Evaluation and Development Trends of Optical Detection Technology for Seed Vigor

High-vigor seeds could show superiority in field production because seed vitality is essentially a comprehensive concept composed of germination rate, germination potential, and vitality index. Seed vitality was a traditional method of measurement, albeit a time-consuming and laborious one accompanied by errors caused by human factors. Optical technology had developed rapidly in the agricultural field, making it possible to detect seeds with high precision, high frequency, and high efficiency. In this article, the basic principles, advantages, and limitations of different optical techniques for obtaining seed vigor estimates are introduced and reviewed, and the key technology of non-destructive optical detection of single seeds will be discussed. In addition, the current situation of optical detection nondestructive technology in single seed detection will be discussed and analyzed, and the three directions of optical principle (seed characteristic spectrum database, intelligent sorting, and grading equipment) will be explored, providing a reference for the research methods using optical technology applied to seed detection.

Seed has been the basis of agricultural production, a national strategic and fundamental core industry, and an important guarantee for achieving food security (1). Increasing yield was the key to producing sufficient food on limited land area to ensure sufficient food production, and screening high-vigor seeds was an important guarantee for increasing yield (2). High-vigor seeds germinate quickly, are very healthy, produce many seedlings, resist harsh conditions, grow more tillers, and last longer in storage. These high-vigor seeds could achieve the goal of less sowing, less labor, lower cost and high benefits, reduce seed waste, improve seed utilization efficiency, and realize the goal of increasing income efficiently.

In 1995, the International Seed Testing Association (ISTA) published a manual on seed vitality testing methods, which introduced eight methods: conductivity testing, accelerated aging testing, freezing testing, low-temperature germination testing, controlled deterioration testing, composite stress vitality testing, brick sand testing, seedling growth testing, and the chlorotriphenyltetrazolium chloride (TTC) staining method (3). These experiments were mainly based on the physiological and biochemical properties of seeds for determination. In 1982, Xu and associates (4) began exploring seed vigor, and determined the vigor of Brassica pekinensis, Phaseolus vulgaris, Raphanus sativus, and other seeds by the germination method (4). In 1996, the vigor of wetland seeds was detected by electrical conductivity and other indicators (5). Traditional methods of seed vigor detection include standard germination tests, TTC staining, conductivity measurement, and accelerated aging tests, which are destructive to seeds. However, these traditional methods for measuring seed vitality commonly have problems such as long experimental cycles, cumbersome operations, and irreversible damage to seeds, making it difficult to meet the needs of modern large-scale breeding efforts (6). From this perspective, the optical technology developed in recent years has become a crucial solution to this issue, thanks to its ability to provide high data throughput, rapid processing, operational simplicity, and non-destructive analysis (7,8).

In 2000, Majumdar and associates (9) used machine vision technology to classify single seeds of different grains, laying the foundation for subsequent determination of seed vitality (9). Machine vision refers to the utilization of computers and cameras to replicate the functions of human vision, specifically for industrial and scientific applications. In the following 20 years, research on seed vigor detection gradually introduced new technologies such as laser speckle technology, ultra-weak luminescence method, hyperspectral imaging (HSI) technology, conductivity detection technology, and fluorescence spectroscopy (8,10–12). The main methods for detecting seed vigor by optical technology at home and abroad include near-infrared (NIR) spectroscopy, hyperspectral imaging, X-ray spectroscopy, infrared thermal imaging, and color selection technology (13–15). At present, in response to the urgent demand for the application of seed vigor classification detection for rice seeds with lemma shells in agriculture planting and selection of seed, a transmission near-infrared absorption spectrum detection system for seed vigor was designed in China, using a wide spectral range and high energy density super-continuum laser light source (NKT SuperK Compact) as the incident light to achieve the rapid and non-destructive classification of single rice seeds (16). The advantages of these detection methods include simple operation, fast detection speed, high accuracy, and non-destructive testing. With the continuous expansion of seed detection technology, non-destructive testing is becoming a new direction for seed vigor determination.

Research Trends in Seed Vigor Measurement by Spectral Techniques

The mainstream spectroscopic techniques and key research objects used in the existing studies were summarized to understand the research progress in the field of determining seed vigor by spectroscopic techniques. Databases of “Web of Science” and “China Knowledge Network” were used to search the academic papers published worldwide in a 10-year period (2013–2022) about the determination of seed vigor by spectral technology, the search terms were “Spectroscopy and seed vigor,” as shown in Figure 1a and 1b.

FIGURE 1: The trend of manuscripts published from 2013 to 2022, showing (a) the number of articles and (b) discipline classification.

A total of 86 articles using spectral techniques to measure seed vigor in the China National Knowledge Infrastructure (CNKI) citation database from 2013 to 2022 were counted, as shown in Figure 1a. Since 2019, there had been an increasing trend year by year, with 70 of the references belonging to crop discipline. A total of 80 directly related articles were retrieved in “Web of Science” in Figure 1b; the largest number of these were distributed in spectroscopy and plant analysis, totaling 26 articles and accounting for 32.5% of the total. The content mainly included research on near-infrared (NIR) spectroscopy, laser spectroscopy, high-time resolution (HIT) spectroscopy, photoacoustic spectroscopy, tunable semiconductor laser absorption spectroscopy (TDLAS) technology, and other technologies applied to the seed vigor of rice, maize, and wheat.

Research Progress in Measuring Seed Vigor Using Optical Non-Destructive Techniques

Method of Seed Quality Detection Based on Optical Characteristics

The tissue of seed would transfer energy in the form of surface reflection and proprioceptive (in depth) reflection when light radiates onto the seed, and the seeds exhibit different characteristics after radiating different intensities and densities of light (Figure 2). Seed vigor was discriminated by correlating the optical feature of seeds with internal vigor indicators, based on the absorption feature of seeds to light. The main methods for seed vigor detection based on optical technology are shown in Table I.

FIGURE 2: Logical mind map for evaluation and development trends of optical detection technology for seed vigor.

Near Infrared Spectroscopy Detection Technology

The principle of NIR spectroscopy uses near-infrared light (wavelength 780-2526 nm) to interact with the functional groups of organic compounds, and the molecules composed of organic compounds absorb the energy of the corresponding part of near-infrared light (17). Because different molecules vibrate at specific frequencies, they generate unique near-infrared absorption spectra. Absorption properties of hydrogen-containing groups (O-H, N-H, C-H, S-H, and P-H) in organic matter with NIR energy are used to study seed composition (18), which correlates to qualitative and quantitative analysis of seed vigor. Moreover, this NIR method had the characteristics of non-destructiveness, low cost, non-pollution, fast speed, no sample damage, and convenient measurement.

Alamery and coauthors (19) introduced a classification system for seed vigor levels by developing a partial least squares (PLS) prediction model. This significant advance validated the viability of NIR technology for non-destructive seed testing. The created binary classification prediction model could identify low, medium, and high-vigor seed groups, with regression results of 80%–100% and 96.3%–96.6%, respectively. Yin Shuxin and colleagues (20) employed vector normalization and principal component analysis to preprocess the data obtained through NIR technology, removing noise, and reducing the complexity of the dataset. A model created using a back propogation (BP) neural network was then used to recognize individual corn seeds, achieving an accuracy of 90.30% and a recognition time of only 27.36 milliseconds (21). Based on the data of maize seeds measured by NIR spectroscopy, Wang and associates used polynomial smoothing preprocessing (SG) curved to reduce noise and applied support vector machine (SVM) to construct a discriminant model to obtain the correct rate of 95.56% for identifying whether seeds were with or without vigor (22). Li and colleagues (23) combined NIR technology with competitive adaptive reweighting (CARS), uninformative variable elimination (UVE), and forward interval partial least squares (FIPLS) to select the optimal modeling band, and used partial least squares regression to establish a prediction model for the related indicators of sweet corn seed vigor, achieving simultaneously measuring three seed vitality related indicators and improving the efficiency of seed non-destructive testing.

However, due to the low sensitivity of NIR technology and its vulnerability to external environmental interference, it is not yet feasible to determine the viability of individual seeds.

Hyperspectral Imaging Detection Technology

Hyperspectral imaging (HSI) detection is an emerging technology that perfectly combines traditional image technology and spectral analysis, possessing ultra-multi band, high spectral resolution, and the ability to merge an image and spectra, enabling comprehensive analysis of samples from multiple perspectives (24). HSI was first applied in the field of remote sensing, and with the continuous development of imaging technology, HSI has gradually been applied to fields such as medicine and agriculture. Currently, due to the advantages of both machine vision and NIR, HSI has become a powerful tool for seed vitality detection (25–28).

In 2016, Ambrose and associates. (25) first introduced HSI into seed vitality monitoring, and successfully distinguished living and dead corn seeds, with a correct classification rate of 95.6% for the prediction set. Kandpal and colleagues combined partial least squares-discriminant analysis (PLS-DA) with three different spectral feature screening algorithms—variable importance in prediction (VIP), selectivity ratio (SR) and significance multivariate correlation (sMC)—to construct a hyperspectral based muskmelon seed vitality recognition model. Among them, the correct recognition rate of the PLS-DA-SR model on the verification set was 94.6% (27). In China, many scholars have studied the application of HSI in seed vitality monitoring, including major cash crops such as rice, wheat, corn, soybeans, and peanuts. In 2023, Zou and associates (8) used HSI to collect spectral information from three types of single peanut seeds with different vitality levels and combined various pretreatment methods and logistic regression (LR), light gradient-boosting machine (LightGBM) and Xgboost3 feature band extraction algorithms to build five classification models including xgboost, catboost, random forest, support vendor classifier (SVC), and gradient boosting decision trees (GBDT). The results showed that the MF-LighTBM-random forest (RF) model emerged as the more effective, achieving a prediction set accuracy of 92.59%. This provided theoretical underpinnings for the rapid, non-destructive, and efficient detection of peanut seed vitality (29). With the continuous development of related technologies, the combination of deep learning and HSI has become a new trend. Wu and associates (28) proposed a deep convolutional neural network combined with HSI for batch detection of rice seeds under unbalanced sample conditions, achieving a recognition accuracy of 97.69%.

In conclusion, due to the unique spectral and imaging characteristics of HSI, domestic and international research has been conducted on seed vitality detection utilizing HSI. This research has encompassed diverse vitality identification algorithms and various seed types. Hyperspectral technology has emerged as a promising research direction in the field of seed vigor detection. Specifically, the ability of HSI to concurrently capture spectral information and seed image data has paved the way for the commercial application of rapid, non-destructive, and batchwise testing of single seed vitality.

Photoacoustic Spectroscopy Detection Technology

Photoacoustic spectroscopy (PAS) is a detection technology developed by utilizing the photoacoustic effect. It has the advantages of high sensitivity and broad applicability and is particularly suitable for the detection of strongly scattering and opaque samples (30). The principle of photoacoustic spectroscopy is that the light source irradiates the target, causing it to absorb energy and heat up. The periodic modulation of the light leads to rapid thermal expansion, which induces pressure waves in the surrounding medium, generating weak sound waves. These sound waves, known as photoacoustic signals, can be detected and amplified to obtain spectroscopic data of the target sample. Li and associates. (31) obtained the photoacoustic spectroscopy of rice seeds by using the Fourier transform infrared spectrometer Nicolet 6700 and the detection system built with PA 300 photoacoustic accessories in the experiment, and established a rapid and non-destructive sorting model for rice seed vigor by combining the least-squares support-vector regression (LS-SVR) algorithm. The main result was that the correlation coefficient and standard deviation of the mixed model building correction set were 0.9701 and 0.4657, respectively, and the correlation coefficient and standard deviation of the prediction set were 0.9562 and 0.5729, respectively. The results showed that photoacoustic spectroscopy had high accuracy in seed vigor detection. At present, photoacoustic spectroscopy is rarely used, but its high accuracy makes it suitable for vigor detection of seeds with rough surfaces and has shown potential in exploring the analysis of seed vigor.

Tunable Diode Laser Absorption Spectroscopy Technology

Many studies have shown that seeds consume oxygen and produce carbon dioxide during aerobic respiration, and the intensity of aerobic respiration of seeds is significant and positively correlated with the vitality intensity of seeds (32–34). TDLAS technology is a trace gas detection technology that has the advantages of high detection limit, good resolution, and strong real-time performance (35). Therefore, TDLAS had been applied to the research of seed vitality.

Based on wavelength modulation spectroscopy (WMS)-TDLAS technology, Gorim and colleagues developed a high-sensitivity CO₂ concentration detection system for seed respiration and applied it to measure the respiration intensity of corn seeds. The results showed that the time resolution was tens to hundreds of times higher than gas chromatography and the non-dispersive infrared (NDIR) CO₂ sensor, providing a basis for further measuring seed vitality intensity (36). Zhai and associates (37) used TDLAS technology and other nondestructive testing techniques,, such as machine vision, electronic nose sensor, HSI, laser speckle detection, NIR spectroscopy, and the H₂O₂ flow rate method through overall quantitative assessment. The results showed that the TDLAS method has better cost performance.

Near-Infrared Super-Continuum Laser Spectroscopy Technology

Generally, there are two modes of NIR spectroscopy when scanning samples, namely the reflection mode and the transmission mode. The reflection mode is mainly used to analyze solid samples, while the transmission mode is used to analyze liquid samples. However, NIR spectroscopy poses challenges in analyzing seed vigor. As seed vigor is mainly related to the seed’s internal composition, useful spectral information cannot be obtained when scanning seeds in the reflectance mode. Moreover, the seed coat, located on the outside of the seed, becomes an interference problem when the transmittance mode is used (16,38).

The NIR supercontinuum laser utilizes a supercontinuum light source, which offers a broad and continuous spectral range with high energy density. This results in a more uniform output spectrum compared to the tungsten-halogen lamps commonly used in traditional NIR spectroscopy. While providing a sufficiently broad spectral range, the optical power output of the supercontinuum laser can be several orders of magnitude higher—typically 4 to 6 orders of magnitude (10³ to 10⁶ times brighter)—than that of a tungsten-halogen lamp. This enhanced power enables high-quality spectral measurements of rice seeds with lemma shells, yielding data with a high signal-to-noise (S/N) ratio across a wide spectral band (39).

Jin and colleagues (16) obtained near-infrared spectral information of three different years of Nipponbare seeds in the 1100-2100 nm NIR band. The analysis results showed that the vigor gradients of rice seeds were correlated with the absorption peaks of their NIR absorption spectra. When the number of principal components was 3, the cumulative contribution rate of the model was 93.5%. The PLS-DA model had a high predictive ability for the grading of rice seed vitality gradients in different years, and the grading accuracy of both the calibration and validation sets could reach over 90%. After screening, the germination rate of rice seeds could reach over 95%.

The detection system of this method and optimized spectral processing method had been used to achieve rapid and non-destructive classification of single grain rice seeds with different vigor, providing an effective technical means for accurately improving seedling quality, screening backlog seeds, selecting and producing high-quality rice, and playing an important guiding role in the development of the seed industry and efficient utilization of germplasm resources.

Seed Quality Detection Method Based on Appearance Features

The seed quality detection method based on appearance features uses machine vision to simulate human visual functions, digitized the obtained image information, and then establishes a discriminative model between the detection index and vigor, which could distinguish the seed vigor (40,41). Machine vision technology is used to identify differences in vitality between different varieties of sweet corn seeds, extract parameters such as color element values, saturation, and projected area of corn seeds, perform correlation analysis of seed vigor, create seed vitality prediction and screening models, and combine artificial neural networks and binary logistic regression data analysis methods to achieve an overall prediction accuracy of up to 77.6%. The results of non-destructive testing of pepper seed vigor using machine vision technology showed that the processing time could be accelerated using threshold processing and flood fill algorithms, and the extracted seed grayscale values could reflect the seed vigor information determined by the seed vigor index testing system with an accuracy of 92% (40). Machine vision technology had high requirements for seed measurement environments and insufficient adaptability in complex agricultural production environments. At the same time, in cases where the seed features were not obvious, it was not only necessary to improve the accuracy of the algorithm, but also to overcome the problem of high computational effort associated with high precision.

Current Status and Prospects for the Application of Optical Nondestructive Testing for Single Seeds

To meet the pressing need for seed selection in agricultural production, the development of a single seed detection and sorting device capable of realizing assembly line seed feeding, detection, and sorting has emerged as an inevitable requirement trend for the seed industry. Traditional seed sorting methods have relied predominantly on the physical characteristics of seed shape for batch screening, aiming to acquire seeds with high uniformity in physical attributes (41). However, such methods often fell short in satisfying the stringent requirements of single seed precision sowing. Therefore, it was important to develop a grain-by-grain non-destructive seed testing and sorting system to screen out moldy and damaged seeds and select high-vigor seeds for ensuring production. Based on machine vision technology, a single maize seed detection and sorting device was designed (42). A single seed feeding structure was designed, and a high-speed industrial camera was used to collect the seed image to obtain the color and morphological characteristics of a single seed. The PLS-DA method was used for discriminant analysis, and the detection models for moldy and damaged seeds were constructed, respectively. The device could achieve single seed detection and sorting, with a sorting accuracy of over 95% for the moldy model and over 89% for the damaged model, and a sorting rate of over 300 seeds/min. The device could realize the full automation of corn seed production from feeding to sorting and could detect and sort moldy and damaged corn seeds in real time. However, the device was not yet sufficient for the screening of high-vigor seeds for industrial development, which was one of the key current research aims.

Jin and associates (16) designed a single seed vigor detection system that included a supercontinuum laser light source, collimator lens, aperture diaphragm, focusing system, quartz optical fiber, and grating spectrometer. By utilizing the wide coverage and strong transmittance of the super-continuum laser spectrum, the light penetrates the seed and obtains information on compositional changes within the seed. It was shown that the germination rate of the Nippon bare seed population with initial germination rates of 78% and 85% had been increased to 98.96%, and a “spectrum-vigor” model for evaluating single seed vigor had been established. By utilizing proteomic research strategies, the differential proteins, and pathways of different groups in the experiment were explored, and it was found that the application of niacin significantly improved seed vigor.

In the pursuit of optical research on seed vitality, researchers might approach the subject from three pivotal aspects. First, the investigation into the optical detection mechanism of seed vigor must be initiated. By scrutinizing the optical properties, material composition, and metabolism pertinent to seed vigor, we can derive clear correlations between optical characteristics and seed vitality, laying a solid theoretical foundation for optical-based seed vigor detection. Second, it is imperative to conduct research on the characteristic spectra of seeds and establish a comprehensive database encompassing spectral signatures of seed vigor. Given the current scarcity of seed varieties and samples in detection research, as well as the absence of a universal spectral database, identifying key spectral features that influence seed vigor offers a viable technical route for optimizing costs and enhancing the efficiency of optical-based non-destructive testing systems in production applications. Finally, with the advancement of electronic information and automation technologies, the integration of optical non-destructive testing for seed vigor with intelligent sorting equipment holds significant potential. Developing cost-effective, rapid, and non-destructive testing and sorting solutions represents a crucial trend. Such advancements in seed grading and processing, when coupled with genetic breeding, can effectively safeguard seed vitality for agricultural production, ultimately leading to improved overall production efficiency.

Acknowledgments

The work was supported by Key Research and Development Program of Hunan Province of China (No. 2021NK2002), Hunan Natural Science Foundation (No. 2022JJ30351) and Hunan Agricultural Science and Technology Innovation Fund Project (No. 2022CX26).

Author Contribution Indication

Xiaoyu Wang (Conceptualization, Data curation, Formal analysis, Methodology, Writing); Mingdong Zhu (Conceptualization); Jie Li (Writing); Yu Yang (Investigation); Hongjun Xie (Project administration); Yonghong Duan (Investigation); Nailiang Cao (Resources); Ruifeng Kan (Software); Yinghong Yu (Conceptualization, Funding acquisition, Project administration, Visualization).

References

1. Zhang, Y.Q.; Wu, X.L.; Li, S.J. Implementing the Seed Project to Open Up a New Stage for Seed Industrialization Development. China Seed Ind. 2018, 11, 1–3. DOI: 10.19462/j.cnki.1671-895x.20181109.012
2. Wu, X.L.; Ning, F.; Hu, X.L. et al. Genetic Modification for Improving Seed Vigor Is Transitioning from Model Plants to Crop Plants. Front. Plant Sci. 2017, 8. DOI: 10.3389/fpls.2017.00008
3. Campbell, M.R.; Sykes, J.; Glover, D.V. Classification of Single- and Double-Mutant Corn Endosperm Genotypes by Near-Infrared Transmittance Spectroscopy. Cereal Chem. 2000, 77 (6), 774–778. DOI: 10.1094/CCHEM.2000.77.6.774.
4. Xu, B.M.; Gu, Z.H.; Zheng, G.H. Discussion on Methods for Measuring Seed Vitality (I) - TTC Quantitative Method. Seed 1982, 02 (021), 12–14. DOI: 10.16590/j.cnki.1001-4705.1982.02.021
5. Li, S.X.; Chen, Y.S.; Wu, Q.M. A Study on the Method for Measuring the Vitality of Wetland Pine Seeds. J. Nanjing For. Univ. 1996, 20 (3), 16–19.
6. Xing, M.Y.; Long, Y.; Wang, Q.Y.; et al. Physiological Alterations and Nondestructive Test Methods of Crop Seed Vigor: A Comprehensive Review. Agriculture-Basel 2023, 13. DOI: 10.3390/agriculture13030527
7. Wang, Y.L.; Peng, Y.K.; Qiao, X.; et al. Discriminant Analysis and Comparison of Corn Seed Vigor based on Multiband Spectrum. Comput. Electron. Agric. 2021, 190. DOI: 10.1016/j.compag.2021.106444
8. Zou, Z.Y.; Chen, J.; Wu, W.J.; et al. Detection of Peanut Seed Vigor based on Hyperspectral Imaging and Chemometrics. Front. Plant Sci. 2023, 14. DOI: 10.3389/fpls.2023.1127108
9. Majumdar, S.; Jayas, D.S. Classification of Cereal Grains using Machine Vision: II. Color Models. Trans. ASABE 2000, 43 (6), 1677–1680.
10. Braga, R.A.; Dal Fabbro, I.M.; Borem, F.M.; et al. Assessment of Seed Viability by Laser Speckle Techniques. Biosyst. Eng. 2003, 86, 287–294. DOI: 10.1016/j.biosystemseng.2003.08.005
11. El Masry, G.; ElGamal, R.; Mandour, N.; et al. Emerging Thermal Imaging Techniques for Seed Quality Evaluation: Principles and Applications. Food Res. Int. 2020, 131, 109025. DOI: 10.1016/j.foodres.2020.109025
12. Grasso, R.; Gulino, M.; Giuffrida, F.; et al. Non-destructive Evaluation of Watermelon Seeds Germination by using Delayed Luminescence. J. Photochem. Photobiol. B, Biol. 2018, 187, 126–130. DOI: 10.1016/j.jphotobiol.2018.08.012
13. de Medeiros, A. D.; Da Silva, L. J.; Ribeiro, J.; et al. Machine Learning for Seed Quality Classification: An Advanced Approach Using Merger Data from FT-NIR Spectroscopy and X-ray Imaging. Sensors 2020, 20 (15), 4319. DOI: 10.3390/s20154319
14. Sun J,; Wang G,; Zhang H, et al. Detection of Fat Content in Peanut Kernels Based on Chemometrics and Hyperspectral Imaging Technology. Infrared Phys. Technol. 2020, 105, 103226. DOI: 10.1016/j.infrared.2020.103226
15. Li, L.; Peng, Y.H.; Li, Y.Y. Design and Experiment on Grading System for Online Non-destructive Detection of Internal and External Quality of Apple. TCSAE 2018, 34 (09), 267–275. DOI: 10.11975/j.issn.1002-6819.2018.09.033
16. Jin, W.L.; Cao. N.L.; Zhu, M.D. et al. Nondestructive Grading Test of Rice Seed Activity using Nearinfrared Super-Continuum Laser Spectrum. Chin. Opt. 2020, 13 (5), 1032–1043. DOI: 10.37188/CO.2020-0027
17. Lin, M.; Lv, J. Determination on Components of Corns Based on Neural Networks and Near Infrared Spectrum. Infrared Technol. 2004, 26 (3), 78–81. DOI: 10.3969/j.issn.1001-8891.2004.03.018
18. Wu, Z.S.; Ouyang, G.Q.; Shi, X.Y.; et al. Absorption and Quantitative Characteristics of C-H Bond and O-H Bond of NIR. Opt. Spectrosc. 2014, 117 (5), 703–709. DOI: 10.1134/S0030400X1411023X
19. Al-Amery, M.; Geneve, R.L.; Sanches, M.F.; et al. Near-infrared Spectroscopy used to Predict Soybean Seed Germination and Vigour. Seed Sci. Res. 2018, 28 (3), 245–252. DOI: 10.1017/S0960258518000119
20. Yin, S.X.; Yang, D.F.; Wang, X.Z.; et al. Application Study of Near Infrared Spectroscopy Technology in Maize Seed Vigor Detection. Mod. Agric. Sci. Technol. 2015, 13, 20–21, 23. DOI: 10.3969/j.issn.1007-5739.2015.13.008
21. Li, J.H.; Yang, Z.L.; Wang, Z.B.; Wang, Z.B. The Corn Content Measurement with Near Infrared Diffuse Transmission. Infrared Technol. 2013, 35 (11), 732–736.
22. Wang, W.X.; Peng, Y.H.; Zhen, X.C et al. Portable Nondestructive Detection Device for Nutrient Components of Pork. TCSAM 2017, 48 (9), 303–311. DOI: 10.6041/j.issn.1000-1298.2017.09.038
23. Li, W.; Li, Y.; Li, G.K. et al. Seed Vigor Detection of Sweet Corn by Near Infrared Spectroscopy Under High Temperature Stress. J. Nucl. Agric. Sci. 2018, 32 (08), 1611–1618. DOI: 10.11869/j.issn.100-8551.2018.08.1611
24. Wang, J.Y.; Sun, L.J.; Feng, G.J. et al. Intelligent Detection of Hard Seeds of Snap Bean Based on Hyperspectral Imaging. Spectrochim. Acta A Mol. Biomol. Spectrosc. 2022, 275, 121169. DOI: 10.1016/j.saa.2022.121169
25. Ambrose, A.; Kandpal, L.M.; Kim, M.S. et al. High Speed Measurement of Corn Seed Viability using Hyperspectral Imaging. Infrared Phys. Technol. 2016, 75, 173–179. DOI: 10.1016/j.infrared.2015.12.008
26. Cui, H.W.; Cheng, Z.S.; Li, P. et al. Prediction of Sweet Corn Seed Germination Based on Hyperspectral Image Technology and Multivariate Data Regression. Sensors 2020, 20 (17), 4744. DOI: 10.3390/s20174744
27. Kandpal, L.M.; Lohumi, S.; Kim, M.S. et al. Near-infrared Hyperspectral Imaging System coupled with Multivariate Methods to Predict Viability and Vigor in Muskmelon Seeds. Sens. Actuators B Chem. 2016, 229, 534–544. DOI: 10.1016/j.snb.2016.02.015
28. Wu, N.; Weng, S.Z.; Chen, J.X. et al. Deep Convolution Neural Network with Weighted Loss to Detect Rice Seeds Vigor Based on Hyperspectral Imaging Under the Sample-Imbalanced Condition. Comput. Electron. Agric. 2022, 196, 106850. DOI: 10.1016/j.compag.2022.106850
29. Zou, Z.Y.; Chen, J.; Zhou, M. et al. Prediction of Peanut Seed Vigor Based on Hyperspectral Images. Food Sci. Technol. 2022, 42, e32822. DOI: 10.1590/fst.32822
30. Wang, Q.; Wang, S. C.; Liu, T. Y. et al. Research Progress of Multi-Component Gas Detection by Photoacoustic Spectroscopy. Spectrosc. Spect. Anal. 2022, 42, 1–8. DOI: 10.3964/j.issn.1000-0593(2022)01-0001-08
31. Li, H.H.; Lu, W.; Du, C.W.; Ma, F.; Luo, H. Study on Rapid and Non-Destructive Detection of Rice Seed Vigor Based on Photoacoustic Spectroscopy Combined with LS-SVR. Chin. J. Lasers 2015, 42 (11), 272–281. DOI: 10.3788/CJL201542.1115003
32. Horbach, M. A.: Dranski, J.; Malavasi, U. C. et al. Physiological Quality and Seed Respiration of Primed Jatropha Curcas Seeds. Acta Bot. Brasilica 2017, 32, 63–69. DOI: 10.1590/0102-33062017abb0231
33. Gorim, L.; Asch, F. Seed Coating Reduces Respiration Losses and Affects Sugar Metabolism during Germination and Early Seedling Growth in Cereals. Funct. Plant Biol. 2014, 42 (2), 209–218. DOI: 10.1071/FP14142
34. Yang Z.R.; Guo F.; Zhang D. et al. Changes of Germination and Respiration of Allium Mongolicum Seeds Under Different Storage Times. J. Northwest A & F Univ. (Nat. Sci. Ed.) 2021, 49 (11), 91–96. DOI: 10.13207/j.cnki.jnwafu.2021.11.012
35. Wang, Q.M.; Wang, Z.Z.; Kamimoto, T. et al. Multi-Species Hydrocarbon Measurement using TDLAS with a Wide Scanning Range DFG Laser. Spectrochim. Acta A Mol. Biomol. Spectrosc. 2022, 265, 120333. DOI: 10.1016/j.saa.2021.120333
36. Gorim, L.; Asch, F. Seed Coating Reduces Respiration Losses and Affects Sugar Metabolism During Germination and Early Seedling Growth in Cereals. Funct. Plant Biol. 2015, 42, 209–218, DOI: 10.1071/FP14142
37. Zhai, Z. J.; Jia, L.Q.; Qi, H. N. Model Evaluation of Seed Viability Detection Without Non-destruction. Seed 2020, 39 (04), 132–135. DOI: 10.16590/j.cnki.1001-4705.2020.04.132
38. Cai, S. Y.; Xu, H.; Wu, L. Y. et al. Physiological Changes Involved in the Acquisition of Seed Vigor during Maturation of Sunflower Seed. Acta Physiol. Plant. 2023, 45 (3), 51. DOI: 10.1007/s11738-023-03534-7
39. Xie, C. Y.; Zhai, W. C.; Li, J. J. et al. System-level Spectral Responsivity Calibration Comparison and Validation of Supercontinuum Laser and Monochromator. Infrared Laser Eng. 2020, 49 (2), 133–140. DOI: 10.3788/IRLA202049.0205005
40. Zhai, Z. J.; Jia, L. Q.; Qi, H. N. et al. Research Advances in Non-Destructive Detection Methods of Seed Vigor. J. Zhejiang A & F Univ. 2020, 37 (02), 382–390. DOI: 10.11833/j.issn.2095-0756.2020.02.025
41. Li, X. F.; Wang, C.; Song, P. et al. Progress in Research on Non-Destructive Testing Methods for Seed Vigor. Seed 2019, 38 (6), 61–65. DOI: 10.16590/j.cnki.1001-4705.2019.06.061
42. Zhang, H.; Dong, H. T.; Li, B. B. et al. Study on Nondestructive Testing Method for Germination Potential of Maize Single Seed. Seed 2021, 40 (11), 144–148. DOI: 10.16590/j.cnki.1001-4705.2021.11.144

Wang Xiaoyu, Li Jie, and Yang Yu are with the Hunan Institute of Agricultural Information and Engineering at the Hunan Academy of Agricultural Sciences, in Changsha, China. Zhu Mingdong, Xie Hongjun, Duan Yonghong, and Yu Yinghong are the Hunan Rice Research Institute at the Hunan Academy of Agricultural Sciences, in Changsha, China. Zhu Mingdong, Cao Nailiang, and Kan Ruifeng are with the Anhui Institute of Optics and Fine Mechanics at the Hefei Institute of Physical Science, in Hefei, China. Direct correspondence to: yyh30678@163.com

Wang Xiaoyu and Zhu Mingdong contributed equally to this work. ●