Novel separation method of black walnut meat from shell using invariant features and a supervised self-organizing map

Fenghua Jin, Lei Qin, Lu Jiang, Bin Zhu, Yang Tao

Research output: Contribution to journalArticlepeer-review

Abstract

A method to automatically separate black walnut meat from shells would be highly beneficial to the walnut processing industry. We describe a machine vision system with backlight illumination. Backlit images of walnut meat and shells showed quite different texture patterns due to their different light transmittance properties. This texture pattern was described by the combination of two complimentary texture description operators: local binary pattern and local variance. The resultant feature vectors were fed into a classifier, the supervised self-organizing map (SOM), to determine if the images were meat or shell. Results showed that the proposed approach was very effective in walnut meat and shell separation, with an overall separation accuracy of 98.2%. The high separation accuracy, fast computation speed, and instrument low cost make the proposed imaging system a great potential in walnut processing industry.

Original languageEnglish (US)
Pages (from-to)75-85
Number of pages11
JournalJournal of Food Engineering
Volume88
Issue number1
DOIs
StatePublished - Sep 1 2008

Keywords

  • Black walnuts
  • Classification
  • Imaging
  • Invariant features
  • Machine vision
  • Self-organizing map
  • Texture

ASJC Scopus subject areas

  • Food Science

Fingerprint Dive into the research topics of 'Novel separation method of black walnut meat from shell using invariant features and a supervised self-organizing map'. Together they form a unique fingerprint.

Cite this