光纤图像的定位二值化  

Positioning Segmentation for Fiber Image

作　　者：王学城[1] 

机构地区：[1]华东理工大学信息学院,上海200237

出　　处：《现代计算机》2004年第10期13-16,共4页Modern Computer

摘　　要：本文介绍一种用于光纤图像分割的定位二值化方法。本方法根据光纤图像中光纤的位置和大小都固定的特点,首先通过两个方向的倾斜投影确定光纤中心的位置,接着由最小误差法确定是否存在光纤的初判阈值,然后根据光纤的大小选择合适的二值化区域,再在所定区域内进行二值化,从而将图像中所有光纤完整地、正确地分割出来。本方法避免了整体阈值法动态范围小和传统自适应阈值法固定分区所造成的误判。In this paper a new approach called projection positioning segmentation is set forth to realize the image segmentation.According to the characters of the fibers whose position and size are constant in the picture,firstly this new approach locates the center of every fiber through slope projection in two directions,and confirms the original threshold using Minimum to judge if fibers exist or not,then selects the suitable size of segmentation region referring to the size of the fibers,finally segments the whole fibers in these selected regions thoroughly and accurately.This method avoids the small dynamic range in global thres-holding and misjudgement caused by fixed divided regions in adaptive thresholding.

关 键 词：二值化方法 图像分割 光纤 自适应阈值 最小误差法 分区 大小 动态范围 阈值法 误判 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]