Adjustment of Alignment Errors of Disparity Images Around a Crane

Kojiro Kazama, Junichi Susaki*, Yoshie Ishii

Abstract

Automated construction sites require rapid three-dimensional mapping of crane surroundings. Three-dimensional maps are generated using disparity images from a monocular camera mounted on a crane boom. Existing methods use homography matrices repeatedly to align disparity images, which accumulates errors and causes a misalignment in the origin position. In this study, a new alignment method was developed to address this challenge using the features of the crane orbit, such as the fixed base, boom length, and elevation angle. This approach uses rotation angles and swing radii estimated from homography matrices with adjustment and optimization through loop closure of the round-trip orbit of a crane. Parameter adjustments were based on fixed-point and angle comparisons between outward and return trips. The optimization was performed using the least-squares method. Experiments using round-trip images generated by the simulator yields reduced alignment errors. In future studies, we intend to verify whether the proposed method can be applied to real round-trip video images.

Keywords: Keywords: Photogrammetry, Disparity image, Alignment error, Three-dimensional map

Full Text: