一种新的刀具磨损面图像边界提取方法  

Extraction Method for Image Boundary Based on Tool Wear

作　　者：杨吟飞[1] 李亮[1] 何宁[1] 

机构地区：[1]南京航空航天大学机电学院,南京210016

出　　处：《南京航空航天大学学报》2007年第6期786-789,共4页Journal of Nanjing University of Aeronautics & Astronautics

基　　金：国家自然科学基金(50175051)资助项目

摘　　要：介绍了一种新的基于机器视觉的磨损区域几何尺寸的测量方法。该方法在人工指定含有边缘的某一微小区域后,通过一系列窗口对磨损区域的边缘进行跟踪以实现对磨损区域的几何描述。所有的图像处理仅在含有边缘的窗口中进行,减少了噪声对边缘提取的干扰,也减少了数据处理量,因此具有良好的鲁棒性和较高的处理效率。该方法的测量精度与图像分辨率有关,本文的测量精度优于5μm。A method based on machine vision is developed to measure the wearing area, Firstly a small area pointed by the user is used to provide the start point for tracking and extraction. Then a series of windows are used to track and extract the edges of wearing area and the geometrical description of wearing area is acquired. As all image processing is carried out within those windows, the method avoids the disturbance from the noise and reduces the quantity of processed data, and therefore obtaining good robustness and high efficiency. Meanwhile the method enhances the measurement precision to the range of pixels （less than 5μm）.

关 键 词：机器视觉 图像处理 跟踪提取 刀具磨损 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]