基于机器视觉的枸杞外观品质检测与评级方法研究

发布时间：2018-12-23 21:01

【摘要】：枸杞的分级目前主要依靠人工进行,这种分级方法受人的主观意愿的影响,分级不准确,同时费时费力,生产效率不高,不仅不能保证枸杞的外观品质,而且无法满足市场上品种多样化的需求。因此,本文采用机器视觉技术对枸杞的表面品质进行了检测识别及分级,实现枸杞的自动、无损检测。针对枸杞的外观品质,主要涉及枸杞的大小、表面缺陷、颜色等三个特征,主要研究内容和结果如下:(1)针对枸杞分级检测中对图像特征提取的需要,对几种常用颜色空间模型的特点分别进行了系统的分析和阐述,同时对图像的灰度化、二值化、图像的分割、图像的边缘检测、图像滤波等图像的基础算法进行了分析和研究,选择出适于枸杞外观品质检测分级的预处理算法。(2)为了实现对枸杞大小的分级,通过研究获取枸杞的投影表面积、长轴直径、短轴直径、偏心率、圆度等几何特征参数的方法和技术,建立了枸杞长轴径的人工测量值与像素值之间的相关模型。(3)根据色度值识别出破损和油粒区域,再通过滤波处理,统计破损和油粒区域的面积像素个数,根据这部分区域与整个健全枸杞面积的比值进行对缺陷程度的判别,实现对枸杞表面缺陷程度的检测与判断,正确率达93.4%。(4)以HSI颜色空间模型为依据,提取枸杞的颜色特征参数。用色度的范围对枸杞进行分级,最后计算色度、饱和度、亮度的均值、方差,判断分级的准确性。(5)利用Matlab/GUI软件平台开发一套枸杞自动检测评价系统,该软件采用模块化设计,包括文件模块、图像采集模块、图像预处理模块,特征参数提取模块和分级模块。实现了人机合一的可视化操作,界面友好,操作简单,易于维护。试验结果证明,利用该系统对枸杞进行检测分级速度快,准确率高。
[Abstract]:At present, the classification of Lycium barbarum is mainly carried out manually. This classification method is influenced by people's subjective will. It is inaccurate, laborious and inefficient in production. It is not only unable to guarantee the appearance quality of Lycium barbarum. And unable to meet the market variety of demand. Therefore, the surface quality of Lycium barbarum was detected and classified by machine vision technology, and the automatic and nondestructive testing of Lycium barbarum was realized. Aiming at the appearance quality of Lycium barbarum, it mainly involves the size, surface defect and color of Lycium barbarum. The main research contents and results are as follows: (1) aiming at the need of image feature extraction in classification detection of Lycium barbarum, The characteristics of several commonly used color space models are systematically analyzed and expounded. At the same time, the basic algorithms of image grayscale, binarization, image segmentation, image edge detection, image filtering and so on are analyzed and studied. In order to realize the classification of the size of Lycium barbarum, the projection surface area, long axis diameter, short axis diameter, eccentricity rate of Lycium barbarum were obtained by studying the surface area, long axis diameter, short axis diameter and eccentricity of Lycium barbarum. Based on the method and technique of geometric characteristic parameters such as roundness, the correlation model between the measured value and pixel value of the long axis diameter of Lycium barbarum is established. (3) according to the chroma value, the damaged and oil particle regions are identified, and then processed by filtering. According to the ratio of this area to the whole intact wolfberry area, the defect degree is judged, and the surface defect degree of Lycium barbarum is detected and judged. (4) based on HSI color space model, the color characteristic parameters of Lycium barbarum were extracted. This paper classifies Lycium barbarum with the range of chrominance, calculates the mean value, variance of chrominance, saturation and brightness, and judges the accuracy of classification. (5) A set of automatic detection and evaluation system of Lycium barbarum is developed by using Matlab/GUI software platform. The software is designed by modularization, including file module, image acquisition module, image preprocessing module, feature parameter extraction module and classification module. The visual operation is realized, the interface is friendly, the operation is simple and easy to maintain. The experimental results show that the system is fast and accurate in the detection and classification of Lycium barbarum.
【学位授予单位】：兰州理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：S567.19;TP391.41

【参考文献】