基于近红外高光谱图像的冬枣损伤早期检测

发布时间：2018-06-05 06:26

本文选题：冬枣 + 高光谱成像　；参考：《食品科学》2017年02期

【摘要】：为了对冬枣损伤进行早期检测,采用近红外高光谱图像技术对损伤区域成像。针对高光谱图像波长多的特点,分别采用连续投影算法、相关特征选择算法、一致性(Consistency)算法选择冬枣损伤的特征波长,对提取的特征波长分别应用k-邻近、朴素贝叶斯(naive bayes,NB)、支持向量机(support vector machine,SVM)3种分类方法进行损伤区域识别。结果表明:所有方法选择的一致特征波长在1 353 nm和1 691 nm附近。Consistency算法选择的特征波长在SVM分类器下分类识别正确率达到95.16%,一致特征波长在NB分类器下分类识别正确率达到84.26%,验证了一致波长的有效性,为多光谱成像技术实现在线检测冬枣损伤提供参考依据。
[Abstract]:In order to detect the damage of winter jujube early, the near infrared hyperspectral image technique was used to image the damaged area. Aiming at the characteristics of multiple wavelengths in hyperspectral images, continuous projection algorithm, correlation feature selection algorithm and consistent Consistencyalgorithm are used to select the characteristic wavelengths of winter jujube damage respectively. Three classification methods, namely naive Bayes and support Vector machines (SVM), are used to identify damage regions. The results show that the correct rate of classification and recognition of the characteristic wavelength selected by all the methods is 95.16% under SVM classifier at 1 353 nm and 1 691nm respectively, and the consistent feature wavelength is classified and recognized by NB classifier. The accuracy is 84.26, which verifies the validity of the consistent wavelength. It provides a reference for multi-spectral imaging technology to detect the damage of winter jujube on-line.
【作者单位】：西北农林科技大学机械与电子工程学院;霍恩海姆大学农业工程研究所;
【基金】：陕西省自然科学基础研究计划-青年人才项目(2015JQ3065) 中国博士后科学基金项目(2015M572602) 西北农林科技大学国际科技合作种子基金项目(A213021505)
【分类号】：S665.1;TP391.41

【参考文献】