The Study Of SAR Image Speckle Filter

With its ability to image the Earth's surface in nearly all weather conditions, together with its high spatial resolution, SAR has shown its potential for classifying and monitoring geophysical parameters both locally and globally. However, speckle in SAR images disturbs its applications. The dissertation is with focus on speckle filtering.The study of speckle filters and the problem encountered by speckle filters are introduced in the first chapter. The second chapter is concerned with how speckle originate. In the third chapter we introduce the fundamental properties of SAR images. This chapter is the foundation of SAR image processing including principles of SAR image formation, the nature of information in SAR images, the statistical properties of SAR images and the data models. A new signal processing method named EMD is exploited to process SAR images in the fourth chapter. We first research how speckle in SAR images influences the modes obtained from SAR images by EMD. We found that speckle in SAR images influences the first and the second mode. The speckle filter and multiscale edge detection technique are proposed respectively based on the effects of speckle on the modes. In the fifth chapter, we study the speckle filters in context of estimate theory, we find that the problem of the speckle filters is edge preservation. A method named adaptive multi-look processing is proposed, it reduces speckle while preserving the edges. At last, we evaluate the proposed methods using the simulated SAR images, Chinese L-band airborne SAR images, SIR-C/X-SAR images and ERS-1 images. The sixth chapter is conclusion. The innovations in the dissertation as bellow:1. The definition of edge in SAR images is confused, the definition of edge is important for the study of edge preservation speckle filters. In this paper, we define an edge in an image as an area which intensity is sharply variation. The edge preservation is to preserve the gradient of edge areas. Based on edge definition, we develop a method to evaluate speckle filters in tern of edge preservation: compute ratio between the sum of gradient absolute value in the edge area of filtered image and original.2. The prime problem of speckle filters is edge preservation, the traditional speckle filters cannot effectively preserve edge, to preserve the edge, a practical and reliable way can accomplish this mission, say, smoothing is running within a width of single pixel chain along edge direction.3. EMD is a novel signal processing method, the study prove that this method can be used to processing SAR images, speckle in SAR images influences the first and the second mode.4. EMD can decompose signal in different scale components, An effective algorithm of multiscale technique for detecting edges in SAR images is proposed based on the character of EMD.5. A speckle filter based on EMD is proposed, it smooth along edge direction in edge area and compute mean of square area in homogenous region, therefore this method can reduce speckle while preserve edge in SAR images.6. The speckle filters in context of estimate theory is essentially to look for homogenous region, then estimate the true image, but the traditional speckle filter only detect the limited shape homogenous region, a speckle filter named adaptive multi-look processing which can detect any possible shape homogenous region is proposed.