Abstract: G-image representing region homogeneity of Synthetic Aperture Radar(SAR) image is presented. A region growing method is used to segment the image roughly based on the G-image. During the mergence of the rough-segmented image, for the SAR image which has rich direction and texture information, NonSampled Contourlet Transform(NSCT) is employed to extract regional feature and a new method is used to merge regions. Simulation results show the effectivity of the segmentation algorithm for SAR image.

Key words: Synthetic Aperture Radar(SAR), image segmentation, region homogeneity, G-image

摘要： 提出反映合成孔径雷达(SAR)图像区域一致性的G值图，在此基础上运用区域生长方法对图像进行粗分割。在对粗分图像进行区域合并的过程中，对于含有丰富方向和纹理信息的SAR图像，使用非下采样Contourlet变换(NSCT)进行区域特征提取，并采用一种新方法进行区域合并。仿真结果证明了SAR图像分割算法的有效性。

关键词: 合成孔径雷达, 图像分割, 区域一致性, G值图

CLC Number: 

• TP391