Dynamic acquisition of SVBRDF, geometry and motion using a single RGBD camera = RGBD 카메라를 이용한 동적 기하 정보 및 표면 반사 성질 동시 추출 기술

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 215
  • Download : 0
We propose a fusion method that simultaneously captures spatially-varying BRDF and geometry of an object in motion using a single RGBD camera. State-of-the-art fusion methods capture geometry and often diffuse albedo, but no specular reflectance. Traditional SVBRDF acquisition methods capture many input images to obtain dense samples of viewing and lighting angles using a single camera or a multiview setup, followed by an offline process of inverse rendering which often takes several hours. In this thesis, we introduce a novel fusion method that can capture SVBRDF, geometry, and motion of dynamic objects in an interactive manner. Based on observation under both environment illumination and infrared light of an RGBD camera, we introduce a novel online optimization method that jointly estimates both diffuse and specular appearance parameters in addition to geometry and motion per frame, by accumulating photometric observations in a half-angle buffer in a voxel grid. We progressively refine dynamic geometry and spatially-varying appearance in clusters, enabling interactive online feedback at an averaged speed of 880\,msec.~per frame. Our experimental results of various dynamic objects that include the mixture of diffuse and specular reflection validate our fusion method's robust performance in estimating both SVBRDF and geometry in motion using a single RGBD camera.
Advisors
Kim, Min H.researcher김민혁researcher
Description
한국과학기술원 :전산학부,
Publisher
한국과학기술원
Issue Date
2019
Identifier
325007
Language
eng
Description

학위논문(석사) - 한국과학기술원 : 전산학부, 2019.2,[iv, 30 p. :]

Keywords

SVBRDF▼ageometry▼amotion▼aRGBD Camera▼adynamic 4D scanning; 표면 반사 성질▼a기하 정보▼a동적 정보▼aRGBD 카메라▼a동적 4D 복원

URI
http://hdl.handle.net/10203/267026
Link
http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=843547&flag=dissertation
Appears in Collection
CS-Theses_Master(석사논문)
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0