| 基于特征荧光信号的去囊衣带芯橘瓣分选 |
| |
| 引用本文： | 王叶群,杨增玲,任卫东,刘婷,杨杰,张绍英.基于特征荧光信号的去囊衣带芯橘瓣分选[J].农业工程学报,2017,33(9):296-301. |
| |
| 作者姓名： | 王叶群 杨增玲 任卫东 刘婷 杨杰 张绍英 |
| |
| 作者单位： | 1. 中国农业大学工学院,北京,100083;2. 汇川盛业技术(北京)有限公司,北京,100176 |
| |
| 基金项目： | 国家重点研发计划（2016YFD0400305） |
| |
| 摘 要： | 为实现去囊衣带芯橘瓣的机器视觉分选,给去囊衣带芯橘瓣的机器识别提供直接、准确的判别信号,采用荧光分光光度计对50颗去囊衣带芯橘瓣的橘芯和砂囊分别进行三维荧光光谱扫描,通过对橘芯、砂囊平均三维荧光光谱分析,确定了橘芯相对砂囊的特征荧光信号,据此对带芯橘瓣荧光图像识别的可行性进行了验证。检测发现,在370~390 nm紫外光激发下,橘芯和砂囊在440 nm处的荧光强度差异较大;橘芯与砂囊在370 nm紫外光激发下,440 nm处荧光强度分布的箱线图表明两者荧光强度分布存在明显差异,且以橘芯在440 nm处荧光强度的下四分位数(Q1)与砂囊在440 nm处荧光强度的上四分位数(Q3)的平均值作为分类标准,去囊衣带芯橘瓣检出准确率可达85%。对(370±2)nm激发下得到的单色(440±5)nm荧光图像进行二值化及形态学处理后,可在以砂囊为背景的橘瓣图形中显现橘芯亮斑。利用橘芯与砂囊的荧光特性差异进行机器视觉成像分析,可作为识别去囊衣带芯橘瓣的一种有效方法。
|
| 关 键 词： | 图像处理 光谱 荧光 去囊衣橘瓣 橘芯 砂囊 分选 |
| 收稿时间： | 2016/11/29 0:00:00 |
| 修稿时间： | 2017/5/13 0:00:00 |
Sorting of peeled mandarin segments with orange core based on characteristic fluorescent signal |
| |
| Wang Yequn,Yang Zengling,Ren Weidong,Liu Ting,Yang Jie and Zhang Shaoying.Sorting of peeled mandarin segments with orange core based on characteristic fluorescent signal[J].Transactions of the Chinese Society of Agricultural Engineering,2017,33(9):296-301. |
| |
| Authors: | Wang Yequn Yang Zengling Ren Weidong Liu Ting Yang Jie and Zhang Shaoying |
| |
| Institution: | 1. College of Engineering, China Agriculture University, Beijing 100083, China;,1. College of Engineering, China Agriculture University, Beijing 100083, China;,2. Huichuanshengye Technologies Co.Ltd, Beijing 100176, China;,1. College of Engineering, China Agriculture University, Beijing 100083, China;,1. College of Engineering, China Agriculture University, Beijing 100083, China; and 1. College of Engineering, China Agriculture University, Beijing 100083, China; |
| |
| Abstract: | Abstract: The peeled mandarin segment with orange core (central column residue) is one kind of defective mandarin segment during the production of canned mandarin. So it is necessary to remove the defective mandarin segments according to the requirement of quality before canning. Due to the differences of physical and chemical properties between orange core and orange gizzard, the fluorescence spectra of orange core and orange gizzard are different theoretically. For the purpose of offering direct and accurate signals to the machine to identify the peeled mandarin segments with orange cores, in this study, the peeled mandarin segments were taken as the samples, and the 3D (three-dimensional) fluorescence spectra of orange core and orange gizzard were determined by Cary Eclipse fluorescent spectrophotometer (excitation wavelength in the range of 250-700 nm, 5 nm interval; emission wavelength in the range of 300-800 nm, 1 nm interval; discharge voltage of 600 V, scanning speed of 1 200 nm/min). The differences of fluorescence characteristics of orange core and orange gizzard were found according to the 3D fluorescence spectra after removing Rayleigh and Raman scatter. In order to verify the differences further, the fluorescence spectra of orange core and orange gizzard of 50 peeled mandarin segments were determined by Cary Eclipse fluorescent spectrophotometer (at an excitation wavelength of 370 nm, an emission wavelength in the range of 400-700 nm with the interval of 1 nm, a discharge voltage of 600 V, a scanning speed of 1 200 nm/min). The differences of fluorescence intensities at 440 nm for orange core and orange gizzard were evaluated by means of drawing the box-plots of the fluorescence intensity at 440 nm. The accuracy rate of the discrimination of orange core and orange gizzard was analyzed, and the average value between fluorescence intensity of orange core at the lower quartile and that of orange gizzard at the upper quartile at 440 nm was used as classification criterion. The camera obscura was taken as a platform, and 2 sets of ultraviolet light sources with center wavelength of 375 nm were selected as imaging light sources. The fluorescence images of the mandarin segments with and without orange core were taken using a monochrome fluorescence image acquisition system, which was set up by putting a band-pass filter (440±5 nm) in the front of Image soft G445 camera objective. The grayscale images were obtained by extracting B color component from RGB images, and then the binary images were generated by applying the threshold values, with the IN Vision Assistant software. The results showed that there were significant differences between orange core and orange gizzard of peeled mandarin segments at an excitation wavelength of 370-390 nm and an emission wavelength of 440 nm. According to the classification criterion the discrimination accuracy of orange core and orange gizzard was 85%. Meanwhile, using the fluorescence image acquisition system, the monochrome images of the mandarin segments with and without orange core were obtained respectively. There was significant difference between them, and thus the mandarin segments with orange core and the mandarin segments without orange core could be distinguished effectively. Overall, the study indicates that the fluorescence signal at an excitation wavelength of 370-390 nm and an emission wavelength of 440-450 nm can be the exclusive light signal of the orange core for on-line, nondestructive sorting, which is help to eliminate the peeled mandarin segments with orange core. |
| |
| Keywords: | image processing spectrum fluorescence peeled mandarin segments orange core orange gizzard sorting |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
| 点击此处可从《农业工程学报》浏览原始摘要信息 |
| 点击此处可从《农业工程学报》下载免费的PDF全文 |
|
