基于分形和Gabor滤波的高分辨率遥感影像分割  

Segmentation of High Resolution Remote Sensing Images Based on Fractals and Image Filtering

作　　者：何文斌[1] 梁利姣[2] 

机构地区：[1]东莞理工学院计算机学院,广东东莞523808 [2]东莞理工学院城市学院,广东东莞523106

出　　处：《东莞理工学院学报》2011年第3期53-57,共5页Journal of Dongguan University of Technology

摘　　要：在城市规划、土地利用等应用中常需要对高分率遥感影像进行分割,而传统的滤波器组分解算法把每个波段的能量作为纹理特征进行分割,存在分割精度低,对噪声、灰度变换敏感等缺点。本文提出一种基于图像预滤波和分形特征的纹理特征提取算法,并对提取的特征向量集用K-m eans算法进行聚类。对天津滨海新区的Qu ickB ird影像的分割实验表明新算法能有效提高分割精度。It is very necessary to segment high resolution remote sensing image in urban planning and land use.Traditionally,filter bank decomposition methods consider the energy in each band as the textural feature,which is sensitive to noise and image intensity.Thus,the segmentation precision is low in some cases.This paper describes a new approach to the segmentation of textured gray-scale images based on image pre-filtering and fractal features.The fractal feature set is clustered by means of K-means.The test of QuickBird image from the new coastal region of tianjin demonstrates that the new method can effectively improve classification accuracy.

关 键 词：纹理分割 GABOR滤波器 分形特征 K均值 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]