An Accurate and Robust Range Image Registration Algorithm for 3D Object Modeling

Abstract

Abstract-Range image registration is a fundamental research topic for 3D object modeling and recognition. In this paper, we propose an accurate and robust algorithm for pairwise and multi-view range image registration. We first extract a set of Rotational Projection Statistics (RoPS) features from a pair of range images, and perform feature matching between them. The two range images are then registered using a transformation estimation method and a variant of the Iterative Closest Point (ICP) algorithm. Based on the pairwise registration algorithm, we propose a shape growing based multi-view registration algorithm. The seed shape is initialized with a selected range image and then sequentially updated by performing pairwise registration between itself and the input range images. All input range images are iteratively registered during the shape growing process. Extensive experiments were conducted to test the performance of our algorithm. The proposed pairwise registration algorithm is accurate, and robust to small overlaps, noise and varying mesh resolutions. The proposed multi-view registration algorithm is also very accurate. Rigorous comparisons with the state-of-theart show the superiority of our algorithm.

Performance Evaluation of 3D Local Feature Descriptors

“…Local features have proven to be very successful in many vision tasks such as 3D object recognition, 3D modeling, 3D shape retrieval, and 3D biometrics [1][2][3][4][5][6]. Local features have been extensively investigated during the last few decades with the aim to design descriptors which are distinctive, robust to occlusions and clutter [7].…”

A novel local surface feature for 3D object recognition under clutter and occlusion

“…That is, the TriSI-based algorithm results in less false positives compared with the other algorithms. 93 100 100 100 100 100 RoPS [17] 97 100 100 100 100 97 100 100 100 100 SHOT+Game [35] Although the proposed TriSI feature is only used for 3D object recognition in this paper, other possible application areas include 3D model/shape retrieval [2], 3D face recognition [47,26,27], and 3D object/scene reconstruction [16,21].…”

Multi-attribute statistics histograms for accurate and robust pairwise registration of range images

“…The development of numerous low-cost 3D sensors, e.g., Microsoft Kinect and Intel RealSense, has further promoted such trend in recent years. Pairwise registration of range images is a fundamental and crucial step in many 3D vision tasks, such as 3D object recognition [1] , surface registration [2] , 3D modeling [3] , cultural heritage [4] and robotics [5] .…”

“…To solve above difficulties, many effort s have been made in the literature. A matured pipeline for range image registration is a coarse-to-fine strategy [3,[6][7][8] , where the coarse registration provides an initial guess for the fine registration as well as accelerates the registration process. The most typical fine registration methods are arguably the iterative closest point (ICP) algorithm and its variants [9][10][11] .…”

“…Since the existing fine registration methods are already able to obtain fairly accurate results [12] , the real challenging turns out to be the coarse registration problem, which has attracted many interests lately. Feature-based coarse registration methods are mostly used nowadays owing to their validated effectiveness and robustness [3,6,[13][14][15][16][17] , their employed features can be classified into local and global features. Owing to the superiority of local features against global features in terms of distinctiveness and robustness to clutter, occlusion, and missing regions [12] , local features are more suitable for generating correspondences between range images.…”