论文部分内容阅读
为了检测在不同时间段下苹果的轻微损伤,选择红富士苹果作为研究对象。分别采集损伤1h内和满24h的32个苹果样本以及16个对照样本。通过对比损伤与正常区域的光谱反射率图像并利用主成分分析(Principle Component Analysis,PCA)方法,识别苹果损伤的最佳波长区域为680—980nm,最佳PCA波段为第四主成分波段,并运用PCA方法对损伤1h内和满24h的富士苹果分别进行检测。研究结果表明,16个对照苹果全正确检出,损伤1h内和满24h的富士苹果检测率分别为78.1%和87.5%,随着时间的推移,检测精度越来越高。该结论将为搭建高光谱检测平台提供一定的理论基础。
In order to detect minor damage to apples at different time periods, Red Fuji apple was selected as the study object. Thirty-two apple samples and 16 control samples were collected within 1 h and at 24 h, respectively. Injury and by comparing the spectral reflectance image and a normal region analysis (Principle Component Analysis, PCA) method using a main component, the best wavelength regions identified as damaged apples 680-980nm, band PCA fourth best band main component, and The PCA method was used to detect the Fuji apples within 1h and 24h respectively. The results showed that 16 control apples were detected correctly, and the detection rates of Fuji apples were 78.1% and 87.5% within 1 h and 1 h, respectively, with the detection accuracy getting higher and higher as time went by. This conclusion will provide a theoretical basis for building hyperspectral detection platform.