BEGIN:VCALENDAR VERSION:2.0 PRODID:Linklings LLC BEGIN:VTIMEZONE TZID:Asia/Seoul X-LIC-LOCATION:Asia/Seoul BEGIN:STANDARD TZOFFSETFROM:+0900 TZOFFSETTO:+0900 TZNAME:KST DTSTART:18871231T000000 DTSTART:19881009T020000 END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20230103T035311Z LOCATION:Room 324\, Level 3\, West Wing DTSTART;TZID=Asia/Seoul:20221208T140000 DTEND;TZID=Asia/Seoul:20221208T153000 UID:siggraphasia_SIGGRAPH Asia 2022_sess165_papers_256@linklings.com SUMMARY:Shape Completion with Points in the Shadow DESCRIPTION:Technical Papers\n\nShape Completion with Points in the Shadow \n\nZhang, Zhao, Wang, Hu\n\nSingle-view point cloud completion aims to re cover the full geometry of an object based on only limited observation, wh ich is extremely hard due to the data sparsity and occlusion. The core cha llenge is to generate plausible geometries to fill the unobserved part of the object based on a partial scan, which is under-constrained and suffers from a huge solution space. Inspired by the classic shadow volume techniq ue in computer graphics, we propose a new method to reduce the solution sp ace effectively. Our method considers the camera a light source that casts rays toward the object. Such light rays build a reasonably constrained bu t sufficiently expressive basis for completion. The completion process is then formulated as a point displacement optimization problem. Points are initialized at the partial scan and then moved to their goal locations wit h two types of movements for each point: directional movements along the l ight rays and constrained local movement for shape refinement. We design n eural networks to predict the ideal point movements to get the completion results. We demonstrate that our method is accurate, robust, and generaliz able through exhaustive evaluation and comparison. Moreover, it outperform s state-of-the-art methods qualitatively and quantitatively on MVP dataset s.\n\nRegistration Category: FULL ACCESS, ON-DEMAND ACCESS\n\nLanguage: EN GLISH\n\nFormat: IN-PERSON, ON-DEMAND URL:https://sa2022.siggraph.org/en/full-program/?id=papers_256&sess=sess16 5 END:VEVENT END:VCALENDAR