Determination of Colour and Kinetic Parameter Differences Between Aflatoxin Contaminated and Uncontaminated Pistachio Nuts Using Machine Vision


Creative Commons License

Özlüoymak Ö. B. , Güzel E.

Journal of Tekirdağ Agricultural Faculty, cilt.18, sa.1, ss.157-168, 2021 (Hakemli Üniversite Dergisi)

  • Cilt numarası: 18 Konu: 1
  • Basım Tarihi: 2021
  • Dergi Adı: Journal of Tekirdağ Agricultural Faculty
  • Sayfa Sayıları: ss.157-168

Özet

Aflatoxins produced by Aspergillus species have a great important in the food industry, especially in dried nuts and fruits. Agricultural products are prone to the aflatoxins during the stages like harvesting, drying and storage. Rapid identification of aflatoxin contaminated products is of great interest to the food industry. The food companies start using screening technologies instead of human labour to become more profitable and accurate. Moreover, economical losses and diseases resulting from aflatoxin contamination are a significant problem. The objective of this study was to develop an image processing based aflatoxin contaminated in-shell pistachio nut identification system in order to separate aflatoxin contaminated pistachio nuts from the healthies one. Bright greenish yellow fluorescence (BGYF), which indicates possible aflatoxin contamination, was investigated as a discriminating factor for identification of contaminated pistachio nuts. A total of 100 pistachio nut samples (50 BGYF+ and 50 BGYF-) were evaluated. In the study, imaging algorithms were developed in order to classify the pistachio nut samples as BGYF+ and BGYF-. The colour (L*, a* and b*) and kinetic (chroma, hue angle and browning index) parameters of each pistachio nut sample were analysed and differences between them were determined statistically. Colour and kinetic parameters were also grouped and associated each other by using factor analysis method to simplify the image processing algorithm. Statistically significant differences were found for all colour and kinetic parameters between two groups. According to the factor analysis results; chroma, a* and browning index values were substantially loaded on Factor 1, while hue angle and b* were substantially loaded on Factor 2. The remaining variable L* was substantially loaded on Factor 3. In future studies, an optimized (more effective and convenient) image processing algorithm for developing a new real-time determination and separation system will be enhanced based on the statistical analysis results. The results obtained from this study will form a basis for further investigations.