Locating objects in spherical panoramic images

In this paper, a method for locating objects in spherical panorama is proposed. Given a set of reference images for reference stored in the database and a query image, we would find the nearest image in the database, and then label the corresponding location on the spherical panorama. Since there are differences between spherical panoramic image and query view which is mostly captured by normal camera, the traditional methods could not be used to solve this problem directly. Therefore, we transform the query image into parts of spherical panorama, and derive the essential matrix and homograph matrix in spherical panoramic image based on the property of spherical panorama. We adopt SIFT to detect the features in spherical panorama, and then reject the wrong matching points using RANSAC according to the essential matrix we derived. The remaining matching feature points are still more than others in spherical panorama database. We calculate the homograph matrix, and then the corresponding position in spherical panoramic image will be detected. The experiments verify the validity of the method proposed in this paper.

[1]  Chak-Kuen Wong,et al.  Worst-case analysis for region and partial region searches in multidimensional binary search trees and balanced quad trees , 1977, Acta Informatica.

[2]  G LoweDavid,et al.  Distinctive Image Features from Scale-Invariant Keypoints , 2004 .

[3]  Sunil Arya,et al.  An optimal algorithm for approximate nearest neighbor searching fixed dimensions , 1998, JACM.

[4]  Ankita Kumar,et al.  Experiments on visual loop closing using vocabulary trees , 2008, 2008 IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops.

[5]  Robert Laganière,et al.  Epipolar Geometry for the Rectification of Cubic Panoramas , 2006, The 3rd Canadian Conference on Computer and Robot Vision (CRV'06).

[6]  Jun Fujiki,et al.  Epipolar Geometry Via Rectification of Spherical Images , 2007, MIRAGE.

[7]  Jana Kosecka,et al.  Experiments in place recognition using gist panoramas , 2009, 2009 IEEE 12th International Conference on Computer Vision Workshops, ICCV Workshops.

[8]  Andrew Zisserman,et al.  Robust computation and parametrization of multiple view relations , 1998, Sixth International Conference on Computer Vision (IEEE Cat. No.98CH36271).

[9]  Michael Kroepfl,et al.  Efficiently locating photographs in many panoramas , 2010, GIS '10.