Efficient discrimination of seed viability based on machine vision color indices: screening for reliable phenotypic indicators

文献类型： 外文期刊

第一作者： Mao, Yilin

作者： Mao, Yilin;Wu, Weifeng;Li, He;Xu, Yanan;Dong, Xuehui;Sun, Qun;Tu, Keling;Zhang, Han

作者机构：

关键词： Machine vision; Hyperspectral imaging; Deep learning; Viability discrimination; Perilla seeds

期刊名称：MICROCHEMICAL JOURNAL （ 影响因子：5.1； 五年影响因子：4.7 ）

ISSN： 0026-265X

年卷期： 2025 年 217 卷

页码：

收录情况： SCI

摘要： Viability, a key indicator for assessing Perilla seeds quality, directly affects its germination performance and field production potential. To address the destructive and inefficient traditional detection methods, machine vision and hyperspectral imaging technology were employed to investigate phenotypic indicators significantly correlated with Perilla seed viability. A deep learning architecture, named PSV-CBLD, combining convolutional neural network and bi-directional long and short-term memory, was utilized to establish a viability discrimination model for Perilla seeds. The findings are as follows: (1) Color indices (R, a*, b, H, S, V), the reflectance at the end of the visible region (700.48-773.00 nm) and the short-wave near-infrared region (773.80-999.94 nm) were significantly correlated with seed germination rate (|r| >= 0.90). (2) As perilla seeds matured, the values of R, a*, b, S, and V gradually decreased, while the H value increased, and the hyperspectral reflectance declined. (3) The PSV-CBLD model, utilizing the color indices set (RabHSV), achieved optimal performance in viability discrimination, with an accuracy of 87.87 %. In this study, an efficient discrimination model for Perilla seed viability was established relying on several reliable phenotypic indicators. This provides a novel perspective for rapidly screening high-viability seeds in agricultural production.

分类号：

  • 相关文献
作者其他论文 更多>>

微信小程序