光谱学与光谱分析, 2018, 38 (8): 2379, 网络出版: 2018-08-26  

近红外光谱的马铃薯环腐病SIMCA模式识别

SIMCA Discrimination of Ring Rot Potatoes Based on Near Infrared Spectroscopy
作者单位
中国农业机械化科学研究院, 北京 100083
摘要
中国是马铃薯生产和消费大国, 伴随马铃薯主粮化战略推进, 马铃薯对中国农业结构和消费者饮食结构的影响与日俱增。 环腐病是制约马铃薯产业发展的常见病害, 对种薯会造成死苗死株, 对加工原料会降低加工效率和成品质量, 严重可达30%~60%。 传统检测马铃薯病害的主要方法是目测、 机器视觉以及高光谱成像等方法, 目测或机器视觉方式鉴别环腐病需要对样品进行破坏; 高光谱成像技术成本高昂, 存在一定的应用局限性。 因环腐病会造成整薯内部品质变化, 利用近红外光谱技术探测整薯内部品质变化, 从而将环腐病马铃薯从健康薯中区别开来, 具有可行性和实用价值。 创新地尝试利用近红外光谱结合SIMCA模式方法来区分马铃薯环腐病及健康薯。 研究结果表明, 基于主成分分析的SIMCA模式识别能有效判别马铃薯环腐病样品, 模型校正集中环腐病和健康薯的识别率、 拒绝率均为100%; 模型验证集中环腐病的识别率、 拒绝率分别为99.00%和100%, 健康薯的识别率、 拒绝率分别为94.12%和100%, 所建模型精度较高。 利用独立的18个样品进行模型外部验证, 环腐病样品识别率为87.50%, 健康薯识别率为80.00%, 均没有错判。 表明所建SIMCA二值识别模型效果良好, 可满足实际应用, 但模型精度需进一步提高。 马铃薯环腐病发病部位接近表皮0.5 cm左右, 近红外光谱对马铃薯样品有一定的透射和漫反射。 可考虑采集马铃薯接近表皮部分的果肉组织内部光谱信息, 结合马铃薯环腐病的发病机理及近红外漫反射光谱的特性, 利用近红外识别模型进行环腐病判别, 具有一定的创新性和应用性。
Abstract
China is one of the world’s largest countries in potato production and consumption. In 2015, the Chinese government put forward a staple-potato development strategy aimed to change the Chinese traditional diet habit of vegetable-potato and promote potato’s status in safeguarding food security. Potato ring rot is a common disease which has restricted the development of potato industry. With the ring rot potato as the seed, it would cause unhealthy plants; with the ring rot potato as the raw materials for processing, it would cause lower efficiency and worse product quality. Visual inspection, machinevisiontechnology and hyperspectral imaging are the traditional methodsto detect potato diseases. However, it is destructive testing when visual inspection and machinevisiontechnology are used to detect ring rot potatoes; and hyperspectral imaging is at a significant cost. There are some limitations of application on these traditional methods. Internal quality changesof potatoes is caused by ring rot diseases. Near infrared spectroscopy (NIRS) could be used to reflect the quality change of the whole potato. Therefore, NIRS can be used to distinguishring rot potatoes from healthy potatoes. It’s feasible and practical to detect potato ring rot nondestructively with near infrared spectroscopy. Combined with NIRS and soft independent modeling of class analogy (SIMCA), this experiment was aimed to identify ring rot potatoes from healthy potatoes. The results showed that, SIMCA mode based on principal component analysis (PCA) was effective to identify ring rot potatoes. In calibration set, the recognition rate and rejection rate of ring rot potatoes and healthy potatoes were both 100%. In validation set, the recognition rate and rejection rate of ring rot potatoes were 99.00% and 100%. The recognition rate and rejection rate of healthy potatoes were 94.12% and 100%. For external validation, the recognition rate of ring rot potatoes and healthy potatoes were 87.50% and 80.00% respectively without misjudge. The SIMCA model was accurate in prediction and suitable to practical application, but the precision would be improved in further research. The pathogenic site of ring rot potatoes was close to epidermis for about 0.5 cm; and there was transmission and diffuse reflection when NIRS Penetrating potatoes. So that it is possible to collect the NIRS information of potato tuber flesh near to potato epidermis. Combined with pathogenic mechanism of potato ring rot disease and characteristics of near-infrared diffuse reflectance spectra, it is innovative and practical to use NIRS to distinguish ring rot potatoes from healthy potatoes.

张小燕, 杨炳南, 曹有福, 李少萍, 赵庆亮, 兴丽. 近红外光谱的马铃薯环腐病SIMCA模式识别[J]. 光谱学与光谱分析, 2018, 38(8): 2379. ZHANG Xiao-yan, YANG Bing-nan, CAO You-fu, LI Shao-ping, ZHAO Qing-liang, XING Li. SIMCA Discrimination of Ring Rot Potatoes Based on Near Infrared Spectroscopy[J]. Spectroscopy and Spectral Analysis, 2018, 38(8): 2379.

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!