SIA OpenIR  > 工业控制网络与系统研究室
基于机器视觉的指针仪表测读方法研究
Alternative TitleResearch on Machine Vision-based Reading Method for Pointer Meters
李栋1,2
Department工业控制网络与系统研究室
Thesis Advisor杨志家
ClassificationTP391.41
Keyword指针仪表 圆心标定 图像分割 Hough变换 中心投影
Call NumberTP391.41/L31/2012
Pages79页
Degree Discipline模式识别与智能系统
Degree Name硕士
2012-05-28
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
Abstract在工业无线技术得到大量应用的今天,利用图像信息对传统指针式仪表进行准确测读具有十分广泛的现实意义。虽然目前有许多图像处理方法可以实现仪表的自动测读,但这些方法在测读的速度、精确度和抗干扰性等方面都有不同程度的缺陷。本文对基于机器视觉的仪表测读过程进行环节划分,针对各个不同的环节,分别寻求相对理想的图像处理方法,从而实现测读方法整体性能的优化,在测读速度、精确度和抗干扰性等方面达到较好的平衡效果。 本文重点对表盘圆心的标定、仪表图像的分割、指针信息的提取这三个关键环节的技术方法进行了研究。  (1)在圆心标定方面,由于仪表的灰度图像在表盘内外存在明显的灰度值差异,根据能量最小原理,可以迭代求取使得能量函数达到极小值的圆心和半径信息。这种方法充分利用了仪表图像的整体灰度信息,受局部区域噪声干扰的影响较小,标定效果准确。   (2)在仪表图像的分割方面,首先利用一种基于像素点比例的自适应阈值法对原始图像进行二值化分割;然后利用Zhang快速并行细化算法对二值化图像做细化处理,进一步简化图像信息,从而有效减少后续的指针信息提取环节的计算量。  (3)在指针信息的提取方面,抓住已标定的表盘圆心信息,分别使用Hough变换法、中心投影法和最小二乘直线拟合法提取指针信息,通过对比计算的速度和精确度等指标,给出最理想的提取方法;同时根据指针与表盘中心的相对位置确定指针方向信息,为指针示值计算提供足够的信息量。   (4)最后,本文介绍了整套算法在STM32嵌入式平台和TI DSP平台上移植和验证的过程。首先描述了硬件平台的主体结构,接着分析了从图像数据采集、处理到最终计算结果输出之间的信息流动控制,最后介绍了算法和控制程序的整体框架,给出了不同平台下算法运行的效果分析。
Other AbstractBecause industrial wireless technology has been widely used today, Using image information to read traditional pointer meters accurately has a wide range of practical significance. Although there are many image processing methods to achieve the automatic measurement of the meter, but these methods have varying degrees of impairment in terms of reading speed, accuracy and robustness. This paper breaks the machine vision-based instrument measuring process down to various aspects, and searches for relatively ideal image processing approach to achieve the optimization of the overall performance of the reading method that can achieve a better balance effect on the reading speed, accuracy and robustness.    This article focuses on technology study of three key aspects: the calibration of meter center, the segmentation of meter image, the extraction of the pointer information.    (1)On the calibration of meter center, because the grayscale images has an obvious difference of gray value inside and outside the instrument dial, based on energy minimization principle, the center and radius information that leads the energy function down to its minimum can be iterated to strike. This method takes full advantage of the image gray level information of the instrument, and takes small disturbances by local area noise so that it gives accurate results of calibration.    (2)On the segmentation of the instrument image, it first uses a kind of method which gets an adaptive threshold based on the proportion of pixels to achieve the binarization segmentation of original image; then it uses the Zhang fast parallel thinning algorithm for binary image refinement process to further streamline image information, thus the subsequent calculation on the extraction of the pointer information can be effectively reduced.    (3)On the extraction of the pointer information, it seizes the calibration information of the meter center and uses the Hough transform, the central projection method and the least squares method to extract the pointer information. Then it takes comparison on the calculation speed and precision degree to choose the best method of extracting. It also determines the direction of the pointer based on the relative position of the pointer and dial center, so it can provide the provision of adequate amount of information for the calculation of the meter value.     (4)Finally, this paper introduces the porting and verification process of the set of algorithms on the STM32 embedded platform and the TI DSP platform. First it describes the main structure of the hardware platform, and then analyzes the control of the information flow from the image data acquisition and processing to the final calculation. Finally it gives the overall framework of the algorithm and control procedures, and shows the analysis of the effect of the algorithm running on different platforms.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/9334
Collection工业控制网络与系统研究室
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
李栋. 基于机器视觉的指针仪表测读方法研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2012.
Files in This Item:
File Name/Size DocType Version Access License
基于机器视觉的指针仪表测读方法研究.pd(1796KB) 开放获取LicenseApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[李栋]'s Articles
Baidu academic
Similar articles in Baidu academic
[李栋]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[李栋]'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.