文档介绍：
基于高光谱反射图像检测鸭梨碰伤#
任梦佳,崔笛**
(浙江大学生物系统工程与食品科学学院, 杭州 310058)
5
10
15
20
25
30
35
40
摘要:鸭梨是我国白梨品系的传统优良主栽品种,在全国出口水果中居重要位置,每年都为
出口创汇做突出贡献。为了提升我国鸭梨在国际市场上的地位和竞争力,满足消费者对外观
美、质量好的高档水果的需求,对鸭梨碰伤进行检测十分必要。本文以鸭梨为研究对象,采
用反射方式采集样本的高光谱图像,初步探索了基于高光谱反射图像分析的鸭梨皮下碰伤检
测方法。通过使样本从不同高度自由下落撞击地面制备不同程度的碰伤。实验结果表明:在
400-1000 nm 波段范围内,无碰伤样本的相对反射率大于碰伤样本,且碰伤程度越严重,其
相对反射率值越小。对 400-1000 nm 波段高光谱反射图像进行主成分分析的结果表明,通过
依次对第四主成分图像进行中值滤波、阈值分割、填充处理和连通区域面积筛选,可以实现
碰伤区域的分割。该结果为后续鸭梨碰伤面积的定量检测奠定了基础。
关键词:生物系统工程;高光谱;反射;图像;鸭梨;碰伤
中图分类号:S2
Detecting 'Yali' Pear Bruise Using Hyperspectral Imaging in
Reflectance Mode
REN Mengjia, CUI Di
(School of Biosystems Engineering and Food Science,Zhejiang University, Hangzhou 310058)
Abstract: 'Yali' pears play an important role in China export fruits. A hyperspectral imaging
system in reflectance mode was applied to investigate the possibility of detecting the bruise in Yali
pears. The buised pears was developed by dropping them to the ground from different heights. The
results showed that the relative reflectance of normal pears was higher than bruised ones in the
range of 400-1000 nm, and severer bruised pears had lower relative reflectance. In addition,
ponent analysis (PCA) was performed by using the images in the range of 400-1000
nm and fourth PC (PC-4) image was selected to segment the bruised regions which was
plished by median filter algorithm, thre