A research team led by Prof. WANG Rujing and WANG Liusan from the Hefei Institutes of Physical Science of the Chinese Academy of Sciences has developed a method to detect internal cracks in rice seeds using near-infrared spectroscopy.

"Cracks can affect the germination rate of seeds," said WANG Liusan, "our study can help select high-quality seeds with a non-destructive method."

The research results were published in Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy.

In agricultural production, rice seed quality is directly related to crop yield and quality. However, cracks in rice seeds are often difficult to detect with the naked eye, which poses a challenge for evaluating the quality of rice seeds.

In this study, the researchers developed the non-destructive method to detect internal cracks in rice seeds by using four machine learning classification algorithms-partial least squares discrimination, support vector machine, k-nearest neighbor, and random forest- combined with four spectral preprocessing methods: standard normal variate, scattering correction, and the first and second derivatives, to build their models.

They tested different models to find the best one for detecting internal cracks in rice seeds and found that using partial least squares discrimination with the original spectral data was the most effective approach. While the support vector machine model didn't perform as well, it still outperformed the random forest and k-nearest neighbor models. The analysis showed that the key factor for crack detection is related to the amylose content in the seeds.

Overall, this study shows that the use of near-infrared spectroscopy combined with partial least squares discriminant analysis is feasible for non-destructive detection of internal cracks in rice seeds.

The application of this technology not only improves the efficiency and accuracy of rice seed quality assessment, but also provides new technological means for seed quality control in agricultural production.

A seed with internal cracks as detected by near-infrared spectroscopy. (Image by WANG Liusan)