摘要
影像分割是面向对象的分类思想应用于遥感影像信息自动提取的基础,纹理是影像的基本特征,是影像分析、理解和识别的重要信息,纹理特性的有效表达和抽取,是基于纹理影像分割的前提。本文系统概述了近年来各类文献中使用频率较高的基于纹理的遥感影像分割方法,并以基于统计的方法、基于纹理结构的方法、基于模型的方法和基于空间/频率的方法四种基本类型为主线,对每一类分割方法的特点进行了分析和总结,在此基础上指出了基于纹理特征遥感影像分割的研究趋势。
Image segmentation is the basis of object-oriented classification approach which applies in automatically extracting information from remotely sensed imagery.Texture is the basic feature of imagery and important information to analyze,understand and recognize imagery.Effective representation and extraction of image texture is indispensable in a texture-based segmentation of imagery.This article gave an overview of the texture-based segmentation methods for remotely sensed imagery,which are frequently utilized in scientific literature.Each of the methods was systematically analyzed and summarized which include statistics-based,texture-structure-based,model-based and spatial/frequency-based.The research tendency of this field was proposed based on the above mentioned study conclusion.
出处
《测绘科学》
CSCD
北大核心
2010年第5期31-34,共4页
Science of Surveying and Mapping
基金
国家973计划项目(2006CB708306)
国家自然科学基金项目(40871206)
关键词
遥感影像
分割
纹理
小波变换
remotely sensed imagery
segmentation
texture
wavelet transform
