Graph based over-segmentation methods for 3D point clouds

Abstract

Over-segmentation, or super-pixel generation, is a common preliminary stage for many computer vision applications. New acquisition technologies enable the capturing of 3D point clouds that contain color and geometrical information. This 3D information introduces a new conceptual change that can be utilized to improve the results of over-segmentation, which uses mainly color information, and to generate clusters of points we call super-points. We consider a variety of possible 3D extensions of the Local Variation (LV) graph based over-segmentation algorithms, and compare them thoroughly. We consider different alternatives for constructing the connectivity graph, for assigning the edge weights, and for defining the merge criterion, which must now account for the geometric information and not only color. Following this evaluation, we derive a new generic algorithm for over-segmentation of 3D point clouds. We call this new algorithm Point Cloud Local Variation (PCLV). The advantages of the new oversegmentation algorithm are demonstrated on both outdoor and cluttered indoor scenes. Performance analysis of the proposed approach compared to state-of-the-art 2D and 3D over-segmentation algorithms shows significant improvement according to the common performance measures.

An Iterative Coarse-to-Fine Sub-Sampling Method for Density Reduction of Terrain Point Clouds

“…A simple area partition method (i.e., square blocks) was used in this study. As a suggestion for further work, it would be interesting to investigate the performance of the coarse-to-fine method when more complex area partition methods (e.g., over-segmentation approaches such as voxel cloud connectivity segmentation and point cloud local variation [29][30][31][32][33]) were used. Such methods may take into account the terrain surface complexity for the area partition.…”

Developing a Semantic-Driven Hybrid Segmentation Method for Point Clouds of 3D Shapes

“…Multiple region-growing processes are performed according to the LCCHs, the time complexity of which is O (pr • neigh), where pr is the number of the points in some particular regions (pr<pn). So the time complexity of the semantic clustering stage is O pn 2 + pn • neigh + pr • neigh = O pn 2 . The variational merging stage includes the process of calculating the errors between the regions and their neighbors and the process of regional merging.…”

“…The time complexity of calculating the errors for all regions is O r_num 2 , and the time complexity of the regional merging is O(r_num). So the time complexity of the traditional method is O r_num 2 + r_num = O r_num 2 . In this paper, in order to reduce the number of error calculations, we use the regions as the unit of a neighbor in the error calculation method.…”

“…Thanks to these developments, the 3D point cloud is able to contain rich information nowadays. There are now a wide range of the applications which make use of 3D point cloud data [2], [3].…”

Modified JSEG algorithm for reducing over-segmentation problems in underwater coral reef images

“…Segmentation of color image has been very popular for various applications and owing to its diverse application types. Still up to today, image segmentation remains a challenging subject [9,10]. One of the most robust methods of segmenting natural imagesis the JSEG algorithm [11].…”