Submitted to: Journal of Applied Spectroscopy
Publication Type: Peer reviewed journal
Publication Acceptance Date: 12/1/2001
Publication Date: 12/1/2001
Citation: Pasikatan, M.C., and Dowell, F.E. 2001. Sorting systems based on optical methods for detecting and removing seeds infested internally by insects or fungi: A review. 36(4):399-416. Interpretive Summary: Insect larvae or pupae inside seeds, or fungi on the surface or in seeds, can pose problems in grain storage, seed quality, and food safety. Timely detection and removal of contaminated seeds from seed lots is essential to control their potential damage. Among available methods, optical methods have the potential to rapidly detect and physically remove contaminated seeds. Therefore, studies related to seed sorting based on optical methods, for these contaminants, were reviewed. Darkening of seed color caused by fungi is generally detectable using optical filters in the visible range of light, and internal seed attributes are detectable in the near-infrared (NIR) range. Models for classifying seeds with these contaminants have succeeded using regions in the visible-NIR range of light or a pair of wavelengths in that region. It is important to identify wavelengths for detection and develop a sorting criterion from these. Indicators of seed fungal contamination are color chitin, ergosterol, or changes in triglycerides. Indicators of internal insects are chitin, protein, phenolic compounds, or changes in starch content.
Technical Abstract: Sorting systems based on optical methods have the potential to rapidly detect and physically remove seeds severely contaminated by fungi, or infested internally by insect larvae or pupae. Thus, the literature on sorting systems based on optical methods for detecting and sorting seeds with these attributes was reviewed. Sorting indices based on wavelengths useful for detecting these attributes were emphasized. Surface characteristics of seeds, like discoloration caused by fungi, are generally detectable in the visible range of the electromagnetic spectrum, whereas internal attributes are detectable in the near-infrared range. The spectral differences between sound and infested seeds are usually subtle, but full-spectrum and two-wavelength classification models have succeeded in detecting and classifying seeds based on these attributes. For high sorting accuracies, wavelength identification and proper selection of a sorting criterion are important. Chitin, ergosterol, or hydrolysis of triglycerides have been identified as indicators of seed fungal contamination whereas chitin, protein, phenolic compounds, or changes in starch have been useful indicators of internal insects in seeds.