基于机器视觉的构件表面缺陷特征提取

电子机电论文编号:JD875  附开题报告,外文翻译,论文字数:18664,页数:40

摘要 
目次 
1 绪论  1
1.1 引言  1
1.2 课题的研究背景及意义  1
1.3 论文的结构  1
2 机器视觉的简介  2
2.1 机器视觉技术  2
2.1.1 机器视觉的构成  2
2.1.2 机器视觉的特点  2
2.1.3 机器视觉系统的组成  3
2.2 机器视觉的应用  5
2.2.1 机器视觉应用的现状  5
2.2.2 机器视觉检测应用  7
2.3 机器视觉的分类  7
2.3.1 定量检测  7
2.3.2 定性检测 8
3 图像处理软件介绍  9
3.1 Matlab系统软件简介  9
3.1.1 Matlab系统包括五个主要部分  9
3.1.2 Matlab的系统开发环境  9
3.2 Matlab图象处理命令  12
4 图像处理基本算法  13
4.1 图像灰度变换  13
4.1.1 灰度直方图  14
4.1.2 灰度的线性变换  14
4.1.3 灰度的阈值变换  14
4.1.4 灰度的窗口变换  15
4.1.5 灰度拉伸  15
4.1.6 灰度均衡  16
4.2 图像增强  16
4.3 图像的分割  17
4.3.1 阈值分割  18
4.3.2 区域生长法分割  20
5 表面缺陷特征的检测的结果  23
5.1 图像的预处理  23
5.2 图像的分割  25
5.2.1 迭代阈值法分割图像 25
5.2.2大津法分割图像  28
5.2.3 区域生长法分割图像  30
6 总结  34
参考文献  35
致谢  36
附录  37

摘要:构件的表面缺陷是影响构件质量的重要因素之一，对它进行检测将达到质量控制的目的,鉴于这样的目的，本设计介绍的是基于机器视觉的构件的表面缺陷特征提取。
本文在现有技术基础上，通过了大量的实验和图像算法仿真，以现有的图像处理理论为基础，设计了适合本设计所涉及的表面缺陷检测的算法，具体的工作如下：
(1).通过确当的途径获取缺陷图像，并以图像进行灰度变换、中值滤波对图像进行增强处理，为下一步图像的进一步处理做前期的准备。
(2).对图像表面缺陷进行研究，这一过程本文采用了大津法、区域生长法、迭代法这三种分割方法，并且认真研究设计了它们的算法。
(3).把各个缺陷图片的分割算法在MATLAB中进行算法仿真，提取正确的实验结果，并且重复多次实验，确保结果的精确性。
关键词:机器视觉；表面缺陷；图像增强；图像分割

The Extraction of Surface Defect Characteristics of Component Parts Based on Machine Vision

Abstract: The surface defects of component parts are an important factor affecting the quality of the component parts, so defect detection is a means to achieve the object of quality control; this article presents the extraction of surface defect characteristics of component parts based on machine vision. Through large amount of experiments and the simulation with image algorithm, and based on the existing image process theory, this article designs an algorithm that is suitable to the surface defect detection related to this design on the basis of existing technology. The specific process is as follows:
    (1) To obtain defect images through appropriate approach and conduct gray-level transformation and the images are enhancement processed with median filtering, so as to get ready for further processing.
(2) To analyze the surface defects of the images using three segmentation methods, i.e. OTSU method, region-growing method, and iteration method, and their algorithms are studied and designed.
(3) To carry out algorithm simulation in MATLAB to simulate the segmentation algorithm of the defect images, and correct experimental results are extracted, and it is repeated for several times so as to ensure the accuracy.
Key words: machine vision; surface defect; image enhancement; image segmentation