Research Of SAR Image Segmentation Based On Variational Method

Digital image processing originated in the 1920s, and then it has extensive applications in the medicine, remote sensing, pattern recognition and etc. These make image processing develop rapidly in the area of modern information processing. Image segmentation is the based of the image processing generally. The purpose of image segmentation is that dividing the image space into many meaningful images to conduct the further processing. To the 1980s, partial differential equation is used in the image segmentation and it formed a complete theoretical system gradually. This opened up a new field of the research on image segmentation.With the widely application of satellite in many fields of society, processing the image of satellite become the important means of getting the information sensed by satellites. In all of the satellite images, the synthetic aperture radar (SAR) satellite images received by ERS-2 are of a large proportion. So the SAR satellite image processing has become a research topic with common concern. In this paper, the segmentation of SAR satellite images is the problem which we want to research.First we discussed the traditional methods of image segmentation and theirs scopes of application in this paper, and then introduced the realization of partial differential method in the field of image segmentation. On the basis of various classical segmentation models, we introduced the principle of the image segmentation with partial differential equation-variational method and its numerical calculation.After knowing about the process of image segmentation based on partial differential equation, we propose a SAR image segmentation technology based on Variational method for the characteristics of SAR satellite images contained lots of noise. The main tasks are:●The process will be divided into two steps as denoising and segmentation, in order to do some work for the characteristics of each step;●In the processing of each step we propose a improved energy model, and do the image processing with the method of partial differential equation;●In the segmentation we use the level set method and fast marching method, in order to save time for evolving and improve efficiency.With the experiments, the method proposed in this paper can make us get a better effect in the segmentation processing of SAR satellite images.