Seed morphological traits were used to identify 67 Italian bean (Phaseolus vulgaris L.) accessions, belonging to 58 Italian landraces. An overall of 138 size, shape and texture descriptors were measured, on each seed, using image analysis techniques. The achieved data, analysed applying the stepwise Linear Discriminant Analysis, allowed to discriminate among bean landraces, also identifying the harvest year and the cropping areas. Comparative analyses were carried out to verify the possibility to distinguish seeds belonging to the same landrace but grown applying different agricultural practices. Preliminarily, it was possible to discriminate three main color categories of bean seeds, with an overall performance of 99.1%. Moreover, for each of these three categories, the belonging bean landraces were identified, with overall correct identification percentages included between 94.3% and 99.7%. Following the same procedure, it was possible to assess the possibility to identify the bean landraces origin, reaching overall correct identification percentage higher than 88%. Also considering the effect of the cropping year, the cultivation region and the agricultural practices, high identification performances were recorded. The results support the application of the computer vision system not only for the identification, classification or grading purpose, but also to define the product traceability, in order to get a “market card” for landrace beans

Characterisation of Italian bean landraces (Phaseolus vulgaris L.) using seed image analysis and texture descriptors

GRILLO, OSCAR;SARIGU, MARCO;
2015

Abstract

Seed morphological traits were used to identify 67 Italian bean (Phaseolus vulgaris L.) accessions, belonging to 58 Italian landraces. An overall of 138 size, shape and texture descriptors were measured, on each seed, using image analysis techniques. The achieved data, analysed applying the stepwise Linear Discriminant Analysis, allowed to discriminate among bean landraces, also identifying the harvest year and the cropping areas. Comparative analyses were carried out to verify the possibility to distinguish seeds belonging to the same landrace but grown applying different agricultural practices. Preliminarily, it was possible to discriminate three main color categories of bean seeds, with an overall performance of 99.1%. Moreover, for each of these three categories, the belonging bean landraces were identified, with overall correct identification percentages included between 94.3% and 99.7%. Following the same procedure, it was possible to assess the possibility to identify the bean landraces origin, reaching overall correct identification percentage higher than 88%. Also considering the effect of the cropping year, the cultivation region and the agricultural practices, high identification performances were recorded. The results support the application of the computer vision system not only for the identification, classification or grading purpose, but also to define the product traceability, in order to get a “market card” for landrace beans
Computer vision; EFDs; Haralick’s features; LDA; Seed morphology; Traceability
File in questo prodotto:
File Dimensione Formato  
Lo Bianco et al. 2015 Fagioli.pdf

Solo gestori archivio

Tipologia: versione post-print
Dimensione 1.33 MB
Formato Adobe PDF
1.33 MB Adobe PDF   Visualizza/Apri   Richiedi una copia

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11584/174402
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 22
  • ???jsp.display-item.citation.isi??? ND
social impact