首页 | 本学科首页   官方微博 | 高级检索  
   检索      

流域变换的串行与并行策略研究
引用本文:周海芳,蒋艳凰,杨学军.流域变换的串行与并行策略研究[J].国防科技大学学报,2002,24(6):71-76 ,90.
作者姓名:周海芳  蒋艳凰  杨学军
作者单位:国防科技大学计算机学院,湖南,长沙,410073
基金项目:国家杰出青年科学基金项目资助(69825104)
摘    要:流域变换是数字形态学中用于图像分割的一种经典方法,其并行化问题成为近年来研究的重点。首先给出了流域变换的数学模型,并归纳列举了几种离散情况下的形式化定义;其次分类总结了近年来流域变换串行算法研究的新进展,从而在此基础上重点讨论了相应的并行化策略。详细分析了设计并行流域算法需要考虑的几个问题;并比较评价了现有并行算法的性能特点,得出了一些结论;最后提出了有待进一步研究的问题。

关 键 词:流域变换  图像分割  积水盆  分水岭  并行算法  分布存储  域分解  加速比
文章编号:1001-2486(2002)06-0071-06
收稿时间:5/8/2002 12:00:00 AM
修稿时间:2002年5月8日

Research on Serial and Parallel Strategies of Watershed Transform
ZHOU Haifang,JIANG Yanhuang and YANG Xuejun.Research on Serial and Parallel Strategies of Watershed Transform[J].Journal of National University of Defense Technology,2002,24(6):71-76 ,90.
Authors:ZHOU Haifang  JIANG Yanhuang and YANG Xuejun
Institution:College of Computer, National Univ. of Defense Technology, Changsha 410073,China;College of Computer, National Univ. of Defense Technology, Changsha 410073,China;College of Computer, National Univ. of Defense Technology, Changsha 410073,China
Abstract:Watershed transform is a classical method of image segmentation in mathematical morphology, and its parallelism is an important problem. This paper starts with a mathematical model of watershed transform in topography, followed by its several formal definitions for digital case. Then some new serial watershed algorithms proposed in recent years are classified and analyzed, based on which their related parallel strategies are discussed in detail. Then we analyze some issues, which should be considered when designing parallel watershed algorithm, and give an evaluation of existing parallel algorithms, and draw some conclusions. Finally we point out the problems and challenges of future research.
Keywords:watershed  transform  image segmentation  catchment basins  watershed line  parallel algorithms  distributed memory  domain decomposition  speedup  
本文献已被 CNKI 万方数据 等数据库收录!
点击此处可从《国防科技大学学报》浏览原始摘要信息
点击此处可从《国防科技大学学报》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号