MORPHOLOGICAL FILTERING OF SAR INTERFEROMETRIC IMAGES S. Réjichi (1) , F. Chaabane (1) , F. Tupin (2) , I. Bloch (2) (1) URISA, École Supérieure des Communications de Tunis (SUP’COM), Tunisie. (2) Institut TELECOM, TELECOM ParisTech, CNRS LTCI, France ABSTRACT This paper proposes a new morphological filter for SAR interferograms. It is based on a modified version of alternate sequential filters with reconstruction (MASF), in which the structuring elements are adaptively defined according to the fringe directions. This provides a good fidelity to the fringe information while efficiently removing noise. Another feature of the proposed approach is to apply the filter on the original interferogram and on shifted version, to overcome the wrapping of the phase, and to combine the two results. The proposed filtering technique is then tested on both simulated and real data with different levels of noise. It is also compared to previous techniques according to simplicity and noise reduction. Index Terms— SAR interferometry, morphological filtering, alternate sequential filter, geodesic reconstruction. 1. INTRODUCTION SAR and differential SAR interferometry are operational tools for monitoring surface deformation and topographic profile reconstruction. This technique is based on the fact that the phase difference between two different satellite position SAR signals is directly related to the elevation of a ground surface point. The major problems of SAR interferometry are the phase ambiguity and the temporal and geometric decorrelation [1]. First, as the phase information is included in [0,2π], an unwrapping step is needed to correctly understand the interferometric data. Second, we are talking about geometric decorrelation when the angle of sight i.e. acquisition geometry changes between the two acquisitions dates. Then, SAR images are affected by temporal decorrelation when the considered surface inside the resolution element is changed due to long period cover. These effects are translated to an additional noise which is detrimental to the use of interferometric fringes. Particularly, these disturbances strongly compromise the phase unwrapping process thus the accuracy of the results i.e. the resulting Digital Elevation Model (DEM) or the deformation map. Therefore, a suitable method has to be used to improve the quality of the data. An appropriate filter should be applied on the complex interferometric data, before exploiting the phase. This procedure increases the signal to noise ratio of the phase field and decreases the number of residues. In the literature, several methods have been proposed to reduce the interferometric phase noise [2] [3] [4] and some of them are based on basic mathematical morphology operators [5]. All these techniques, however, involve the loss of image details to a certain extent. In this paper, we introduce a new morphological filter which is based on a modified alternate sequential filter. This filter is applied to the wrapped phase field and takes into account the properties of this type of images (fringes pattern). The performances of this filter are discussed and compared to other existing filters under different conditions. Both simulated and real data have been used for that purpose. This paper is organized as follows. The proposed filtering method of SAR interferometric images is presented in Section 2. In Section 3 the obtained results on both simulated and real data with different levels of noise are discussed. Evaluation results are given for simulated images and the robustness of this filter is also tested by comparing it to other existing filters [2]. 2. MORPHOLOGICAL FILTERING: ADAPTATION TO THE FRINGES PATTERN In this section, we discuss a new strategy taking into account features which characterize the interferometric images i.e. the transition between the fringes and the value of each pixel which in fact reflects an altitude. This strategy estimates the direction of the gradient and includes it in the design of the structuring elements used in morphological filters. According to the method diagram (see Figure1), this section begins by presenting the interferogram shifting step. Secondly, the gradient estimation is explained. Thirdly, the principle of filtering taking into account the gradient direction is exposed. Finally, the recombination of filtered interferograms and the last filtering step are explained. 1581 978-1-4244-9566-5/10/$26.00 ©2010 IEEE IGARSS 2010