Quality Assessment of Crop Seeds by Near-Infrared Hyperspectral Imaging
Non-destructively analyzing the quality of crop seeds is very important for early generation screen-ing in crop breeding. In this study, winter wheat and soybean seeds were measured by a near-infrared (NIR) pushbroom hyperspectral imaging system. Hyperspectral imaging has advantages over conventional NIR spectroscopy by providing both spectral and spatial information simultane-ously. The reflectance spectral images were collected at 850-1700 nm with a resolution of 2.7 nm. The spectrum of each sample was extracted from the data cube using image processing method. Partial least square regression (PLSR) was then used to construct the calibration models. For the determination of crude protein of winter wheat, the correlation coefficient of calibration was r=0.973, the standard deviation of prediction was SEP=0.556, and the relative of SEP was SEP％=3.399％. For the determination of crude protein and crude fat of soybean, the results were r=0.902, SEP=1.332, SEP％=3.195％ and f=0.901, SEP=0.613, SEP％=3.148％, respec-tively. The results showed that NIR hyperspectral imaging could accurately evaluate the component of grain seeds. The extracted spectra from different seed positions and the gaps between them have significant difference. The data of hyperspectral image contained a lot of redundant informa-tion. Therefore, genetic algorithm (GA) was applied to select sensitive wavelengths for hypercube. The results showed that GA did not significantly improve the model performance; however, it could simplify the calculations. Moreover, based on the selected sensitive wavelengths, the low-cost multi-spectral imaging system could be developed specially for the quality assessment of wheat or soybean seeds. It was concluded that hyperspectral imaging was useful for the quality assessment of breeding materials and had potential application for assisting crop breeding.
Hyperspectral Imaging Near Infrared Genetic Algorithm Wheat Soybean Seed
Dazhou Zhu Kun Wang Dongyan Zhang Wenjiang Huang Guijun Yang Zhihong Ma Cheng Wang
National Research Center of Intelligent Equipment for Agriculture, Beijing 100097, P. R. China Natio National Research Center of Intelligent Equipment for Agriculture, Beijing 100097, P. R. China National Engineering Research Center for Information Technology in Agriculture, Beijing 100097, P. R Beijing Research Center for Agri-Food Testing and Farmland Monitoring, Beijing 100097, P. R. China
国际会议
南昌
英文
1144-1150
2010-10-22（万方平台首次上网日期，不代表论文的发表时间）