高光谱图像技术检测柑橘果锈 detection of rust in citrus with hyperspectrai imaging technology.pdfVIP

第 36 卷第 6 期 光电工程 Vol.36, No.6 2009 年 6 月 Opto-Electronic Engineering June, 2009 文章编号：1003-501X(2009)06-0026-05 高光谱图像技术检测柑橘果锈 蔡健荣，王建黑，黄星奕，陈全胜 ( 江苏大学 食品与生物工程学院，江苏 镇江 212013 ) 摘要：高光谱图像技术作为农产品无损检测的新技术，探讨了其在柑橘外部品质检测的可行性。以检测柑橘果锈 为目的，首先对经预处理的高光谱图像数据进行主成分分析，优选出 571 nm、652 nm 和 741 nm 三个特征波长组 成新的图像块；再进行第二步主成分分析，得到的第三主成分图像为最适宜检测柑橘果锈的图像；最后对该图像 进行中值滤波、平方根变换、阈值分割和数字形态学运算完成特征提取。试验结果表明，此算法对柑橘果锈检测 的正确率可达到 90%。研究表明，利用高光谱图像技术结合两步主成分分析算法检测柑橘果锈是可行的。 关键词：高光谱图像技术；主成分分析；柑橘；果锈检测 中图分类号：O433.1; TN919. 81 文献标志码：A doi：10.3969/j.issn.1003-501X.2009.06.006 Detection of Rust in Citrus with Hyperspectral Imaging Technology CAI Jian-rong，WANG Jian-hei ，HUANG Xing-yi ，CHEN Quan-sheng ( School of Food Biological Engineering, Jiangsu University, Zhenjiang 212013, Jiangsu Province, China) Abstract : Hyperspectral imaging technology, a new Non-destructive Testing (NDT) technology of agricultural products, was attempted to detect external quality of citrus. Detection of rust in citrus was studied in this work. Firstly, Principal Component Analysis (PCA) was performed on hyperspectral imaging data after preprocessing to determine three dominant wavelengths (i.e. 571, 652, 741 nm). Secondly, the datacube composed of three dominant wavelength images was analyzed based on PCA again, and the third Principal Component (PC3) image was found to be most suitable for identifying the presence of rust. Finally, rust features on citrus’s surface were segmented by median filtering, square root transformation, determination of optimal threshold and morphological imaging processing. Experimental results show that th