黄桃在线分级时, 表面损伤和可溶性固形物同时在线检测。 损伤和可溶性固形物是评价黄桃品质好坏的重要指标。 采用高光谱成像技术, 尝试对黄桃损伤和可溶性固形物进行同时检测。 利用主成分分析法, 首先对高光谱图像进行主成分分析得到最佳PC(principal component)图像, 其次根据PC图像中各波长对其贡献率的大小确定最佳特征波长(550和720 nm)并结合二值化, 图像掩膜和阈值分割以及相关的图像处理技术对最佳光谱图像进行定性判别。 其准确率最高达到94.6%, 同时建立偏最小二乘定量回归模型对正常样品SSC(soluble solid content)含量进行预测, 通过对模型的不断优化, 实现了基于高光谱成像技术对黄桃碰伤和可溶性固形物同时检测。 可溶性固形物分选准确率为79.2%。 实验结果表明, 利用高光谱成像技术可以实现对黄桃碰伤和可溶性固形物同时检测, 该研究可以为实际在线分选提供理论依据和参考。

The surface damage and soluble solid content were detected simultaneously in online grading of yellow peach, and the damage level and soluble solid content are the important criteria for evaluating the quality of yellow peach. Hyperspectral imaging technology was used to detect the damage level and soluble solid content of yellow peach simultaneously. The principal component analysis was used to obtain the best PC image firstly. Then according to the contribution rate of characteristic wavelength to PC image, the best wavelength of the image (550 and 720 nm) was determined. In the last, the binaryzation, image masking, threshold segmentation and the related image processing technology were combined to qualitatively discriminate the spectral images corresponding to the best characteristic wavelength. Its accuracy was up to 92.9%. At the same time, partial least squares regression model was established to predict the SSC content of normal samples, and by the continuous optimization of the model, online simultaneous detection of yellow peach bruise and soluble solids based on the hyperspectral imaging technology was finally realized. The sorting accuracy of soluble solids was 79.2%. The experimental results show that the yellow peach bruise and SSC can be detected on-line simultaneously by using hyperspectral imaging technology. This research can provide references and basis for the online sorting.