A review of recent range image registration methods with accuracy evaluation

Abstract

The three-dimensional reconstruction of real objects is an important topic in computer vision. Most of the acquisition systems are limited to reconstruct a partial view of the object obtaining in blind areas and occlusions, while in most applications a full reconstruction is required. Many authors have proposed techniques to fuse 3D surfaces by determining the motion between the different views. The first problem is related to obtaining a rough registration when such motion is not available. The second one is focused on obtaining a fine registration from an initial approximation. In this paper, a survey of the most common techniques is presented. Furthermore, a sample of the techniques has been programmed and experimental results are reported to determine the best method in the presence of noise and outliers, providing a useful guide for an interested reader including a Matlab toolbox available at the webpage of the authors.

Scan segments matching for pairwise 3D alignment

“…Standard alignment techniques use the Iterative Closest Point (ICP) algorithm [22] which solves data association at a point-to-point level 1 . However, as noted in the literature [23] and as shown in the proposed experiments, this method fails when nearest points do not correspond to correct association (illustrated in Fig. 3).…”

“…A recent survey [23] shows that 3D alignment techniques have converged around one main pipeline which involves two broad steps: coarse alignment followed by fine alignment. Coarse alignment uses for instance Principal Component Analysis [15] or feature based matching with features such as the Spin Image [14] to compute an initial coarse registration.…”

A robotic concept for remote maintenance operations: A robust 3D object detection and pose estimation method and a novel robot tool

“…Salvi et al [13] found the DARCES method [17] to be the most robust method for range image registration, and DARCES is also the closest published method to the one we are using in this paper. The most important difference between DARCES and the presented method is that DARCES is based on the presence of three points, while our method only requires two points.…”

“…Existing methods in this field can be grouped into coarse and fine registration methods [13]. The goal of coarse registration methods is to compute an initial estimate of the rigid motion of the object, which eventually can be refined by more precise methods later.…”

A robust and fast method for 6DoF motion estimation from generalized 3D data

“…Other approaches used to reject invalid matches based on some compatibility test Turk and Levoy (1994); Masuda et al (1996); Dorai et al (1998) also depend on the 3D sensor used. In Salvi et al (2007) a deeper study comparing different solutions for ICP in terms of speed and accuracy is performed. Nowadays, more ICP variations are still appearing Du et al (2007a,b); Nuchter et al (2007);Censi (2008); Armesto et al (2010).…”