基于图像纹理分析的车削表面粗糙度检测
发布时间:2018-05-11 07:57
本文选题:表面粗糙度 + 纹理分析 ; 参考:《沈阳建筑大学》2015年硕士论文
【摘要】:表面粗糙度是评定零件表面质量的重要指标,它直接影响到零件的使用性能、安全和寿命,尤其对于具有特殊功能(密封、相对移动等)的零件更是如此。因此,快速准确、无损地检测零件工作表面的粗糙度对于零件的正常使用性能和系统的安全性具有重要意义。本文基于计算机视觉理论,采用图像纹理分析方法,实现了对车削工件表面粗糙度的非接触式无损检测。本文研究的主要内容为:(1)搭建了以VHX-1000型超景深三维显微系统为核心的测量系统硬件平台,实现了车削工件表面清晰显微图像的获取。(2)基于灰度共生矩阵方法(GLCM),对车削工件表面图像进行了纹理统计分析和特征提取。首先,根据车削表面图像特征,确定了灰度共生矩阵的最优构造参数,为后续的特征提取提供准确的数据。其次,根据车削原理和车削痕迹图像特征,提取并分析了基于灰度共生矩阵的14个统计特征参数,分别为角二阶矩、对比度、相关性、差分矩、逆差分矩、和平均、和方差、和熵、熵、差方差、差熵、相关信息测度Ⅰ、相关信息测度Ⅱ和最大相关系数。建立了灰度共生矩阵统计特征参数和对应的表面粗糙度的关系模型数据库。(3)采用多元回归分析方法构建了车削工件表面粗糙度的检测模型,实现了车削工件表面粗糙度的定量计算。分别建立了多元线性回归检测模型与多元非线性回归检测模型,拟合了工件表面纹理特征参数与工件表面粗糙度评定参数Ra映射关系的数学表达式。通过测试样本检验,两种多元回归分析方法均具有较高的检测精度,能够满足表面粗糙度测量的精度要求。实验表明,非线性多元回归检测模型检测精度优于线性多元回归检测模型。(4)采用BP神经网络算法,建立了车削表面粗糙度的检测模型。以车削表面图像纹理特征参数为输入量,对应的表面粗糙度评定值Ra为期望输出,构建了神经网络检测模型。实验结果表明,BP神经网络检测模型具有较高的检测精度,且其检测精度高于多元回归检测模型。(5)以MARLAB软件为开发平台,完成了车削表面粗糙度检测系统的软件设计,并开发了图形用户界面。最后,应用该检测系统对检测模型进行了实验研究,并将该检测结果与传统探针式测量结果进行了对比分析。结果表明,本文检测模型的检测平均误差率均在允许范围内,满足测量要求,可以实现车削表面粗糙度的快速准确、无损检测。
[Abstract]:Surface roughness is an important index to evaluate the surface quality of parts, which directly affects the performance, safety and life of parts, especially for parts with special functions (seal, relative movement, etc.). Therefore, it is very important to detect the roughness of the working surface quickly, accurately and nondestructive for the normal performance of the parts and the safety of the system. Based on the theory of computer vision and the method of image texture analysis, the non-contact nondestructive testing of the surface roughness of turning workpiece is realized in this paper. The main content of this paper is: (1) the hardware platform of measurement system based on VHX-1000 type hyper-depth of field 3D microscope system is set up. Based on the method of gray level co-occurrence matrix (GLCM), texture statistic analysis and feature extraction of turning workpiece surface image are carried out. Firstly, according to the feature of turning surface image, the optimal construction parameters of gray level co-occurrence matrix are determined, which can provide accurate data for the subsequent feature extraction. Secondly, according to turning principle and turning trace image feature, 14 statistical characteristic parameters based on gray level co-occurrence matrix are extracted and analyzed, which are angular second order moment, contrast, correlation, differential moment, deficit moment, average, and variance. Sum entropy, differential variance, differential entropy, correlation information measure 鈪,
本文编号:1873096
本文链接:https://www.wllwen.com/kejilunwen/jinshugongy/1873096.html