【作者】 江伟；

【导师】 储开宇；

【作者基本信息】 华北电力大学 ， 机械制造及其自动化， 2015， 硕士

【摘要】 机器视觉中特定目标的识别与处理是机器视觉的热点研究问题,是机器智能领域的关键技术,是实现机器人感知或认知能力的基础;而视觉图像中的文本信息是一类特殊的视觉目标,能够在一定程度上反映图像中的重要内容,可以对图像中的内容简练的描述或说明,是理解图像或视频内容高层语义的重要途径;自动定位到图像中的目标区域,并提取出来,经过识别与分析,在机械智能化和其他领域有较好的应用前景。机器视觉图像处理方法是机器视觉技术得以应用的基础和关键。本文以研究视觉图像中文本区域作为目标区域进行检测与处理为实现目标,提高机器视觉相关处理方法研究及应用水平。本文的主要工作为:首先,本文介绍了在机器视觉中常用的图像处理方法:彩色图像灰度化、图像增强技术、基于阈值的图像分割方法、数学形态学的基本运算以及几种常用的边缘检测算子;能够做到熟练运用机器视觉常用的处理方法。然后,本文总结了视觉图像中作为目标区域的文本区域的特征,综述了目前存在的目标区域定位算法,分析比较了算法的优缺点,提出了一种提基于边缘和连通域分析的目标区域定位方法。最后,在对目标区域进行恢复处理时,采用基于纹理修复结构的图像修复技术恢复出被目标区域遮挡的背景区域;并针对纹理修复算法中修复优先级和匹配块搜索方式进行了改进;通过对不同大小图像的实验,结果证明了本文算法可以降低图像的误匹配,提升图像修复速度;最终本文实现了视觉图像中目标区域的自动检测与去除。更多还原

【Abstract】 It’s a hot issue in the study of machine vision about target recognition and processing methods. That is a key technology in the field of machine intelligence. That is also the basis for machine to realize perceptual or cognitive ability. Text information in visual image, is a kind of special visual object, can reflect the important details of the image, and concisely descript or explain the content of image, which is a significant clue to get comprehension of high- level semantics of the image or video. Automatically fix the target region, extract, identify and analyze. That has good application in machine intelligence and other areas.Processing methods of machine vision image is the basis and key contents of application of machine vision technology, The main tasks of this paper are:First, the paper introduces the common knowledge used in machine vision image processing: color images gray processing, image enhancement technology, image segmentation method based on a threshold, the basic operations of mathematical morphology and several commonly used edge detection operators; To achieve skilled in using machine vision commonly used processing methods.And then, it summarizes the text region’s characteristics in visual image, reviews the target region localization algorithm in current research, analyses and compares the advantages and disadvantages of the algorithm, and thereby proposes text localization method based on the edge detection and the analysis of connected domain.At last, in the target region restoration, it adopts to image restoration technology based on the structure of texture restoration to recover the background region shadowed by target regions, and ameliorate the algorithms for texture restoration priority and matching block searching method; after the experiments for different sizes of the image, the results prove that the algorithm this paper proposes can reduce the mismatch of the image, and enhance speed of the image restoration; finally the paper implements the automatic detection and removal of target in visual images.更多还原