Facultad de Ingeniería, Universidad de la Sabana, Campus Universitario del Puente del Común, Chía, Cundinamarca, Colombia.
J Sci Food Agric. 2012 Oct;92(13):2709-14. doi: 10.1002/jsfa.5693. Epub 2012 Jun 1.
Digital image analysis has an important role in geographical provenance of grains, as it can provide parameters of size, shape and color, which are important quality parameters for the design of engineering processes such as drying and milling of grains. In this study, digital image analysis was used to classify nine rice cultivars based on different morphometric parameters using the three sides of the grain (lateral, ventral and axial), Feret diameter, and 10 different form factors and color parameters (CIE L*, a* and b*).
Result of principal component analyisis was an equation with seven variables (area, perimeter, length, width, thickness, sphericity and color), which was useful for distinguishing between nine different cultivars. The morphometric and color parameters for the Mor A-98 and Mor A-92 varieties showed they had 88% similarity. The variability was expressed with a confidence of 95%.
Multivariate analysis indicated that the lateral side is the most sensitive for the classification of Mexican rice grains because of its color and morphometric characteristics. These results showed the application of image analysis for the future classifications of grains.
数字图像分析在谷物的地理起源研究中具有重要作用，因为它可以提供大小、形状和颜色等参数，这些参数是谷物干燥和碾磨等工程过程设计的重要质量参数。在这项研究中，使用数字图像分析根据谷物的三个侧面（侧面、腹面和轴向）、Feret 直径以及 10 个不同的形状因子和颜色参数（CIE L*、a和 b）对 9 个水稻品种进行分类。
主成分分析的结果是一个包含七个变量（面积、周长、长度、宽度、厚度、球形度和颜色）的方程，该方程有助于区分 9 个不同的品种。Mor A-98 和 Mor A-92 品种的形态和颜色参数显示它们具有 88%的相似度。变异性的置信度为 95%。
多元分析表明，由于墨西哥稻谷的颜色和形态特征，侧面是分类的最敏感部位。这些结果表明图像分析可用于未来的谷物分类。
