Homogeneous Region Segmentation And Real-time Pinpoint Location For SAR Image

Synthetic aperture radar(SAR)technique has been paid great attention to on earth science research for its ability to obtain images under any weather condition anytime while covering large area.With the development of SAR technology and the application expansion,the quality and quantity of the acquired SAR image have been greatly improved.Automatic interpretation for SAR image becomes a severe bottleneck to its application.SAR image homogeneous region segmentation and real-time location are two important research directions in the field of SAR image interpretation that includes many aspects.Due to SAR imaging principle,it is difficult to apply the processing methods for optical image on the SAR image with speckle noise and geometry distortion.Combining SAR imaging principle with the processing method of optical image,this article studies on the SAR image homogeneous region segmentation algorithm and real-time location algorithm.Based on the characteristics of SAR image,this work can be seperated into several parts described below: 1)After a brief introduction to several common SAR image statistical models,these statistical models will be evaluated by some parameter estimation methods and evaluation criteria of modeling precision.2)SAR image homogeneous region segmentation algorithm.The proposed algorithm includes two steps for segmentation,a rough segmentation and a fine segmentation.In the rough segmentation,SAR image will be roughly segmented into several homogeneous regions.In the fine segmentation,a pixel-level segmentation is applied,aiming to obtain a more smooth segmentation edge.3)A real-time location algorithm based on the fusion of digital elevation model(DEM)and SAR image is proposed.After using DEM to simulate SAR images,SAR image localization is achieved by matching the real SAR image corresponding simulated SAR image.