|
Title:
|
POINT CLOUD REGISTRATION WITH SURFACE DESCRIPTORS |
|
Author(s):
|
Luis Gerardo de la Fraga, Daniel López-Escogido |
|
ISBN:
|
978-989-8533-52-4 |
|
Editors:
|
Katherine Blashki and Yingcai Xiao |
|
Year:
|
2016 |
|
Edition:
|
Single |
|
Keywords:
|
Point cloud registration, surface registration, robust registration, surface descriptors, Levenberg-Marquardt |
|
Type:
|
Short Paper |
|
First Page:
|
309 |
|
Last Page:
|
314 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
In computer vision a common problem is the registration of 3D data, which consists in to minimize the sum of squares of the distance between metrical marker correspondences. Commonly used marker correspondences are points, in two or more datasets. The most popular algorithm used to solve this problem is the Iterative Closest Point (ICP) due to its easy implementation and fast convergence. In this paper is introduced a methodology using some features of ICP and a general-use-optimizer like Levenberg-Marquardt. The proposed approach also uses vectors as surface descriptors, and those vectors are used to describe planes, cylinders, and cones, contained in synthetic and real shapes. Additionally in the sphere model is used its center point. Using this information, it is proposed the minimization of an objective function which gives more weight to the part of surface alignment. Experimental results show us that the registration using surface descriptors is more efficient, it can be solved both with fewer iterations and less execution time, than using only points. |
|
|
|
|
|
|
