2D PHASE UNWRAPPING BY FUSION CUT-LINE AND QUALITY-MAPS Faiza HOCINE, Sawssen BELHADJ-AISSA, Mohamed FEKIR, Aichouche BELHADJ-AISSA University of Sciences and Technology Houari Boumediene (USTHB). Laboratory of Image Processing and Radiation, Faculty of Electronic and Computer Science. BP. 32, El Alia, Bab Ezzouar 16011, Algiers, Algeria, Email: faiza_ho@yahoo.fr, h.belhadj@lycos.com. ABSTRACT Phase unwrapping is one of the critical step in the feasibility of InSAR processes. It consists in retrieving the absolute phase from modulo 2π. Through this paper, we propose a method of phase unwrapping that combines /fusions the advantages of two methods, namely: The method of phase unwrapping by avoidance of discontinuities, based on creating cut lines and the second method, it is based on local/regional propagation of phase unwrapping by referring to quality maps. The idea is that the quality value between the two target edges of the cut-line is set at zero in the quality maps, then the progress of the unwrapping is performed according to the process of the second method where unwrapping phase by referring to edge quality map. To validate the results, the method has been tested using real interferogram. Keywords: InSAR, phase unwrapping, cut lines, quality maps, edge quality. 1. INTRODUCTION The phase unwrapping consists at lifting the 2π ambiguity in the phase and produce a continuous phase image proportional to the ground measurement, by seeking the integer multiple of 2π for each pixel of the interferogram. In the ideal case where the phase difference between two adjacent pixels does not exceed π (knowing that the Nyquist condition is met) [1], the unwrapping solution is the same regardless of the path of integration. Otherwise, such a problem has no unique solution. To this end, various phase unwrapping methods, techniques and algorithms have been proposed and developed [2] whose main idea is to minimize the phase differences between adjacent pixels: i) either, locally by correcting the phase step by step and imposing certain restrictions on the unwrapping path. [3] [4], ii) or, globally by modeling and optimization of possible solutions [5][6]. Through this paper, we propose a method of phase unwrapping that combines /fusions the advantages of two methods, namely: The method of phase unwrapping by avoidance of discontinuities, based on creating cut lines. It is extremely fast and generally satisfactory, is founded on residue localization and it proceeds to their effect neutralization to avoid them in the unwrapping step. They are created through connecting the positive and negative residues, either by straight lines or by lines following the low values of the coherence image. The disadvantage of this method is that it creates isolated areas "due to sudden cuts in the phase" and the second method, it is based on local/regional propagation of phase unwrapping by referring to quality maps. The quality of a pixel is calculated compared to all these neighbors, unlike other methods of calculation. The fact that each phase unwrapping operation is performed between two adjacent pixels, then instead to interested on the quality of isolated pixel, we look on the pixel edge quality, which is the intersection of two pixels connected horizontally, vertically or on the diagonally direction. Then the edge quality of the pixels is used to guide the path of unwrapping phase. However, this method has a tendency to create an effect of blocks in unwrapped phase space and in the presence of noise or discontinuities of phase, phase unwrapping is itself unresolved and hence quality-guided phase unwrapping is not sufficient. To improve results, our method creates the cut lines prohibiting the path of progress unwrapping along the path of quality calculated from the interferogram. To validate the results, the method has been tested using real interferogram product from tandem pair ERS1/ERS2 taken in the region of Algiers, Algeria. 2. PHASE UNWRAPPING The concept of phase unwrapping consists to seek for a single reconstruction path of a curve or a continuous surface. It progresses through the calculation of the phase difference between two successive pixels in a predefined way. If the difference is less than or greater than ∓, successive pixels in the path are shifted by ±2. The trivial process flow is summarized as follows:  |∆( 1 , 2 |≤ ℎ ∆( 1 , 2 ) = ∆( 1 , 2 )  ∆( 1 , 2 )> ℎ ∆( 1 , 2 ) = ∆( 1 , 2 ) − 2,  ∆( 1 , 2 ) < − ℎ ∆( 1 , 2 ) = ∆( 1 , 2 ) + 2 Where  1 , 2 are two neighboring pixels,  1 represent the reference point,  2 is unwrapped according to  1 , ∆ et ∆ are the difference of the wrapped phase and the unwrapped phase respectively. As a result of this step we obtain a continuous phase as shown in the profile of a line of the unwrapped phase 'Figure 1'. This curve contains the information of the unwrapped phase in respect to the first pixel of the line. The coherence of the _____________________________________ Proc. ‘Fringe 2015 Workshop’, Frascati, Italy 23–27 March 2015 (ESA SP-731, May 2015)