The use of maximum curvature points for the recognition of partially occluded objects

Min Hong Han, Dongsig Jang

Research output: Contribution to journalArticlepeer-review

83 Citations (Scopus)


A graph-theoretic optimization method is used to recognize partially occluded objects from a 2-D image through the use of maximal cliques and a weight matching algorithm. The vertices of an occluded object image with high curvature values are classified by the objects which are hypothesized to be involved in the occlusion. A heuristic method is also developed to further improve the computational speed. A few typical examples are given to illustrate the accuracy of the optimization model as well as the simplicity of the companion heuristic method.

Original languageEnglish
Pages (from-to)21-33
Number of pages13
JournalPattern Recognition
Issue number1-2
Publication statusPublished - 1990
Externally publishedYes

Bibliographical note

Funding Information:
The problem of identifying partially occluded objects from a vision image has drawn much research attention in the area of computer vision. The occlusion takes place when an object is either overlapped or touched by another object. This problem has significant importance in industrial environment as well as military applications. Suppose parts are flowing on a conveyor for visual inspection. When parts touch or overlap each other, the vision system should be able to recognize each part involved in the occlusion rather than to reject the occluded objects as a single unidentifiable part. A similar situation arises when a robot tries to pick a particular part from a bin in which different part types are jumbled together. In this paper, an object image refers to the image of an object free of any occlusion, whereas the occluded image refers to an image of multiple objects when they are part of an occlusion. Past approaches for identifying the occluded objects from a vision image have relied upon many different means such as Fourier descriptors, statistical pattern matching, symbolic matching, syntactic and relaxation methods, and others. Each of these approaches can be largely classified as either a boundary based method or a local feature based method. The boundary based method uses the boundary information, whereas the local feature based method uses local features such as holes and corners, tS) An extensive research effort has been made for the recognition of partially occluded objects using the boundary based methods. Bhanu and Faugeras t2} * This work was partially supported by the Engineering Excellence Fund of Texas A&M University.

Copyright 2014 Elsevier B.V., All rights reserved.


  • Clique
  • Occlusion
  • Polygonization
  • Weight matching

ASJC Scopus subject areas

  • Software
  • Signal Processing
  • Computer Vision and Pattern Recognition
  • Artificial Intelligence


Dive into the research topics of 'The use of maximum curvature points for the recognition of partially occluded objects'. Together they form a unique fingerprint.

Cite this