Share Email Print

Proceedings Paper

360x180 stereoscopic panoramas from static light field capturing
Author(s): Kai Wang; Chunhong Wu; Shaojie Ren; Feng Xiao; Yunluo Liu
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

With the increasing demand in visual experience, visual effects with good immersion and stereoscopic sense have become one of the research hotpots in computer vision and computer graphics. This paper presents a stereoscopic panoramic image generation method based on static light field capture. The work can be divided into two parts: the acquisition of the static light field and the stereoscopic panoramic image generation. The light field data is recorded by two mutually perpendicular cameras rotated around a fixed axis. The two cameras capture images with every 9 rotating degrees and a total of 40x2 images were obtained after spin a circle. Both two cameras are calibrated and the recorded images are re-projected to a spherical surface. The optical flow information was calculated from the overlapping part of the spherical projection using a block matching method based on pyramid model. As a result, the light filed information which contains texture and disparity information is acquired based on the spherical expression. On the basis of the above work, the stereoscopic panoramic images under the new viewpoints were then resampled densely based on a pixel level from the obtained light field. A Poisson blending algorithm is used afterward to fuse the image strips from different perspectives and eliminate the brightness difference between those different strips. The high quality stereoscopic panoramic images with an 8k resolution were achieved in the end. Compared with the commonly used panoramic image generating method, our method has the advantages of low cost, simple operation, and high automation. More importantly, it can generate stereoscopic panoramic images containing parallax information with the vertical 180 degrees of view.

Paper Details

Date Published: 6 May 2019
PDF: 11 pages
Proc. SPIE 11069, Tenth International Conference on Graphics and Image Processing (ICGIP 2018), 1106938 (6 May 2019); doi: 10.1117/12.2524410
Show Author Affiliations
Kai Wang, Univ. of Science and Technology (China)
Chunhong Wu, Univ. of Science and Technology Beijing (China)
Shaojie Ren, Univ. of Science and Technology Beijing (China)
Feng Xiao, Beijing FengYun Vision Technology Co., Ltd. (China)
Yunluo Liu, Univ. of Science and Technology (China)

Published in SPIE Proceedings Vol. 11069:
Tenth International Conference on Graphics and Image Processing (ICGIP 2018)
Chunming Li; Hui Yu; Zhigeng Pan; Yifei Pu, Editor(s)

© SPIE. Terms of Use
Back to Top
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?