Conclusions
-
Both (700-1050nm) and (900-1700nm) NIR spectroscopic techniques showed good potential
-
Neuralnetwork-based classifier showed higher classification accuracies than statistical classifiers
-
Spectral information in 900-1700 nm with NN classifier showed highest accuracies (100%) for classifying good, moldy, and rancid kernels as compared to these in 700-1050 nm
Other contributions
-
Establish fatty acid profiles
-
Oil content prediction
Conclusions
-
Both (700-1050nm) and (900-1700nm) NIR spectroscopic techniques showed good potential
-
Neuralnetwork-based classifier showed higher classification accuracies than statistical classifiers
-
Spectral information in 900-1700 nm with NN classifier showed highest accuracies (100%) for classifying good, moldy, and rancid kernels as compared to these in 700-1050 nm
Other contributions
-
Establish fatty acid profiles
-
Oil content prediction
Spectroscopic Techniques for Discriminating Confectionary Sunflower
Needs
- Portable
- Instrument-sample
- Fast & Accurate
- Cost-effective
Parameters
- Good (edible)
- Moldy
- Rancid
Objectives
-
Develope a portable sensing system for quality evaluation of sunflower kernels
-
Evaluate the capability of the sensing system for discriminating good, moldy, and rancid kernels
Methodology
Sensing System Development
- Hardware
- NIR technology
- Portable spectroscopy
- 700-1050nm, 900-1700nm
- Material handling system
- Signal analysis and prediction software
Samples and Spectral Signals
- Samples
- 40 good sunflower confectionary
- 40 moldy
- 40 rancid
- Spectral signals
- 700-1050nm, PC based Ocean Optics
- 900-1700nm, PC based CDI
