基于高光谱图像的小麦种子年份快速鉴别分析研究

Rapid Research on the Storage Year of Wheat Seed Based on Hyperspectrum Imaging

小麦种子的发芽率对小麦的增产增收至关重要,而小麦种子的发芽率与小麦种子的储藏年份有着密切关系。一般而言,随着储藏年份的增加,小麦种子的发芽率降低,因此小麦种子的储藏年份的鉴别研究具有重要意义。研究了不同年份的小麦种子,通过主成分分析,去除波段之间的多余信息,根据PCA变换的RGB彩色合成,可以更客观地看到不同年份小麦种子在图像显示上的差别。然后在不同年份种子的基础上提取平均值、标准差,对得到的光谱曲线对比分析,选取了500 nm和700 nm 2个波段分别构建进行归一化植被指数(NDVI)和增强型植被指数(EVI)光谱指数。利用7个年份的小麦种子建模,分析NDVI和EVI与其相应年份的散点图和趋势线,得到了决定系数0.7624和0.8585,从这2个决定系数可以看出EVI的建模高于NDVI。利用构建的建模模型对2010、2012、2013和2014年份的小麦种子进行年份预测,预测的年份与实际年份达到100%。本研究以小麦种子为例,利用成像高光谱技术鉴定小麦种子的年份,为农产品时间鉴定提供一种新的技术参考。

The germination rate of wheat seed was very important to increase the yield and income of wheat,and it was closely related to the storage year of wheat seed.The germination rate of wheat seed decreased with the increase of storage year.Therefore,the identification of storage year of wheat seed was of great significance.In this paper,the wheat seeds of different years were studied.Through the principal component analysis,the redundant information between bands was removed.According to the RGB color synthetic image,the differences of wheat seeds of different years were objectively seen.Then the spectral curves of wheat seeds of different years were analyzed by extracting the average value and standard deviation,and the normalized vegetation index(NDVI)was determined at 500 nm and 700 nm.And enhanced vegetation index(EVI).After the abnormal data were deleted,seven years of wheat seeds were used to model,and four years of seeds were used as test data.The scatter plot and trend line of NDVI、EVI and their corresponding years were analyzed.The determination coefficients were 0.7624 and 0.8585,respectively.It was obvious that EVI modeling was higher than NDVI.The model was used to predict the year of wheat seeds in 2010,2012,2013 and 2014,and the predicted year was 100%of the actual year.Taking wheat seeds as an example,this study used imaging hyperspectral technology to identify the year of wheat seeds,providing a new technical reference for time identification of agricultural products.