P. Forsman (Finland)
3-D object extraction, range image processing
In this paper a method for the modeling of cylindrical objects from range image data is presented. The contribution is related to the iterative algorithm for approximating (curved) elongated objects with a circular cross-section by a sequence of circle models across its’ visible length. Moreover, the method can be configured to accept objects, or rather, measurement echoes from objects, which indicate different levels of circularity in the object cross-section. This is particularly important due to the wide laser beam pattern and significant divergence of the beam, which increases the noise in the measurement data as a function of the measured distance. Also, tree trunks, which are the principal target object group of the method in natural environments, do not have an absolutely circular cross section.
Important Links:
Go Back
