Integrating optical imaging techniques for a novel approach to evaluate Siberian wild rye seed maturity

Zhicheng Jia, Chengming Ou, Shoujiang Sun, Juan Wang, Jingyu Liu, Ming Sun, Wen Ma, Manli Li, Shangang Jia, Peisheng MaoFrontiers in Plant Science 2023 · DOI

Abstract

Advances in optical imaging technology using rapid and non-destructive methods have led to improvements in the efficiency of seed quality detection. Accurately timing the harvest is crucial for maximizing the yield of higher-quality Siberian wild rye seeds by minimizing excessive shattering during harvesting. This research applied integrated optical imaging techniques and machine learning algorithms to develop different models for classifying Siberian wild rye seeds based on different maturity stages and grain positions. The multi-source fusion of morphological, multispectral, and autofluorescence data provided more comprehensive information but also increases the performance requirements of the equipment. Therefore, we employed three filtering algorithms, namely minimal joint mutual information maximization (JMIM), information gain, and Gini impurity, and set up two control methods (feature union and no-filtering) to assess the impact of retaining only 20% of the features on the model performance. Both JMIM and information gain revealed autofluorescence and morphological features (CIELab A, CIELab B, hue and saturation), with these two filtering algorithms showing shorter run times. Furthermore, a strong correlation was observed between shoot length and morphological and autofluorescence spectral features. Machine learning models based on linear discriminant analysis (LDA), random forests (RF) and support vector machines (SVM) showed high performance (>0.78 accuracies) in classifying seeds at different maturity stages. Furthermore, it was found that there was considerable variation in the different grain positions at the maturity stage, and the K-means approach was used to improve the model performance by 5.8%-9.24%. In conclusion, our study demonstrated that feature filtering algorithms combined with machine learning algorithms offer high performance and low cost in identifying seed maturity stages and that the application of k-means techniques for inconsistent maturity improves classification accuracy. Therefore, this technique could be employed classification of seed maturity and superior physiological quality for Siberian wild rye seeds.

Introduction

Siberian wild rye (Elymus sibiricus L.) is a critical species for grassland restoration and livestock fodder in cold regions. However, its cultivation is hampered by a significant "shattering" problem: seeds fall off the stalk easily once mature. Harvesting too early results in low-quality, immature seeds, while delaying harvest by even a few days can lead to losses of up to 80% of the yield. Traditional maturity assessment relies on subjective visual cues or destructive laboratory tests, creating a need for a rapid, objective, and non-destructive monitoring method to identify the optimal "Goldilocks" window for harvest.

Methodology

The researchers developed a multi-modal imaging pipeline using the VideometerLab4 system to capture "optical fingerprints" of seeds, moving beyond simple RGB photography.

Data Acquisition and Processing

The study utilized a dataset of 600 seeds categorized by three maturity stages—Milk-ripe (MRS), Dough stage (DS), and Full-ripe (FRS)—and two grain positions—Superior grains (SG) and Inferior grains (IG). Three types of data were integrated:

  • Morphological: Shape and size characteristics.
  • Multispectral: Reflectance across 19 wavelengths from UV to NIR.
  • Autofluorescence: Light emission under specific excitation, serving as a proxy for internal chemical changes like chlorophyll degradation.
Workflow of the integrated optical imaging and machine learning approach for seed maturity classification.

Feature Selection and Modeling

To optimize for low-cost hardware, three filtering algorithms (JMIM, Information Gain, and Gini Impurity) were tested to reduce the 42 initial features to the top 20%. Classification was performed using Linear Discriminant Analysis (LDA), Random Forests (RF), and Support Vector Machines (SVM). Additionally, K-means clustering was employed to reclassify seeds based on physiological state, addressing the "Position Paradox" where grains at different positions on the spikelet mature at different rates.

Key Insights and Results

The study demonstrated that autofluorescence and specific color coordinates (CIELab A and B) are highly predictive of maturity, capturing chemical changes invisible to the naked eye.

  • Model Performance: The LDA model was the standout performer, achieving an accuracy of 0.94 and an AUC of 0.97 with the full feature set.
  • The K-means Update: Reclassifying seeds using K-means clustering significantly boosted model performance, increasing accuracy by 5.8% to 9.24%.
  • Efficiency: Feature reduction via JMIM allowed for processing times as low as 0.01 seconds, making the system suitable for high-speed sorting machinery.
  • Biological Correlation: Optical features showed a strong correlation with shoot length, indicating that these non-destructive scans are reliable predictors of seed vitality and growth potential.

Conclusion

This research establishes a robust framework for non-destructive seed maturity assessment in Siberian wild rye. By combining multi-source optical fusion with efficient machine learning and clustering techniques, the study provides a path toward automated, real-time harvest monitoring. Future work will focus on validating these models across diverse environmental conditions and developing simplified sensors targeting the specific wavelengths identified here.

Summary updated

Abstract

Advances in optical imaging technology using rapid and non-destructive methods have led to improvements in the efficiency of seed quality detection. Accurately timing the harvest is crucial for maximizing the yield of higher-quality Siberian wild rye seeds by minimizing excessive shattering during harvesting. This research applied integrated optical imaging techniques and machine learning algorithms to develop different models for classifying Siberian wild rye seeds based on different maturity stages and grain positions. The multi-source fusion of morphological, multispectral, and autofluorescence data provided more comprehensive information but also increases the performance requirements of the equipment. Therefore, we employed three filtering algorithms, namely minimal joint mutual information maximization (JMIM), information gain, and Gini impurity, and set up two control methods (feature union and no-filtering) to assess the impact of retaining only 20% of the features on the model performance. Both JMIM and information gain revealed autofluorescence and morphological features (CIELab A, CIELab B, hue and saturation), with these two filtering algorithms showing shorter run times. Furthermore, a strong correlation was observed between shoot length and morphological and autofluorescence spectral features. Machine learning models based on linear discriminant analysis (LDA), random forests (RF) and support vector machines (SVM) showed high performance (>0.78 accuracies) in classifying seeds at different maturity stages. Furthermore, it was found that there was considerable variation in the different grain positions at the maturity stage, and the K-means approach was used to improve the model performance by 5.8%-9.24%. In conclusion, our study demonstrated that feature filtering algorithms combined with machine learning algorithms offer high performance and low cost in identifying seed maturity stages and that the application of k-means techniques for inconsistent maturity improves classification accuracy. Therefore, this technique could be employed classification of seed maturity and superior physiological quality for Siberian wild rye seeds.
BibTeX
@inproceedings{jia2023integrating,
  author = {Zhicheng Jia and Chengming Ou and Shoujiang Sun and Juan Wang and Jingyu Liu and Ming Sun and Wen Ma and Manli Li and Shangang Jia and Peisheng Mao},
  title = {{Integrating optical imaging techniques for a novel approach to evaluate Siberian wild rye seed maturity}},
  booktitle = {Frontiers in Plant Science},
  year = {2023},
  doi = {10.3389/fpls.2023.1170947},
}
Figure