Single-shot capture of panoramic images has become possible in recent years, expanding applications for omnidirectional cameras. Retrofitting omnidirectional cameras onto existing objects can be problematic, however, because placing lenses on either side can cause the mounting structure to be included in captured images. To address this issue, we propose using two hemispherical cameras equipped with 185° field-of-view fisheye lenses and combining the images from each while compensating for differences in separation and height to produce a panoramic image like that from an omnidirectional camera. In other words, we investigate acquisition of omnidirectional panoramic images from two hemispherical cameras installed at some distance from each other. In this article, we propose a method for generating spherical images by stitching together images from two cameras. Specifically, we prepare a circular object, include it in images from two separated hemispherical cameras, detect the circular object in each image, and match the two detected circles. We also describe multiple experiments confirming the utility and effectiveness of the proposed method. This work is expected to contribute to expanding the range of applications for omnidirectional imaging.
ASJC Scopus subject areas
- Human-Computer Interaction
- Hardware and Architecture
- Computer Science Applications
- Electrical and Electronic Engineering