Quantitative analysis of starch and amylose in rice using near-infrared hyperspectroscopy and data extraction algorithms combined with GOA-SVR

Abstract

The rice starch and amylose content are important indicators as values of rice nutrition and economy. The paper aimed at analyzing the feasibility of near-infrared hyperspectroscopy as well as data extraction algorithms combined with chemometrics to quantify starch and amylose in rice. Simultaneously, a model based on grasshopper optimization algorithm-support vector regression (GOA-SVR) was suggested for the detection of starch as well as amylose content in rice. Three modeling algorithms (partial least squares regression (PLSR), extreme learning machine (ELM), as well as GOA-SVR) were combined with the hyperspectral data of experimental samples from the correction set to develop the near-infrared hyperspectral-based models to detect rice starch as well as amylose. The experimental results revealed the near-infrared hyperspectroscopy combined with GOA-SVR and the multi-dimensional scaling data extraction algorithm could relatively better detect the rice starch as well as amylose content compared to the PLSR and ELM modeling algorithms. Compared with previously published near-infrared hyperspectral studies, the results of this study are relatively accurate and rapid.

Plant phenotyping relevance

米のデンプンおよびアミロース含量という植物器官の形質を、近赤外ハイパースペクトル計測とデータ抽出・回帰モデルで定量する方法を開発・比較しており、表現型取得が中心である。

Code and data availability

公開状態または取得可能な本文経路を確認できませんでした。

No evidence-backed public reproduction asset is currently recorded.