Use of spectral information and machine vision for bruise detection on peaches and apricots

E. Garcia-Pardo, Q. Yang, C. R. Bull, Reyer Zwiggelaar

Research output: Contribution to journalArticlepeer-review

33 Citations (SciVal)

Abstract

This paper addresses the problem of detecting bruises on peaches and apricots using machine vision. Bruises were created in a controlled manner on freshly harvested sample fruits. The spectral reflectance characteristics of both bruised and non-bruised surfaces were measured and analysed. From the analyses, the appropriate wavelengths, at which the two types of surface are well separated, were determined. Subsequently spectral filters centred at these selected wavelengths were used for image capture. Image analysis algorithms were developed to detect bruises in images. The most successful methods of detecting bruises were found to be ratio and normalised difference imaging at two wavelengths. The success rate for bruise detection was approximately 65%.
Original languageEnglish
Pages (from-to)323-331
Number of pages9
JournalJournal of Agricultural Engineering Research
Volume63
Issue number4
DOIs
Publication statusPublished - 1996

Fingerprint

Dive into the research topics of 'Use of spectral information and machine vision for bruise detection on peaches and apricots'. Together they form a unique fingerprint.

Cite this