Skip to main navigation Skip to search Skip to main content

A Blind Comparison Between Results of Four Image Analysis Systems using a Photo-Library of Piles of Sieved Fragments

  • John-Paul Latham
  • , John Kemeny
  • , Norbert H. Maerz
  • , Michael Noy
  • , Jacques Schleifer
  • , Simon Tose

Research output: Contribution to journalArticlepeer-review

Abstract

Four image analysis systems for measuring rock fragmentation: FragScan, PowerSieve ® , Split and WipFrag, have been compared under conditions necessary to provide an objective though limited assessment of their capabilities. The analysis of results is based on a sample of ten photographs taken from a series of photographs of controlled artificial muckpiles. These were created from dumping a blended mixture of sieved samples of limestone aggregate, in order to create a range of near perfect Rosin-Rammler sieve size distributions. Results from the various systems are compared with sieved results using both histogram and cumulative forms, with and without fines corrections in the case of Split and Wipfrag. Statistical indicators are evaluated to examine the match between system prediction values and sieving values. Commentaries on the results by the inventors of each system have been incorporated. All four systems were found to perform both well in some cases and poorly in others. From a detailed examination of the results, some insight into the strengths and weaknesses of the various systems is presented.

Original languageAmerican English
JournalFragblast
Volume7
DOIs
StatePublished - Jun 1 2003

Keywords

  • Fragmentation
  • Image analysis
  • Limestone
  • Rock fragmentation
  • Rocks
  • Sampling bias
  • Size distribution
  • Statistical methods
  • comparative study
  • particle size
  • photograph

Disciplines

  • Geology
  • Petroleum Engineering

Fingerprint

Dive into the research topics of 'A Blind Comparison Between Results of Four Image Analysis Systems using a Photo-Library of Piles of Sieved Fragments'. Together they form a unique fingerprint.

Cite this