Abstract: In interferometric synthetic aperture radar filtering, denoising has smoothing effect on interference fringes, which results in detail information loss of interference fringes. To solve this problem, this paper proposes an improved adaptive gradient estimation and morphological reconstruction filtering algorithm based on mathematical morphology theory. The algorithm firstly adopts interferogram conversion to protect the phase stripes effectively. Then improving adaptive gradient estimation for better line structure elements protects the phase details. Finally, the paper combines modified morphological opening and closing operation and morphological reconstruction with expansion and erosion operation to carry out filtering steps. Simulation and real data experiments show the proposed algorithm not only has effective performance on denoising, but has fine fidelity to the fringe information of interferograms.