{$cfg_webname}
主页 > 电子信息 > 论文 >

基于数字图像处理的织物缺陷检测研究

来源:56doc.com  资料编号:5D21208 资料等级:★★★★★ %E8%B5%84%E6%96%99%E7%BC%96%E5%8F%B7%EF%BC%9A5D21208
资料以网页介绍的为准,下载后不会有水印.资料仅供学习参考之用. 帮助
资料介绍

基于数字图像处理的织物缺陷检测研究(任务书,开题报告,论文13000字)
摘要
本文采用Gabor滤波器来检测常见织物图像中的疵点,目的是分离图像中的缺陷区域,从而实现纺织品缺陷检测的自动化。本研究基于机器视觉在工业检测中的应用,大大提高了生产过程中的检测效率,具有一定的应用价值。
当研究信号的频率特性时,傅里叶变换一直是我们常用的工具,但是傅里叶变换对于频域局部特性的表现能力较差,在纹理图像的解析过程中并没有很好的表现。而Gabor滤波器是一种有效检测织物疵点的工具,完善了傅里叶变换的不足,并且通过对于Gabor滤波器参数的研究,我们可以在时间域和频率域均保持良好的识别精度。
在对织物的图案进行初步研究后,我们可以发现,纺织物图像一般含有一定的纹理特征,而其中的疵点往往聚集在某一区域。对应到频率域,纹理部分对应的是高频部分,而相对集中的疵点则对应的是低频部分,在选取合适的Gabor滤波器参数后,我们可以将图像的纹理部分与缺陷部分分离开来,在经过阈值分割后,将疵点标记出来。
为综合各个方向的纹理特征,本文同时采用了多通道的Gabor滤波器,在此基础上进行图像融合,进一步提高了识别的准确率。
关键字:傅里叶变换;Gabor滤波器;图像融合;阈值分割

Abstract   
In this paper, Gabor filter is used to detect the defects of fabric images. The purpose is to separate the defect area in the image, so as to realize the automatic realization of textile defect detection. This study is based on the application of machine vision in industrial inspection, which greatly improves the detection efficiency in the production process and has a certain application value.
When studying the frequency characteristics including the signal, the Fourier transform has always been our helpful tool, but the Fourier transform has poor performance in the local characteristics of the frequency domain, and does not perform well in the process of analyzing the texture image. The Gabor filter is a tool for effectively detecting fabric defects. The Gabor filter has better resolution in the spatial domain and frequency domain. By studying the Gabor filter parameters, we can maintain both the time domain and the frequency domain with good recognition accuracy.
After a preliminary study of the pattern of the fabric, we can find that the textile image generally contains certain texture features, and the defects are often concentrated in a certain area. Corresponding to the frequency domain, the texture part corresponds to the high-frequency part, and the relatively concentrated defect corresponds to the low-frequency part. After selecting the appropriate Gabor filter parameters, we can separate the texture part and the defect part of the image. After the threshold segmentation, the defects are marked.
In order to synthesize the texture features in all directions, a multi-channel Gabor filter is adopted in this paper. Based on this, image fusion is performed to further improve the recognition accuracy.
Key Words:Fourier Transform; Gabor filter; Image fusion; Threshold segmentation

目录
摘要    I
Abstract    II
第1章绪论    1
1.1 课题背景及意义    1
1.2 国内外研究现状    1
1.2.1 统计法    2
1.2.2 频域法    2
1.2.3 模型法    3
1.3 研究内容与预期目标    4
第2章傅里叶变换    5
2.1 傅里叶变换原理    5
2.1.1 一维傅里叶变换    5
2.1.2 二维傅里叶变换    5
2.2 简单图像滤波    6
2.3 傅里叶变换的意义及不足    7
第3章 Gabor滤波器    9
3.1 一维Gabor滤波器    9
3.2 二维Gabor滤波器    10
3.3 波长λ参数影响    10
3.4 角度θ参数影响    12
3.5 标准差σ参数影响    13
第4章检测系统设计    16
4.1 OpenCV简介    16
4.2 算法思路综述    16
4.3 获得滤波器参数    17
4.4 多角度滤波与图像融合    18
4.5 阈值分割    19
第5章实验结果及分析    21
5.1 织物疵点描述    21
5.2 实验环境    21
5.3 实验结果展示    22
5.4 实验结果分析    25
第6章结论    26
参考文献    27
致谢    29

推荐资料