图像分割的分级性研究  

Hierarchy of Image Segmentation

作　　者：沈兰荪[1] 王爱民[1] 

机构地区：[1]北京工业大学信号与信息处理研究室,北京100022

出　　处：《北京工业大学学报》2000年第3期33-38,共6页Journal of Beijing University of Technology

基　　金：国家自然科学基金资助项目!(69971004)

摘　　要：图像分割是指将一幅图像分解为若干互不交选区域的集合，是图像处理与机器视觉的基本问题之一．本文对传统的分割定义进行分析，指出其局限性，并提出了新的广义定义；根据不同的分割目的，提出了“４级分割”的分级性框架；并简要介绍了各级分割的应用和常用算法．该成果有利于进一步对各种分割算法进行有效的客观评价．Image Segmentation denotes a process by which a raw image is partitioned into nonoverlapping regions. The image segmentation problem is treated as a key issue in image processing and machine vision. In this article, a new generalized definition for segmentation is presented. A 4-level scheme for generalized segmentation is constructed according to the hierarchy of segmentation purposes. This article also surveys applications and current methods for the 4-level segmentation tasks. The results are beneficial to further research on the applicability of current segmentation algorithms.

关 键 词：图像分割 计算机视觉 目标识别 可分级性 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]