摘要
结合高光谱图像处理和光谱分析方法,通过一次图像扫描同时对苹果的表面摔伤和糖分含量进行检测。苹果第一主成分图像与794 nm的图像相减后进行去噪和阈值分割处理,摔伤检测的准确率为92.6%。对感兴趣区域的反射光谱曲线进行多元散射校正、一阶导数和SG平滑处理后利用偏最小二乘回归方法建立糖分含量的预测模型,校正集相关系数Rc为0.93,SEC为0.47°Brix,验证集相关系数Rv为0.92,SEV为0.67°Brix。结果表明:利用高光谱成像技术可以实现苹果内部品质和外部品质的同时检测。
The hyperspectral imaging technology was used to detect the bruises and solid soluble content（SSC） of apples simultaneously.The image at 794nm from the first principle component was subtracted,the series of de-noising and threshold processing were performed and the bruises on apples were predicted with the accuracy of 92.6%.In the hyperspectral imaging,region of interest（ROI） was determined and spectrum reference curve was calculated.The original reflectance spectrum curve was processed by multiplicative scatter correction（MSC）,a first derivative and Savitzky-Golay（SG） smoothing,and PLSR model was developed to predict SSC with Rc of 0.93,SEC of 0.47°Brix,and Rv of 0.92,SEV of 0.67°Brix.The research showed that it was feasible to detect the SSC and bruises on apples simultaneously with hyperspectral imaging technology.
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2011年第3期140-144,共5页
Transactions of the Chinese Society for Agricultural Machinery
基金
"十一五"国家科技支撑计划资助项目(2006BAD11A12-11)
国家农业智能装备工程技术研究中心开放课题资助项目
关键词
苹果
糖分含量
摔伤
高光谱
无损检测
Apples
Solid soluble content
Bruises
Hyperspectral
Nondestructive detection
