数字图像边缘的快速提取算法研究  

Study on the Image Edge Extracting Algorithm for Machine Vision Measuring System

作　　者：段振云[1] 单紫薇 赵文辉[1] 杨旭[2] 

机构地区：[1]沈阳工业大学机械工程学院,沈阳110870 [2]中国科学院沈阳自动化研究所,沈阳110016

出　　处：《组合机床与自动化加工技术》2017年第12期12-14,20,共4页Modular Machine Tool & Automatic Manufacturing Technique

基　　金：十二五国家科技支撑计划(2014BAF08B01)

摘　　要：在机器视觉测量系统中,图像边缘提取是进行后续坐标计算的必要前提,为了提高图像边缘提取的速度与精度,提出一种新的数字图像边缘提取算法。用高斯滤波对图像进行平滑处理,根据八邻域的位置关系,经过灰度值的比较确定图像边缘。通过与常用的边缘检测算子Sobel算子和Canny算子的边缘提取效果的实验比较,该算法运算量小,速度快,且能直接提取出图像连续的单像素边缘,优于传统的图像边缘提取算法,实用性强。Image edge extraction is the necessary premise in the machine vision measuring system. In order to improve the speed and precision of image edge extraction,this paper proposes a newdigital image edge extraction algorithm. By Gaussian filter to smooth the image,according to the position of 8 neighborhood relations,through the comparison of grey value image edge is determined. With the commonly used edge detection operator Sobel operator and Canny operator edge detection result comparison,the experimental results showthat the operator does not need to calculate the derivative,small computational complexity,fast speed,it can directly extract image continuous single pixels edge,superior to the traditional image edge extraction algorithm,strong practicability.

关 键 词：机器视觉 边缘提取 高斯滤波 八邻域 

分 类 号：TH122[机械工程—机械设计及理论] TG506[金属学及工艺—金属切削加工及机床]