摘要本文研究了基于机器视觉的缺陷检测系统。首先对可能存在的缺陷进行分类并简单描述了缺陷检测的基本流程。再把图像灰度化以及去噪的预处理方法进行了介绍,分别比较了均值滤波、中值滤波以及高斯低通滤波的去噪效果,最后选用了加权平均值法来进行图像灰度化处理以及中值滤波来进行图像去噪。接着简要地说明了目前主要的配准算法,包括基于边缘特征的配准算法、SIFT算法配准、SURF算法配准的简单原理。基于边缘的图像配准由于无法提取完整的边缘信息,会使得配准的效果不理想,所以不适用于本文的缺陷检测方法。基于SIFT的特征点匹配则会产生不必要或是错误的匹配特征点,所以也存在一定的局限性。在SURF的算法基础上,采用欧氏距离进行特征点检测,把匹配的特征点用RANSC、设置点对间的连线长度和斜率的阈值来对匹配的特征点对进行筛选,接着计算出刚性变换矩阵,用双线性插值法进行图像配准。最后本文讲述了MATLAB的GUI界面的设计流程和部分功能介绍。83128
毕业论文关键词:机器视觉、缺陷检测、预处理、图像配准、边缘检测、SIFT、SURF、GUI设计
毕业设计说明书外文摘要
Title The Application in the Defect Defection about Machine Vision
Abstract The dissertation studies the defect detection system based on machine vision。 First we classify the possible defects and describe the basic process of defect detection briefly。 Then the pre-processing methods about gray image and denoising is introduced。 We compare the mean filter, median filter and Gaussian low-pass filter。 Finally the weighted average method for gray image processing and a median filter is selected for image denoising。 And we describe the principles of the current main image registration methods including the registration methods based on the edges, SIFT and SURF。 The registration methods based on the edges will make the registration result to be not satisfying because the full edge information is difficult to extract。 The SIFT also has some disadvantage since it will produce unnecessary or wrong matching points。 Based on SURF algorithm, we use Euclidean distance to defect the feature points。 Then we select the feature points with RANSC, setting the threshold of the length and slope。 Next we calculate the rigidity transformation matrix and use bilinear interpolation for image registration。 At last, this dissertation introduces the process of MATLAB GUI interface design and some main function。
Keywords:Machine vision, defect defection, pre-processing, image registration, edge defection, SIFT, SURF, GUI design
目 次
1 绪论 1
1。1 研究的背景及意义 1
1。3 本文主要内容及结构安排 5
2 连接器表面缺陷分类及缺陷检测基本流程 6
2。1 连接器表面缺陷分类 6
2。2 连接器表面缺陷检测基本流程介绍 6
2。3 缺陷识别相关图像算法 7
2。4 缺陷识别方法 8
2。5 本章小结 10
3 图像预处理 11
3。1 图像灰度化 11
3。2 图像去噪