高光谱成像技术与迁移学习在单倍体玉米种子识别中的应用

摘要：在玉米育种过程中，正确识别玉米种子的品种极其重要。本文采用深度学习中的迁移学习方法，结合高光谱成像 技术建立模型，通过该模型可以从大量二倍体玉米种子中准确识别出单倍体种子。首先利用高光谱成像技术采集玉米种 子在各个波段的信息，然后基于大型计算机视觉数据库（（Image-Net）预训练的VGG-19网络建立识别模型。利用包含256 个波段（862.5-1704.2nm）的种子光谱图像，模型的正确识别率达到 96.32%，利用包含单波段的种子光谱图像，模型的 正确识别率达到95.75%。该模型的识别效果明显优于现有的DPLS-SVM、PCA-SVM等玉米品种识别算法。实验结果表 明，利用可见光图像数据库预训练的CNN模型可以应用于基于近红外高光谱成像的玉米种子识别，并能获得较高的准确 率。同时，在数据样本量较少的情况下，利用迁移学习仍能实现较高的识别率。该模型不仅满足育种识别的要求，而且 大大降低了样本采集的成本。

关键词：种子识别，迁移学习，高光谱成像技术，玉米单倍体

期刊：2019 International Conference on High Performance Big Data and Intelligent Systems (HPBD&IS)

DOI：10.1109/HPBDIS.2019.8735457

设备：GaiaField

47.Hyperspectral Imaging Technology and Transfer Learning Utilized in Haploid Maize Seeds Identification

Abstract：It is extremely important to correctly identify the cultivars of maize seeds in the breeding process of maize. In this paper, the transfer learning as a method of deep learning is adopted to establish a model by combining with the hyperspectral imaging technology. The haploid seeds can be recognized from large amount of diploid maize ones with great accuracy through the model. First, the information of maize seeds on each wave band is collected using the hyperspectral imaging technology, and then the recognition model is built on VGG-19 network, which is pretrained by large-scale computer vision database (Image-Net). The correct identification rate of model utilizing seed spectral images containing 256 wave bands (862.5-1704.2nm) reaches 96.32%, and the correct identification rate of the model utilizing the seed spectral images containing single-band reaches 95.75%. The identification effect of the model is significantly better than existing maize cultivars identification algorithms, such as DPLS- SVM, PCA-SVM. The experimental results show that, CNN model which is pre-trained by visible light image database can be applied to the near-infrared hyperspectral imaging-based identification of maize seeds, and high accurate identification rate can be achieved. Meanwhile, when there is small amount of data samples, it can still realize high recognition by using transfer learning. The model not only meets the requirements of breeding recognition, but also greatly reduce the cost occurred in sample collection.

Keywords：Seeds identification, Transfer learning, Hyperspectral imaging technology, Maize haploid