Study on the quantitative measurement of firmness distribution maps at the pixel level inside peach pulp
Study on the quantitative measurement of firmness distribution maps at the pixel level inside peach pulp
复制标题
桃浆内部像素级硬度分布图定量测量研究
DOI：
10.1016/j.compag.2016.09.018
复制
发表时间：
2016-11-15
影响因子：
8.3
通讯作者：
Chen, Kunsong
中科院分区：
文献类型：
--
作者：
Zhu, Nan;Lin, Menghua;Chen, Kunsong
硬度是桃和许多其它水果的主要品质指标。由于桃果肉不同部位的硬度不同，果肉硬度分布图可作为深入了解和评价果实成熟过程、进一步优化种植模式和采后贮藏策略的重要指标。然而，常用的Magness-Taylor硬度测试仪不能提供像素级的硬度分布，因为它只测量探头所在的一个或几个采样点。在这项研究中，一个基于实验室的高光谱成像系统在380-1030 nm的波长范围内的发展，以检测和可视化的硬度分布的不同横截面的桃果肉。采用偏最小二乘回归（PLSR）和最小二乘支持向量机（LS-SVM）两种校正方法建立了硬度校正模型。此外，分别采用连续投影法、无信息变量剔除法和竞争自适应加权采样法（汽车）选择最佳波长，以提高模型的准确性和鲁棒性。未经预处理的CARS-PLSR被确定为测定桃果肉硬度的最佳模型，得到的预测结果的相关系数为0.852，残差预测偏差为1.739。利用该模型，在像素水平上对桃果肉不同截面的硬度分布进行了定量可视化。硬度直方图显示桃果肉内部存在广泛的硬度。总体结果表明，高光谱成像显示了巨大的潜力，可视化的空间变化的内部桃果肉硬度分布在像素水平上，这可以提供一个更详细的了解成熟过程中的桃果实在采前和采后期间。(C)2016爱思唯尔B. V.保留所有权利。
Firmness is a major quality index for peaches and many other fruits. Because firmness varies among different parts of the pulp, maps of the firmness distribution inside peach pulp can be used as an important indicator for deeply understanding and evaluating the ripening process of fruit and for further optimizing planting patterns and postharvest storage strategies. However, the commonly used Magness-Taylor firmness tester cannot provide the firmness distribution at the pixel level, because it measures only one or a few sampling points where the probe is located. In this study, a laboratory-based hyperspectral imaging system in the wavelength range of 380-1030 nm was developed to detect and visualize the firmness distributions for different cross sections of peach pulp. Two calibration methods based on partial least squares regression (PLSR) and least squares support vector machines were applied to establish firmness calibration models. In addition, the successive projections algorithm, uninformative variable elimination, and competitive adaptive reweighted sampling (CARS) were applied separately to select the optimal wavelengths to improve the models' accuracy and robustness. CARS-PLSR without preprocessing was determined to be the best model for determining the firmness of peach pulp, yielding prediction results with a correlation coefficient of 0.852 and a residual predictive deviation of 1.739. Using this model, the firmness distributions for different cross sections of peach pulp were quantitatively visualized at the pixel level. The firmness histogram revealed the existence of a wide range of firmnesses inside peach pulp. The overall results demonstrated that hyperspectral imaging shows great potential for visualizing the spatial variations in the firmness distribution inside peach pulp at the pixel level, which can provide a more detailed understanding of the ripening process of peach fruits during the preharvest and postharvest periods. (C) 2016 Elsevier B.V. All rights reserved.