文献类型：会议论文

英文题名：Segmentation and classification of PolSAR data using spectral graph partitioning

作者：Zhao, Lei[1] Chen, Erxue[1]

第一作者：赵磊

机构：[1] Institute of Forest Resources and Information Technology, Chinese Academy of Forestry, China

会议论文集：MIPPR 2013: Remote Sensing Image Processing, Geographic Information Systems, and Other Applications

会议日期：October 26, 2013 - October 27, 2013

会议地点：Wuhan, China

语种：英文

外文关键词：Graph structures - Graph theory - Image classification - Image segmentation - Information use - Remote sensing - Synthetic aperture radar

年份：2013

摘要：Polar metric synthetic aperture radar (PolSAR) image classification is an important technique in the remote sensing area, has been deeply studied for a couple of decades. This paper proposes a new approach for segmentation and classification of PolSAR datain two steps. First, segmentation is performed based on spectral graph partitioning using edge information. Graph partitioning process is completed using the normalized cut criterion. Then, classification is performed based on the object level. We use Cloude and Pottiera€ Ys method to initially classify the PolSAR image. The initial classification map defines training sets for classification based on the Wishart distribution. The advantages of this method are the automated classification, and the interpretation of each class based on the regiona€Ys scattering mechanism. We tested this object-based analysis on our study area. It showed that this result well overcome the pepper-sault phenomenon appearing in the one using traditional pixel-based method, providing robust performance and the results more understandable and easier for further analyses. ? 2013 SPIE.