基于机器视觉测量的齿轮图像边界提取算法研究

发布时间：2017-12-28 21:08

本文关键词：基于机器视觉测量的齿轮图像边界提取算法研究　出处：《沈阳工业大学》2017年硕士论文　论文类型：学位论文

更多相关文章： 机器视觉 边缘提取 八邻域 亚像素

【摘要】：齿轮是传递运动和动力的基本组成部分,它的制造精度直接决定了其工作状况的好坏。于是,齿轮的测量工作便成为研究与生产齿轮的过程中极为关键的环节。传统的接触式齿轮测量方法具有精度低、工作量繁重等弊端,于是出现了基于机器视觉的测量方法。其中图像的边缘提取是后续图像处理、求取齿轮参数的前提,所以,本文以基于机器视觉的齿轮尺寸测量为研究背景,以机器视觉技术和图像处理技术为理论依据,提出了针对背光源直齿圆柱齿轮的图像边缘提取算法,课题的主要工作如下:第一,论述了齿轮精密测量的重要性,以及机器视觉技术在齿轮检测中的可行性与必要性,而齿轮图像的边缘检测又是齿轮测量的必要前提,于是,通过对边缘的研究可知,理想边缘主要包括阶跃型和屋脊型两种。然后对齿轮图像的边缘类型进行分析。第二,对几种经典的像素级边缘检测算法进行了研究,并将这些算子作为对比实验。通过对实验结果结合理论基础的研究,总结这些算法的优缺点。基于以上分析,提出基于八邻域搜索的像素级边缘提取算法,以像素八邻域的位置关系为基础,通过高斯滤波对图像进行平滑处理,根据比较目标像素与八邻域像素灰度值的大小关系实现像素级边缘的提取。第三,研究了现有的亚像素级边缘提取算法,包括拟合法、插值法、矩法。通过对现有亚像素级边缘提取算法的研究,提出基于双线性插值与高斯曲线拟合相结合的亚像素级边缘提取算法,实验表明本文提出的算法不但保证了边缘精度,还减少了运算时间。
[Abstract]:Gear is the basic component of transmission motion and power, and its manufacturing precision directly determines its working condition. Therefore, the measurement of gear has become a key link in the study and production of gear. The traditional contact gear measurement method has the disadvantages of low precision and heavy workload, so the measurement method based on machine vision appears. So the image edge extraction is the premise for the subsequent image processing, and take the gear parameters, based on the measurement of gear size based on machine vision as the research background, the machine vision technology and image processing technology as the theoretical basis of image edge extraction algorithm is proposed for the backlight of spur gear, the main subject of the work as follows: first, discusses the importance and feasibility of the gear precision measurement, machine vision technology in detection of gear and gear and the necessity of image edge detection is the necessary premise, gear measurement result, through the research to the edge of the ideal edge including step and roof two. Then the type of the edge of the gear image is analyzed. Second, several classical pixel level edge detection algorithms are studied, and these operators are used as contrast experiments. The advantages and disadvantages of these algorithms are summarized through the research on the theoretical basis of the experimental results combined with the theoretical basis. Based on the above analysis, put forward eight pixel edge extraction algorithm based on neighborhood search, in position between the eight pixel neighborhood based, through the Gauss filter to smooth the image, according to the relationship between pixel size extraction target pixel and neighborhood values to achieve eight pixel edge. Third, the existing sub pixel edge extraction algorithms are studied, including the fitting method, the interpolation method and the moment method. Through the research of the existing sub-pixel edge extraction algorithm, a sub-pixel edge extraction algorithm based on bilinear interpolation and Gauss curve fitting is proposed. Experiments show that the algorithm proposed in this paper not only guarantees the edge accuracy, but also reduces the computation time.
【学位授予单位】：沈阳工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TH132.41;TP391.41

【参考文献】