurate 3D shape descriptor suited to mechanical parts having "holes" and "surface roughness", with which we received the best paper award at an International Conference in 2013. Meanwhile, by applying our 3D segmentation technologies, we have developed a new method for 3D partial shape retrieval as well as a method to search 3D shapes from a 2D picture as query.To prove the effectiveness of our methods, we participated in SHREC2013 (Shape Retrieval Contest) and won the world best accuracy for 3D shape search.