Abstract: An improved matching method based on Scale Invariant Features Transform (SIFT) algorithm is proposed in this paper. The Infinite Symmetric Exponential Filter (ISEF) algorithm is adopted to reduce speckle noise before computation of the scale space pyramid. SIFT algorithm is utilized to detect the feature points and skip the first scale-space octave to reduce processing time. And then false matches are deleted in the Euclidean space. Experiments show that the proposed method increases the number of the features and improves the robustness. The match accuracy could meet the requirement of sub-pixel matching and the processing time has been cut by 60%. Finally, earthquake change detection is implemented from ALOS PALSAR images, and the building damage information detected is consistent with the results from high spatial resolution aerial image.

Key words: ALOS PALSAR, Image Registration, the 2010 Yushu Earthquake