【作者】 周武；

【导师】 汪仁煌；

【作者基本信息】 广东工业大学 ， 检测技术与自动化装置， 2008， 硕士

【摘要】 指针式万用表表壳图像的自动处理和识别,是基于机器视觉的指针式万用表的自动检定系统的一个重要组成部分。本文采用机器视觉、图像处理等技术实现了指针式万用表表壳图像的自动处理和识别,并重点研究了表壳图像的分割、倾斜表壳图像的处理和表壳字符的分割识别。在图像的分割、倾斜处理校正及字符识别方面,国内外都取得了一些研究进展,产生了一些比较成熟的处理方法。但是,针对常用的指针万用表表壳图像的处理,在国内外的研究中比较少见,研究资料也相当匮乏。因此,基于机器视觉的指针万用表表壳图像的自动处理和识别,具有重大的实际意义和经济效益。本文主要做了以下几个方面的工作:1、研究了指针式万用表表壳图像的预处理和分割方法:对指针万用表表壳图像进行了较好的预处理,分割提取了多量程切换开关的指针、黑色圆盘刻度线,并设计算法对刻度线进行精确修补。分割出了黑色圆盘周围的刻度值,对指针对应刻度值位置进行了识别,并分割提取出刻度线所指的字符区域。研究了针对指针万用表表壳图像的字符切分和粘连字符的处理方法。2、研究了指针式万用表表壳图像的字符识别和类别识别方法:采用模板匹配实现了字符识别,详细介绍了模板的制作以及字符模板的匹配计算方法。针对指针万用表表壳图像的特点,研究了字符的类别识别方法:区域颜色识别方法和弧度线包括刻度线的范围识别方法,并设计实现了字符识别算法和字符类别识别算法。3、设计实现指针万用表表壳图像处理系统,实现了自动识别,具有一定的实用价值:通过Matlab7.0软件实现了上述方法和算法,对指针万用表表壳图像进行了仿真分析和处理。设计了表壳图像处理系统的软件,通过实际表壳图像处理,取得了较好的效果。更多还原

【Abstract】 The automatic processing and recognition of the shell image of pointer multi-meter is one of an important component for the automatic calibration devices which is based on machine vision. In this thesis, the technology of machine vision and image processing has been applied to process the automatic processing and recognition of the shell image of pointer multi-meter, which is stressed on image segmentation, tilt correction and character segmentation and recognition.In the aspect of segmentation, tilt correction and character recognition, the technology has become mature and it has acquired great achievements all around the world. However, the research of the shell image of the pointer multi-meter is inadequate and the analytic documents about it are also seldom. As a result, it is of great significant to research on the shell image processing.This thesis mainly did something and got some achievements as follows:1. Research the pre-processing and segmentation of the shell image of the pointer multi-meter.Research the pre-processing of the shell image of the multi-meter, segment and extract the pointer of the switch to multi-meter measuring range and the compass display of the black disk. Design algorithms to realize accurate image inpainting for the compass display. Segment and extract the graduated values round the black disk, then recognize the location of the pointer which relevant to the graduated value. At last, extract the text region. Research the character segmentation and character conglutination for the shell image of the point multi-meter.2. Research the text recognition and classification recognition of the shell image of the pointer multi-meter.Realize the text recognition through template matching. Introduce the methods of how to make the template and how to calculate the template matching. According to the feature of the shell image of the pointer multi-meter, research the character category recognition: the method of regional color recognition and the method of the range of curve contain the compass display. Then design and realize the algorithm of the text recognition and character category recognition.3. Design and realize the image processing system of the shell image of pointer multi-meter.Through using the software of Matlab7.0, good results have acquired from the experiment of the automatic processing and recognition of the shell image of pointer multi-meter. The algorithms and methods have been realized and applied to the automatic calibration system.更多还原