[POSTER] Position Estimation of a Strongly Occluded Object by Using an Auxiliary Point Cloud in Occluded Space

A method is proposed for estimation of occluded space and generation of auxiliary points for 3D position estimation of strongly occluded objects. First, occlusion space detection calculates 3D keypoints at the rear side of a target object, thus obtaining a silhouette around the object on the near si...

Full description

Saved in:
Bibliographic Details
Published in2017 IEEE International Symposium on Mixed and Augmented Reality (ISMAR-Adjunct) pp. 194 - 199
Main Author Sumiyoshi, Shinichi
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.10.2017
Subjects
Cameras
Estimation
Image edge detection
Portable computers
Robustness
Sensors
Three-dimensional displays
Online AccessGet full text

Cover

More Information
Summary:A method is proposed for estimation of occluded space and generation of auxiliary points for 3D position estimation of strongly occluded objects. First, occlusion space detection calculates 3D keypoints at the rear side of a target object, thus obtaining a silhouette around the object on the near side, as found from a camera image by an object detector. The method calculates the space containing the 3D keypoints and defines it as the occluded space. The method then generates an auxiliary point cloud for unobserved regions of this space. By matching the detected point cloud and the auxiliary point cloud, the method can estimate the position of an occluded object that has been difficult to localize with a conventional method, thus giving a general matching technique. The centroid position accuracy of the proposed method was experimentally evaluated to demonstrate its effectiveness and confirm its validity.
DOI:10.1109/ISMAR-Adjunct.2017.64