An image processing technique for high-speed measurement of particle-size distributions

K. Deguchi

Research output: Contribution to journalArticlepeer-review

3 Citations (Scopus)

Abstract

This paper describes an image processing technique for the high-speed measurement of particle-size distributions. When an image of many small particles distributed on a plane is given, it is not difficult to determine the size of each particle by measuring the area of the particle region. However, since we have usually a large number of particles on an image, this conventional technique requires considerable processing time. When real-time processing is essential, a new high-speed processing technique must be developed. The technique proposed in this paper is one based on sampling. We use a number of parallel lines to obtain a distribution of 'run-length'. By employing a TV camera for image pick-up, we can use its scanning lines as the sampling line. When a scanning line moves across a particle region, a run-length is obtained. After the scanning of one image frame, we obtain a distribution of run-lengths for the image frame. Next, the run-length distribution is converted to the particle size distribution. If each particle region is a circle, this conversion can be done by solving a simple linear equation. This calculation can be performed in real-time processing even with a microcomputer. Theoretically, this conversion becomes unstable if we select the size resolution too small. Then some considerations for stable and high-speed calculation have been discussed for practical applications.

Original languageEnglish
Pages (from-to)128-133
Number of pages6
JournalMeasurement
Volume4
Issue number4
DOIs
Publication statusPublished - 1986 Jan 1

Keywords

  • Image processing
  • particle-size distribution
  • stereology

ASJC Scopus subject areas

  • Instrumentation
  • Electrical and Electronic Engineering

Fingerprint Dive into the research topics of 'An image processing technique for high-speed measurement of particle-size distributions'. Together they form a unique fingerprint.

Cite this