红枣品质检测部分系统设计及分级软件研究

发布时间：2018-04-19 22:33

本文选题：红枣 + 机器视觉技术　；参考：《西北农林科技大学》2016年硕士论文

【摘要】：红枣不仅具有丰富的营养价值,而且具有很高的经济效益,红枣出口在我国水果总出口中占有十分重要的地位,然而现有的红枣分级方式多采用人工分级和机械式分级,人工分级导致分级质量参差不齐、生产效率低下,机械分级易造成机械损伤,极大地影响了红枣的采后附加值,降低了红枣产业的经济效益。因此,实现红枣快速高效检测,以保证红枣的分级质量,提高经济效益是广大果农迫切需要解决的问题。本研究以红枣为研究对象,运用机器视觉技术和近红外光谱技术对红枣的内外部品质进行检测研究。主要研究工作如下:(1)完成了红枣品质在线检测装置的搭建,包括机器视觉系统、近红外光谱系统和传输机构。用机器视觉系统实现红枣外部品质的检测,用近红外光谱系统实现红枣内部品质的检测,传输机构完成红枣的输送工作。(2)完成了基于机器视觉的红枣外部品质分级软件的设计,包括图像数据的读取模块、设置模块、图像处理模块以及结果统计模块,图像处理模块包括图像预处理模块、大小检测模块、病害检测模块和裂纹检测模块。采用灰度化、中值滤波以及二值化分割对红枣图片进行预处理,采用最小外接矩形法确定红枣的大小,采用HSI颜色模型中的H分量来判别红枣的病害区域,采用I分量来判别红枣裂纹区域。并对外部品质检测软件进行测试,分级正确率达到88.24%。(3)运用近红外光谱技术检测红枣水分和总糖含量两个品质指标,分别建立了对应的定量检测模型,包括预处理方法、建模方法、特征波长以及多元线性回归方程的确定,最终得到较高的相关系数和预测准确度。并开发了红枣内部品质检测与分级软件,主要功能模块包括光谱数据的读取模块、有效光谱信息提取及红枣水分和总糖检测模块、分级设置模块和系统界面模块等。开发近红外光谱处理软件,主要包括光谱数据的载入、校正集和预测集的选择、光谱预处理、校正模型的建立、保存,未知样品预测结果显示和保存等功能。
[Abstract]:Red jujube not only has rich nutritional value, but also has high economic benefit. The export of red date occupies a very important position in the total export of fruit in China. However, the existing classification methods of red jujube are mostly classified by artificial and mechanical classification. Artificial grading results in uneven grading quality, low production efficiency and mechanical damage, which greatly affects the post-harvest added value of jujube and reduces the economic benefit of jujube industry. Therefore, it is an urgent problem for the fruit farmers to realize the rapid and efficient detection of jujube, to ensure the classification quality of jujube and to improve the economic benefit. In this study, the internal and external quality of jujube was detected by machine vision and near infrared spectroscopy. The main research work is as follows: (1) the establishment of on-line testing device for jujube quality includes machine vision system, near infrared spectrum system and transmission mechanism. The machine vision system is used to detect the external quality of jujube, the near infrared spectrum system is used to detect the internal quality of red jujube, and the transmission mechanism is used to carry out the transporting work of red jujube. (2) the software for classifying the external quality of red jujube based on machine vision is designed. The image processing module includes image preprocessing module, size detection module, disease detection module and crack detection module. The image of jujube was preprocessed by gray scale, median filter and binary segmentation. The size of jujube was determined by the method of minimum external rectangle. The H component of HSI color model was used to distinguish the disease area of jujube. I component is used to judge the crack area of jujube. The external quality testing software was tested, and the classification accuracy reached 88.24.K3) two quality indexes, water content and total sugar content of jujube, were detected by near infrared spectroscopy, and the corresponding quantitative detection models, including pretreatment methods, were established, respectively. The method of modeling, the determination of characteristic wavelength and multivariate linear regression equation are used to obtain high correlation coefficient and prediction accuracy. The software for internal quality detection and classification of jujube was developed. The main function modules included reading module of spectral data, extracting effective spectral information, detecting moisture and total sugar of jujube, grading setting module and system interface module, etc. Near-infrared spectrum processing software is developed, including spectral data loading, selection of calibration and prediction sets, spectral preprocessing, establishment of calibration model, preservation, display and preservation of unknown sample prediction results, etc.
【学位授予单位】：西北农林科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP317.4;S665.1

【参考文献】