基于太赫兹成像技术的发芽葵花籽无损检测研究

Nondestructive Testing of Germinated Sunflower Seeds Based on THz Imaging Technology

发芽葵花籽中脂肪和蛋白质等营养物质的含量会减少,影响油脂产品的产量及品质,因此需要保证葵花籽的品质安全。在对发芽葵花籽进行检测研究时,通常需要剥除葵花籽的外壳,破坏样本,这一操作不利于后续的贮藏加工。文章对发芽葵花籽的带壳无损检测方法进行了研究,采用太赫兹透射成像技术结合最大峰值(Max Peak Size)图像重构方法获取发芽葵花籽图像,应用自适应阈值分割法提取图像特征,应用连续投影(Successive Projections Algorithm,SPA)算法提取光谱特征,并通过串行融合实现特征融合,采用误差反向传播(Back propagation,BP)神经网络算法建立了发芽葵花籽在不同频率段内的识别模型。结果表明,采用SPA-BP算法建立的基于融合特征的发芽葵花籽识别模型效果较好,模型整体识别准确率为100%。由此可得,利用太赫兹时域透射成像技术可以实现发芽葵花籽的带壳无损检测,这对未来带壳油料作物品质的无损检测提供新思路。

The content of fat,protein and other nutrients in germinated sunflower seeds will decrease,affecting the yield and quality of oil products,so it is necessary to ensure the quality and safety of sunflower seeds.In the detection and research of germinated sunflower seeds,it is usually necessary to remove the shell of sunflower seeds and destroy the samples,which is not conducive to subsequent storage and processing.In this paper,the shell nondestructive testing method of germinated sunflower seeds was studied.The terahertz transmission imaging technology combined with the maximum peak size image reconstruction method was used to obtain the germinated sunflower seeds image,and the adaptive threshold segmentation method was used to extract image features.SPA algorithm was used to extract spectral features,and the feature fusion was realized by serial fusion.Error back propagation neural network algorithm to establish the identification model of germinated sunflower seeds in different frequency bands.The results show that the SPA-BP algorithm based on the fusion features of germinated sunflower seed recognition model has a good effect,and the overall recognition accuracy of the model is 100%.It is concluded that the shell nondestructive testing of germinated sunflower seeds can be realized by using terahertz time-domain transmission imaging technology,which provides a new idea for the quality nondestructive testing of oil crops with shell in the future.