Assessing the impact of low-temperature stress during anthesis stage on winter wheat grain development through...

Extreme weather events, particularly spring low-temperature stress exacerbated by global warming, have become increasingly prevalent in the Huang-Huai-Hai Basin over the past 40 years, a key wheat-producing area in China. This study aims to assess the impact of low-temperature stress during the anthesis stage on the shape and texture parameters of RGB images of both superior and inferior grains of Jimai 22, utilizing computer vision and machine learning techniques.

Results showed that decreasing temperatures and extended durations led to significant reductions in shape parameters and increased uniformity in texture parameters. Superior grains exhibited greater sensitivity to low-temperature stress, with developmental delays postponing parameter extremums from 21 to 28 days after anthesis. A low-temperature stress index (Z) was constructed using principal component analysis and validated with grain weight, achieving coefficients of determination (R²) of over 0.93. Self-organizing maps clustering based on Z and grain weight indicated that increased treatment duration more significantly affected grain parameters than temperature decrease.

This research demonstrates that the combination of RGB imaging with machine learning techniques provides a methodology for assessing the severity of low-temperature stress in winter wheat grains. © 2025 Society of Chemical Industry.