Hyperspectral imaging for nondestructive determination of internal qualities for oil palm (Elaeis guineensis Jacq. var. tenera).
Hyperspectral imaging for nondestructive determination of internal qualities for oil palm (Elaeis guineensis Jacq. var. tenera).
复制标题
DOI：
10.3173/air.18.130
复制
发表时间：
2009-01-01
期刊：
影响因子：
--
通讯作者：
Banshaw Bahalayodhin
中科院分区：
文献类型：
--
作者：
Junkwon, P.;Takigawa, T.;Banshaw Bahalayodhin
本研究的目的是建立一种测定油棕内在品质的方法。变种tenera）。本研究使用了属于4种成熟度（过熟、成熟、欠熟和未成熟）的束和果实。对这些串进行了成熟度、含油量和游离脂肪酸含量三项内在品质的测定。由于内部质量的估计的基础上的整体数据的一束是困难的，我们专注于平均反射率和平均相对反射率值的水果，没有隐藏在束叶。通过我们的方法，有必要在确定油含量和游离脂肪酸含量之前估计串的成熟度。为了对一串葡萄的成熟度进行分类，利用不同成熟度等级的葡萄的平均相对反射率值，并基于欧氏距离进行分类。此外，叶绿素与类胡萝卜素的比值（Rp）也被用来估计一串的成熟度。根据成熟度等级建立相应的校正模型，预测油分和游离脂肪酸的含量。这两种方法在所有成熟度等级中均获得了正确的估计结果。在油含量和游离脂肪酸模型的验证中，决定系数（R2）分别为99.7%和99.5%，标准预测误差（SEP）分别为0.421和0.190。对于油棕果实，开发了估计果实成熟度的方法。将使用果实下部的平均相对反射率值的成熟度估计与使用非浅绿黄色区域、非黄色区域和非红橙子区域与果实整个区域的比率的成熟度估计进行比较。用果实下部平均相对反射率对各成熟度等级进行了正确的估测，用果实面积比对各成熟度等级的估测正确率为97.92%。由于利用果实的面积比进行成熟度估计可以自动完成，因此可以为工厂中的果实成熟度评估提供更实际的应用。
The goal of this study is to develop an approach to determine the internal qualities in oil palm (Elaeis guineensis Jacq. var. tenera). Bunches and fruits belonging to 4 classes of ripeness (overripe, ripe, underripe and unripe) were used for this study. For these bunches, three of internal qualities as ripeness, oil content and free fatty acid content were examined. Since the estimation of internal qualities based on the overall data for a bunch was difficult, we focused on the average reflectance and the average relative reflectance values of fruits that were not concealed by fronds in bunch. By our approach, it was necessary to estimate the ripeness of the bunch before the oil content and free fatty acid content were determined. To classify ripeness of a bunch, the average relative reflectance values of bunches in different classes of ripeness were used and classified based on Euclidean distance. In addition, ratio of chlorophyll to carotenoids (Rp) was also used for estimating ripeness of a bunch. Then oil content (OC) and free fatty acid (FFA) content were predicted by calibration models corresponding to the class of ripeness. Correct estimation results in all classes of ripeness were obtained by both methods. The coefficients of determination (R2) were 99.7% and 99.5% with a standard error of prediction (SEP) of 0.421 and 0.190 in the validation of oil content and free fatty acid models, respectively. For oil palm fruits, methods to estimate the ripeness of the fruits were developed. Ripeness estimation using the average relative reflectance values in lower part of the fruit was compared with ripeness estimation using the ratio of a not-pale greenish yellow area, a not-yellow area and a not-reddish orange area to the entire area of fruit. The correct estimation in all classes of ripeness was obtained by using the average relative reflectance at lower part of fruit while a correct ripeness estimation rate of 97.92% was gained by using ratio of area in fruit. Since the ripeness estimation using the ratio of the area of the fruits can be done automatically, it may provide more practically applicable for the assessment of fruit ripeness in the factory.