SIA OpenIR  > 智能产线与系统研究室
基于计算机视觉的轮毂自动识别与分类
Alternative TitleAutomatic Wheel Identify and Classify Based on Computer Vision
杨飞1,2
Department现代装备研究室
Thesis Advisor刘伟军 ; 夏仁波
ClassificationTP391.41
Keyword轮毂 图像处理 边缘提取 识别分类
Call NumberTP391.41/Y27/2009
Pages54页
Degree Discipline计算机应用技术
Degree Name硕士
2009-05-27
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
Abstract随着全球汽车产量的逐年提高,相应地轮毂市场需求也在迅猛增长,国内80%的汽车轮毂是由铸造而成,铸造而成的轮毂需要进行精加工,而由混流生产线生产的轮毂在再加工时若对其类型进行正确划分将提高生产效率。以往的类型识别是依靠人工识别,工人劳动负荷较大,产生的视觉疲劳和轮毂生产线的速度等多方面原因致使误判情况时常发生,因此,轮毂企业急需一种自动的轮毂识别分类系统来提高生产效率。计算机视觉是利用计算机对景物的图像进行识别,以实现对人视觉功能的扩展的一门高新技术。利用这一技术可以解决许多工业图像识别和检测环节的问题,以取代落后的人工识别,提高识别效率和工业自动化水平。基于计算机视觉的轮毂自动识别与分类系统首先由图像采集系统从生产流水线上采集到要处理轮毂的图片,之后进行图像的预处理,包括轮毂图像区域的提取,图像去噪等操作,构造了四个具有平移不变性、比例不变性和幅度线性变换不变性的一维不变量,用以表征轮毂图像的灰度特征,利用边缘检测Canny算子提取轮毂外圆轮廓,之后对外圆轮廓运用改进的Hough变换方法计算出轮毂最大直径,最后运用马氏距离方法判别出轮毂类型。本文对基于计算机视觉的轮毂自动识别与分类系统进行了总体设计,实现了图像采集、图像处理与轮毂型号识别等基本功能,克服了传统的人工识别的弊端,适应了混流生产线上快速识别分类的需要,开展了面向型号识别的计算机视觉系统的研究工作,为实现工业生产现代新型的图像识别技术做了一些有益的尝试和探讨。
Other AbstractWith the increase of global automobile production year by year,the market demand of wheel grows rapidly. The 80% of the domestic automotive wheel is made by the casting,while casting wheel need finishing,by mixed-flow production line together with the wheel hub in their working hours the right type will increase production efficiency. In the past the job of identification the type relied on manual identification, so the workers under greater workload and the speed of production line resulting in visual fatigue, so the misjudgments occur from time to time. Therefore, enterprises need for a wheel hub automatic identification classification system to improve production efficiency. Computer vision is a high-tech for identification the images of the scene, in order to achieve the expansion of human visual function. Take advantage of this technology we can solve lots of the problem in image recognition and detection, replace the outdated manual identification and improve efficiency and identify the level of industrial automation. The wheel automatic identification and classification system based on computer vision consist of several parts: firstly the image acquisition system collects the wheel images from the production line and deals with the picture; then the image preprocessing system includes the image extraction of wheel regions, image denoising, and so on; finally recognition system realizes the structure four one-dimensional non-invariant variable in translational invariance, the proportion of invariance and the range of linear transformation for the characterization of the gray-scale image wheel characteristics, using the Canny edge detection operator cylindrical wheel contour extraction which is followed by a round of external outline the use of improved Hough transform method to calculate the largest diameter wheel, and at last classify the wheel type by the Mahalanobis distance discriminant method. In this paper, the system of wheel automatic identification and classification was designed, the realization of the image acquisition, image processing and recognition, such as wheel basic functions of models to overcome the traditional disadvantages of artificial recognition to adapt to the rapid identification of mixed-flow production line classification needs, carried out for the model of the computer vision system to identify the study of modern industrial production for the realization of a new type of image recognition technology has made some attempts and discussion.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/582
Collection智能产线与系统研究室
Affiliation1.中国科学院沈阳自动化研究所
2.中国科学院研究生院
Recommended Citation
GB/T 7714
杨飞. 基于计算机视觉的轮毂自动识别与分类[D]. 沈阳. 中国科学院沈阳自动化研究所,2009.
Files in This Item:
File Name/Size DocType Version Access License
10001_20062801472905(7985KB) 开放获取--Application Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[杨飞]'s Articles
Baidu academic
Similar articles in Baidu academic
[杨飞]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[杨飞]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

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