The study of different panorama generation methods and their applications

In this paper, a new method is proposed to reconstruct a 3D image from video captured by a rotating camera. Our approach uses a large collection of images taken by a camera whose motion has been constrained to planar concentric circles. We resample regular perspective images to produce a set of multi-perspective panoramas and then compute depth maps directly from these resampled panoramas. Our panoramas sample uniformly in three dimensions: rotation angle, inverse radial distance and vertical elevation. The use of multi-perspective panoramas eliminates the limited overlap present in the original input images and, thus, problems such as those in conventional multi-baseline stereo can be avoided. Our approach differs from stereo matching of single-perspective panoramic images taken from different locations, where the epi-polar constraints are sine curves. For our multi-perspective panoramas, the epi-polar geometry, to the first order approximation, consists of horizontal lines. Therefore, any traditional stereo algorithm can be applied to multi-perspective panoramas with little modification. The method is based on a temporal-spatial slice, which allows fast and reliable 3D reconstruction from common video without a special camera. Experiments show that our algorithms are capable of producing comparable high-quality depth maps, which can be used for applications such as view interpolation and for 3-D measuring of real objects. Also in this work, we propose a simple yet effective method for improving the performance of local feature matching among rectangular cylindrical images, which brings more stable and complete 3D reconstruction by incremental SfM (incremental/sequential Structure-from-Motion). The key idea is to explicitly generate synthesised images by rotating the spherical panoramic images so that circular panoramic images are converted to spherical panoramic images, and to detect and describe features only from the less distorted area in the rectified panoramic images. We demonstrate that the proposed method is advantageous for both rotational and translational camera motions compared with the standard methods on the synthetic data.