欧亚国际

最近，来自巴西的科学家利用VideometerLab多光谱成像系统发表了题为Using Multispectral Imaging for Detecting Seed-Borne Fungi in Cowpea的文章，揭示了MSI多光谱成像系统在种子和植物病害领域的潜在应用潜力。

北京欧亚国际科技有限公司是丹麦Videometer公司中国区总代理，全面负责其系列产品在中国市场的推广、销售和售后服务。

Using Multispectral Imaging for Detecting Seed-Borne Fungi in Cowpea

by Carlos Henrique Queiroz Rego 1,*,Fabiano França-Silva 1,Francisco Guilhien Gomes-Junior 1,Maria Heloisa Duarte de Moraes 2,André Dantas de Medeiros 3 andClíssia Barboza da Silva 4

1、Department of Crop Science, College of Agriculture “Luiz de Queiroz”, University of São Paulo, Piracicaba 13418-900, SP, Brazil

2、Department of Plant Pathology and Nematology, College of Agriculture “Luiz de Queiroz”, University of São Paulo, Piracicaba 13418-900, SP, Brazil

3、Department of Agronomy, Federal University of Viçosa, Viçosa 36570-900, MG, Brazil

4、Laboratory of Radiobiology and Environment, Center for Nuclear Energy in Agriculture, University of São Paulo, Piracicaba 13416-060, SP, Brazil

Figure 1. Raw RGB images of cowpea seeds and corresponding transformed images into grayscale and by canonical discriminant analysis (nCDA) captured at 780 nm, with reflectance patters in classes of healthy seeds, Fusarium pallidoroseum, Rhizoctonia solani and Aspergillus sp. before incubation (a), and after incubation (b).

Abstract: Recent advances in  multispectral imaging-based  technology  have provided  useful information on seed health in order to optimize the quality control process. In this study, we verified the efficiency of multispectral imaging (MSI) combined with statistical models to assess the cowpea seed health and differentiate seeds carrying different fungal species. Seeds were artificially inoculated with Fusarium pallidoroseum, Rhizoctonia solani and Aspergillus sp. Multispectral images were acquired at 19 wavelengths (365 to 970 nm) from inoculated seeds and freeze-killed ‘incubated’ seeds. Statistical models based on linear discriminant analysis (LDA) were developed using reflectance, color and texture features of the seed images. Results demonstrated that the LDA-based models were efficient in detecting and identifying different species of fungi in cowpea seeds. The model showed above 92% accuracy before incubation  and 99% after incubation,  indicating  that the MSI technique in combination with statistical models can be a useful tool for evalsuating the health status of cowpea seeds. Our findings can be a guide for the development of in-depth studies with more cultivars and fungal species, isolated and in association, for the successful application of MSI in the routine health inspection of cowpea seeds and other important legumes.

Keywords: Vigna unguiculata (L.) Walp; seed health spectroscopy