Coupling orientation diffusion with coherence-enhancing diffusion: a fingerprint case

Mohammad A. U. Khan, Tariq M. Khan*, Donald G. Bailey

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

1 Citation (Scopus)


The enhancement of coherent flow-like structures is desired for many image processing tasks, such as segmentation and feature detection. This task can be accomplished in a natural way by adopting anisotropic diffusion filtering using a diffusion matrix adapted to the local structure. This method is referred to as coherence-enhancing diffusion (CED). The performance of CED can be analyzed by observing the evolution of the orientation field (OF) associated with an evolving diffusion matrix. It was revealed from a series of experiments that the final OF from a CED-enhanced image sometimes strays from its true underlying OF (marked by a human expert), degrading its performance. In this paper, a strategy is proposed which repeatedly cleans the OF associated with a diffusion matrix. Thus, for every iteration of CED, its OF is diffused separately until it converges and is then fed back to the CED process to move forward. This hypothesis is tested with the motive of getting an enhanced CED performance. The proposed scheme is validated using fingerprint data, and their numerical results are displayed.

Original languageEnglish
Pages (from-to)513-521
Number of pages9
JournalSignal, Image and Video Processing
Issue number3
Publication statusPublished - Mar 2018


  • Orientation diffusion
  • Coherence-enhancing diffusion
  • Fingerprint image enhancement
  • Orientation field estimation


Dive into the research topics of 'Coupling orientation diffusion with coherence-enhancing diffusion: a fingerprint case'. Together they form a unique fingerprint.

Cite this