Image registration by matching relational structures

J. K. Cheng, T. S. Huang

Research output: Contribution to journalArticlepeer-review


We present a method of image registration where we use relational structures as representations of images and we register two images by finding a match between the two relational structures. It is better than traditional cross-correlation like techniques in several aspects: it is insensitive to scaling, rotation, gray level modification and small relative motion between the two images. The matching process makes use of star structures, which have several advantages over conventional node structures.

Original languageEnglish (US)
Pages (from-to)149-159
Number of pages11
JournalPattern Recognition
Issue number1
StatePublished - 1984


  • Correspondence
  • Image registration
  • Matching
  • Motion estimation
  • Relational structures

ASJC Scopus subject areas

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


Dive into the research topics of 'Image registration by matching relational structures'. Together they form a unique fingerprint.

Cite this