1 May 2007 Fast Hough transform for automated detection of spheres in three-dimensional point clouds
Author Affiliations +
Abstract
The calibration of 3-D optical sensors normally requires the use of a calibration artifact of known dimensions. By labeling regions within the measured point clouds as belonging to a known region of the artifact, camera and projector parameters can be optimized. A novel 3-D Hough transform has been developed to extend the well-known strategy for detecting circles in 2-D images to detecting spheres in a 3-D point cloud. In its standard form, the Hough transform suffers from excessive memory storage requirements for the intermediate Hough accumulator space, which can make its application to 3-D problems impractical. We describe an accumulator implementation using an optimized sparse 3-D matrix model that provides compact data storage and efficient data access. Application of this method to experimental shape data for spheres is discussed, demonstrating its memory-saving benefits, computational efficiency, and 3-D feature detection capability.
Tokunbo Ogundana, Charles Russell Coggrave, Richard Burguete, Jonathan Mark Huntley, "Fast Hough transform for automated detection of spheres in three-dimensional point clouds," Optical Engineering 46(5), 051002 (1 May 2007). https://doi.org/10.1117/1.2739011
JOURNAL ARTICLE
11 PAGES


SHARE
Back to Top