Convex Variational Methods on Graphs for Multiclass Segmentation of High-Dimensional Data and Point Clouds

被引:17
作者
Bae, Egil [1 ]
Merkurjev, Ekaterina [2 ]
机构
[1] Norwegian Def Res Estab, POB 25, N-2027 Kjeller, Norway
[2] Michigan State Univ, 220 Trowbridge Rd, E Lansing, MI 48824 USA
关键词
Variational methods; Graphical models; Convex optimization; Semi-supervised classification; Point cloud segmentation; DIFFUSE INTERFACE METHODS; MARKOV RANDOM-FIELDS; IMAGE SEGMENTATION; GLOBAL MINIMIZATION; WEIGHTED GRAPHS; CLASSIFICATION; REGULARIZATION; FRAMEWORK; ALGORITHMS; MODELS;
D O I
10.1007/s10851-017-0713-9
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Graph-based variational methods have recently shown to be highly competitive for various classification problems of high-dimensional data, but are inherently difficult to handle from an optimization perspective. This paper proposes a convex relaxation for a certain set of graph-based multiclass data segmentation models involving a graph total variation term, region homogeneity terms, supervised information and certain constraints or penalty terms acting on the class sizes. Particular applications include semi-supervised classification of high-dimensional data and unsupervised segmentation of unstructured 3D point clouds. Theoretical analysis shows that the convex relaxation closely approximates the original NP-hard problems, and these observations are also confirmed experimentally. An efficient duality-based algorithm is developed that handles all constraints on the labeling function implicitly. Experiments on semi-supervised classification indicate consistently higher accuracies than related non-convex approaches and considerably so when the training data are not uniformly distributed among the data set. The accuracies are also highly competitive against a wide range of other established methods on three benchmark data sets. Experiments on 3D point clouds acquire by a LaDAR in outdoor scenes and demonstrate that the scenes can accurately be segmented into object classes such as vegetation, the ground plane and human-made structures.
引用
收藏
页码:468 / 493
页数:26
相关论文
共 102 条
  • [21] MULTIDIMENSIONAL BINARY SEARCH TREES USED FOR ASSOCIATIVE SEARCHING
    BENTLEY, JL
    [J]. COMMUNICATIONS OF THE ACM, 1975, 18 (09) : 509 - 517
  • [22] DIFFUSE INTERFACE MODELS ON GRAPHS FOR CLASSIFICATION OF HIGH DIMENSIONAL DATA
    Bertozzi, Andrea L.
    Flenner, Arjuna
    [J]. MULTISCALE MODELING & SIMULATION, 2012, 10 (03) : 1090 - 1118
  • [23] Markov random fields with efficient approximations
    Boykov, Y
    Veksler, O
    Zabih, R
    [J]. 1998 IEEE COMPUTER SOCIETY CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, PROCEEDINGS, 1998, : 648 - 655
  • [24] Boykov Y, 2001, LECT NOTES COMPUT SC, V2134, P359
  • [25] Fast approximate energy minimization via graph cuts
    Boykov, Y
    Veksler, O
    Zabih, R
    [J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2001, 23 (11) : 1222 - 1239
  • [26] Bresson X., 2012, ADV NEURAL INFORM PR, V25, P1394
  • [27] Fast global minimization of the active Contour/Snake model
    Bresson, Xavier
    Esedoglu, Selim
    Vandergheynst, Pierre
    Thiran, Jean-Philippe
    Osher, Stanley
    [J]. JOURNAL OF MATHEMATICAL IMAGING AND VISION, 2007, 28 (02) : 151 - 167
  • [28] Multi-class Transductive Learning Based on a"" 1 Relaxations of Cheeger Cut and Mumford-Shah-Potts Model
    Bresson, Xavier
    Tai, Xue-Cheng
    Chan, Tony F.
    Szlam, Arthur
    [J]. JOURNAL OF MATHEMATICAL IMAGING AND VISION, 2014, 49 (01) : 191 - 201
  • [29] Completely Convex Formulation of the Chan-Vese Image Segmentation Model
    Brown, Ethan S.
    Chan, Tony F.
    Bresson, Xavier
    [J]. INTERNATIONAL JOURNAL OF COMPUTER VISION, 2012, 98 (01) : 103 - 121
  • [30] Buhler T., 2009, P 26 ANN INT C MACH, P81, DOI DOI 10.1145/1553374.1553385