None ghosting artifacts stitching based on depth map for light field image

Zhang, Wenyuan, Zhao, Shengyang, Zhou, Wei and Chen, Zhibo 2018. None ghosting artifacts stitching based on depth map for light field image. Presented at: 19th Pacific-Rim Conference on Multimedia, Hefei, China, 21-22 Sept 2018. Published in: Hong, Richang, Cheng, Wen-Huang, Yamasaki, Toshihiko, Wang, Meng and Ngo, Chong-Wah eds. Advances in Multimedia Information Processing – PCM 2018. Lecture Notes in Computer Science. Lecture Notes in Computer Science , vol.11164 Springer, pp. 567-578. 10.1007/978-3-030-00776-8_52

Full text not available from this repository.

Abstract

Due to hardware limitations, existing light field (LF) capturing devices cannot offer sufficient field of view for building 6 degrees of freedom (6 DOF) VR applications. LF image stitching methods can be used to address this problem. The state-of-the-art LF stitching methods highly depend on the stitching accuracy of center view which is essentially a 2D image stitching task. However, conventional 2D image stitching methods usually suffer from the ghosting artifacts. In this paper, a None Ghosting Artifacts (NGA) stitching method is proposed to tackle this problem. We theoretically reveal the intrinsic cause of the ghosting artifacts and then further verify that different depth scenes require different homography matrices for warping. Therefore, the clustered depth map is employed to segment the scene into several layers, and the layer-specific homography matrix is computed for warping. An interpolation mechanism is also proposed to ensure that each layer has its own transformation. Compared with state-of-the-art stitching methods aiming to alleviate ghosting artifacts, experimental results show that the proposed method not only stitches images without ghosting artifacts, but also achieves realistic perspective transformation.

Item Type:	Conference or Workshop Item (Paper)
Date Type:	Published Online
Status:	Published
Schools:	Computer Science & Informatics
Publisher:	Springer
ISBN:	978-3-030-00775-1
ISSN:	0302-9743
Last Modified:	24 Aug 2023 10:56
URI:	https://orca.cardiff.ac.uk/id/eprint/161677

Actions (repository staff only)

Edit Item