Calibrate a moving camera on a linear translating stage using virtual plane + parallax

Xiaoqiang Zhang,Yanning Zhang,Tao Yang,Zhengxi Song

DOI: https://doi.org/10.1007/978-3-642-36669-7_7

2013-01-01

Abstract:For a fixed camera array, the limited number of views restricts the quality of the imaging result. In this paper, we build a virtual dense camera array using different views from a single moving camera to increase the number of view conveniently. A virtual plane + parallax based calibration algorithm is proposed to estimate the relative positions between different views. The main characteristics of this system and calibration approach include: (1) Due to the mobility, one can vary views in the light of different scene configurations. (2) Via calibrating the moving camera under the plane + parallax framework, the computation cost of aligning images from all views is reduced. (3) A moving camera system is set up by mounting a camera on a translating stage which can slide on a linear track. Extensive experiment results demonstrate that our approach can enhance detail and quality of the occluded objects effectively compared to the fixed camera array synthetic aperture imaging approach.