SAR图像的自动分割方法研究Automatic-电子与信息学报.PDFVIP

第28 卷第5 期 电 子 与 信 息 学 报 Vol.28No.5 2006 年5 月 Journal of Electronics Information Technology May 2006 SAR 图像的自动分割方法研究 李 映 史勤峰 张艳宁 赵荣椿 (西北工业大学计算机学院 西安 710072) 摘 要 由于存在相干斑噪声的影响，给SAR 图像分割造成很大的困难，该文提出了一种SAR 图像的自动分割方 法。首先在特征提取阶段，通过计算小波能量提取纹理信息，用邻域统计量提取灰度信息，用保边缘平均灰度提 取边缘信息，以确保边缘准确。然后提出一种改进的完全无监督的聚类算法进行图像分割，该算法可以自动确定 分割的类型数目。由于该方法充分考虑了SAR 图像的纹理、灰度和边缘信息，因而极大地提高了其最终分割性能。 实验结果证明了该方法的有效性。 关键词 SAR 图像，特征提取，无监督聚类，分割 中图分类号：TN957.52 文献标识码：A 文章编号：1009-5896(2006)05-0932-04 Automatic Segmentation for Synthetic Aperture Radar Images Li Ying Shi Qing-feng Zhang Yan-ning Zhao Rong-chun (School of Computer, Northwest Polytechnical University, Xian 710072, China) Abstract The multiplicative nature of the speckle noise in SAR images is a big problem in SAR image segmentation. A novel method for automatic segmentation of SAR images is proposed. The wavelet energy is used to extract texture features, the regional statistics is used to extract gray-level features and the edge preserving mean of gray-level features is used to ensure the accuracy of classification of pixels near to the edge. Three representative kinds of features of SAR image are extracted, so the segmentation performance is enhanced. Besides, an improved unsupervised clustering algorithm is proposed for image segmentation, which can determine the number of classes automatically. Segmentation results on real SAR image demonstrate the effectiveness of the proposed method. Key words SAR image, Feature extraction, Unsupervised clustering, Segmentation 1 引言