高光谱成像
化学
数学
分析化学(期刊)
色谱法
人工智能
计算机科学
作者
Yuanzhe Chen,Qiaohua Wang,Fan Wang,Bin Xu
标识
DOI:10.1016/j.fbio.2023.102605
摘要
The gel springiness of preserved egg during pickling is an important indicator of the maturity and gel quality of preserved eggs (PEs), for which there are no validated rapid means of detection. This work, hyperspectral imaging was used to non-destructively detect the gel springiness of preserved eggs (PEs) at different maturity levels during the curing period. Firstly, principal component analysis was used to explain the characteristics of the changes in gel springiness during the curing process of PEs of different maturity levels. Secondly, based on two-dimensional correlation spectroscopy(2DCOS), the 427–843 nm band range was explored as the optimal study region for gel springiness when gel springiness was used as a perturbation factor. In turn, spectral factors representative of the change in gel springiness are stripped in this band region. Comparing three different variable selection methods (SPA、CARS and UVE), it was found that the UVE-PLSR model had the highest detection accuracy: The predicted coefficient of determination Rp2 and the predicted standard deviation SEp of the predicted gel springiness are 0.846 and 1.526, with a residual prediction deviation RPD of 2.35. Finally, the above optimal prediction model is applied to the pixel spectrum to calculate the springiness value of each pixel point on the hyperspectral image and is supplemented by a pseudo-color technique to visualize the prediction of the spatial distribution of the gel springiness value. The results show that using hyperspectral imaging can detect the gel springiness value of the PE during the curing process, which will provide a theoretical basis for future determination of the ripeness of the PE as well as high-throughput online sorting.
科研通智能强力驱动
Strongly Powered by AbleSci AI