ETRI-Knowledge Sharing Plaform

KOREAN
논문 검색
Type SCI
Year ~ Keyword

Detail

Conference Paper A View Warping and Blending Algorithm for Full-Parallax Virtual View Image Synthesis on Light-Field Content
Cited - time in scopus Share share facebook twitter linkedin kakaostory
Authors
Youngsoo Park, Hong-chang Shin, Gwangsoon Lee, Won-sik Cheong, Jeongil Seo
Issue Date
2017-08
Citation
International Conference on 3D Systems and Applications (3DSA) 2017, pp.790-790
Language
English
Type
Conference Paper
Abstract
Light-field content is one of the visual media to provide full-parallax 3D viewing [1]. However, a light-field imaging from a camera-array system or a specialized camera with micro-lens arrays for capturing light field cannot display directly when the number of view is different between a capturing equipment and a light-field display [2]. Therefore, virtual view images synthesis is required to display light-field content. We present a view warping and blending algorithm for full-parallax virtual view synthesis on light-light field content. The proposed algorithm consists of depth map calculation for targeted view, four-directional back-ward warping, and disocclusion hole filling [3], [4]. Fig 1 . shows that the pipeline of the proposed algorithm. For calculating depth map for targeted view, we warped the depth map at the closest view in priority and performed the spatial-view referenced hole filling in remained area on depth map. Then, we execute four-directional back-ward warping with the depth information and dis-occlusion hole filling. Experimental results show that the proposed algorithm can generate full-parallax dense view images. The proposed algorithm can be used for light-field content generation and expected that it can contribute to use light-field products widely in practice environments.
KSP Keywords
Depth Map, Depth information, Light field display, Micro-lens, Visual Media, camera array, content generation, hole filling, image synthesis, light field imaging, virtual view synthesis