DisplayTitle

For millennia grain has been analysed for quality and foreign matter by visual means. While well accepted, these methods, used primarily for the determination of levels of defective grains and foreign matter, are highly subjective and are susceptible to human factors and even potentially manipulation as they are by their nature is neither accurate nor precise. There is strong demand for means of objective analysis to improve accuracy but also, arguably more importantly, to address the repeatability and reproducibility of the analysis. One such approach is the use of Image however, if we consider the objective basis for the calibration process for techniques such as NIR spectroscopy we see the challenges and the paradoxes inherent in the calibration and validation of image analysis instruments. These paradoxes largely stem from the subjectivity of the reference method which remains a human analysis, albeit with a highly experienced operator. This paper will explore advances in image analysis for objective analysis of grain defects and foreign matter and will explore the challenges of the calibration and validation process. It will present results of trails conducted in Australia, Europe and North America which demonstrate the potential of this technique. For example, analysis of barley samples contaminated with wheat resulted in correlations better than 0.99 when comparing a validation set analysed by Image Analysis against a trained analyst. Defects such as skinning in Barley achieved correlations greater than 0.81. Useful levels of repeatability were also found in these studies with repeatability (Pooled SD) being less than half the level of accuracy (measures as RMSEP). Finding of these studies demonstrate that this analytical technique can provide a useful alternative to current methods.