SIA OpenIR  > 自动化系统研究室  > 学位论文
其他题名: Research of Red Dates Automatic Classification Technology Base on Machine Vision
作者: 沈贵萍
导师: 马铖
分类号: TP391.41
关键词: 机器视觉 ; 红枣 ; 图像处理 ; 特征提取 ; 分级
索取号: TP391.41/S42/2012
学位专业: 模式识别与智能系统
学位类别: 硕士
答辩日期: 2012-05-28
授予单位: 中国科学院沈阳自动化研究所
学位授予地点: 中国科学院沈阳自动化研究所
作者部门: 自动化系统研究室
中文摘要: 红枣产业是我国农村经济的一大支柱产业,但是目前红枣分级主要是人工完成,分级精度差,工作效率低,导致我国红枣出口量少,附加值低。由于计算机视觉分级精度和自动化程度高,属于非接触检测过程,因此将其用于红枣外观品质检测具有潜在的应用价值和很好的发展前景。采用机器视觉技术,研究红枣品质的实时检测与分级方法,提高我国红枣市场的竞争力,具有非常现实的社会和经济意义。 本文基于机器视觉技术,对采集的红枣图像进行处理,得到描述红枣外部品质的大小、形状及颜色特征,结合模式识别知识根据提取的特征组合判断红枣所属等级,从而实现红枣的自动分级技术。主要研究工作如下: 1.提出了视频流采集方法,利用DirectShow创建视频流捕获图,在预览时进行抓帧实时处理,相比以前采用的触发方式速度快。 2.红枣图像的分割。研究分析了通常采用的基于边缘的图像分割方法和基于区域的图像分割方法,由于受红枣图像阴影灰度级的影响大,很难完成分割,在颜色空间里处理,实验证明在HSV颜色空间里利用饱和度可以将红枣图像与背景分割,再利用二值形态学消除噪音和毛刺,同时填充前景区域内的孔洞,得到理想的分割后二值图像。 3.提取红枣的外部特征。红枣分级主要依据其大小、形状及颜色特征。本文采用面积和周长描述大小;采用长短轴比、矩形度和傅里叶描述子表示红枣的形状特征,其中,长短轴和矩形度通过红枣图像的最小外接矩形来描述,傅里叶描述子是对红枣的极半径归一化处理后再进行傅里叶变换得到的系数,实验证明取前9个描述子即可很好的重建红枣原始轮廓形状,再对傅里叶描述子归一化处理,使其具有旋转、平移、尺度不变性,仅受形状的影响,利用欧式距离判断待分级红枣与标准样本形状的一致性;采用色调量化直方图描述颜色特征,统计分布在红色区域的像素占整个红枣区域的面积比,判断红枣的成熟度。 4.利用专家规则的阈值判别算法设计红枣分级判别方法,算法简单有效。 由分级结果得出:基于本研究的设计方法结构合理,软硬件可行,并得到较准确的分类结果,系统性能可以满足红枣外观品质检测的实际需求。
英文摘要: Red dates industry is a major pillar industries of China’ rural economy, but presently the Red dates classification is done manually, the classification accuracy and efficiency is low, which result in the low red dates export volume and low value-added. Due to the high classification accuracy and automation of computer vision, a non-contact detection system, so its potential value and good prospects for fruit appearance quality detection. Using machine vision technology and researching the red dates quality real-time detection and classification methods can improve the market competitiveness of red dates, which have real social and economic significance. In this paper, it’s based on machine vision technology to process acquired red date images, obtain red dates external features such as size、shape and hue character, and according to the characteristics of extraction combined judge red dates surbordinate level combining pattern recognition knowledge, consequently, it can realize red dates automatic classification. Main contributions of the dissertation are summerized as follows: 1. Video stream capture method is proposed to obtain red date RGB images. We use DirectShow to create video capture graph, and grab frame during previewing in real-time. This method is faster compared with former trigger mode. 2. Red date image segmentation. We studied and analyzed the usual edge-based segmentation and region-based segmentation, due to the huge impact of the gray-scale of the red date image shadow, the usual segmentation methods are hard to segment the images. So we process images in color space. Experiments show that in HSV color space, we may use saturation to achieve the segmentation between red data image and background, then use the binary mathematical morphological operator method to eliminate noise and burr,at the same time fill the holes of the foreground region of images and finally derive the ideal binary image. 3. Extract external features of red data. The classification of red data is according to size, shape and color. We use the area and perimeter to describe size; use the length of the axial ratio,rectangularity,and fourier operator promotor to describe the shape. The length of the axial ratio and rectangularity can be described through the minimum bounding rectangle.Fourier descriptor is the coefficients of Fourier transform of red dates’ normolized polar radius, experiments show that the front 9 descriptor can reconstruct the original profile very well, then normalize Fourier descriptor to make it have rotating, translation and scale invariance, after that, we use Euclidean distance to judge the shape of consistency. We use the tone quantization histogram to describe the color distribution, we judge the maturity dates according the statistical distribution of the pixels in the red areas of the red date area ratio. 4. We design red dates classification method based on the expert of the rules of threshold discrimnation algorithm, this method is simple and effective. From the classification results we can know that: the structure of design of this study is reasonable. The hardware and software is feasible. The classification result is accurate. The system performance can satisfy the red dates appearance quality detection.
语种: 中文
产权排序: 1
内容类型: 学位论文
Appears in Collections:自动化系统研究室_学位论文

Files in This Item:
File Name/ File Size Content Type Version Access License
基于机器视觉的红枣自动分级技术的研究.pdf(1597KB)----限制开放 联系获取全文

Recommended Citation:
沈贵萍. 基于机器视觉的红枣自动分级技术的研究[D]. 中国科学院沈阳自动化研究所. 中国科学院沈阳自动化研究所. 2012.
Recommend this item
Sava as my favorate item
Show this item's statistics
Export Endnote File
Google Scholar
Similar articles in Google Scholar
[沈贵萍]'s Articles
CSDL cross search
Similar articles in CSDL Cross Search
[沈贵萍]‘s Articles
Related Copyright Policies
Social Bookmarking
Add to CiteULike Add to Connotea Add to Add to Digg Add to Reddit
所有评论 (0)
内 容:
Email:  *
验证码:   刷新
标 题:
内 容:
Email:  *
验证码:   刷新

Items in IR are protected by copyright, with all rights reserved, unless otherwise indicated.



Valid XHTML 1.0!
Copyright © 2007-2017  中国科学院沈阳自动化研究所 - Feedback
Powered by CSpace