Inertia based filtering of high resolution images using a GPU cluster

Daniel Jungblut*, Gillian Queisser, Gabriel Wittum

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

11 Scopus citations

Abstract

The scheme of inertia based anisotropic diffusion is a very powerful noise reducing and structure preserving image processing operator. This paper presents an implementation of this time consuming filter process on a cluster of Nvidia Tesla high performance computing processors, which can be applied to very large amounts of data in only a few minutes. Applying the inertia based diffusion filter to high resolution image stacks of neuron cells provides fully automatic geometric reconstructions of these images on a scale of <1μm. Such a high throughput and automatic image processing tool has great impact on various research areas, in particular the fast growing field of computational neuroscience, where one encounters increasing amount of microscopy data that needs to be processed.

Original languageEnglish (US)
Pages (from-to)181-186
Number of pages6
JournalComputing and Visualization in Science
Volume14
Issue number4
DOIs
StatePublished - Apr 1 2011

Keywords

  • Anisotropic
  • CUDA
  • Diffusion
  • Filtering
  • GPU
  • High resolution images
  • Inertia based
  • Reconstruction

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Software
  • Modeling and Simulation
  • Engineering(all)
  • Computer Vision and Pattern Recognition
  • Computational Theory and Mathematics

Fingerprint Dive into the research topics of 'Inertia based filtering of high resolution images using a GPU cluster'. Together they form a unique fingerprint.

Cite this