Abstract: Due to the fact that unwrapping errors in low coherence areas spread easily to other regions in discrepant coherence interferogram, a new phase unwrapping algorithm based on region-recognition and region-expansion method is proposed. After merging optical image of the same area, irrelevant low coherence regions are removed from the SAR image to avoid their errors spreading. During region expanding process, stable and high coherence pixels are selected as growing seeds firstly, and phase unwrapping is guided by coherence information and weighted numbers of unwrapped neighbor pixels. Pixels will join the unwrapped regions after their ambiguity numbers are qualified. The unwrapped regions expand pixel by pixel from high coherence regions to low coherence regions until the boundary is reached. The experiment results on the PALSAR and ENVISAT-ASAR data in Hong Kong region and Iran Bam region demonstrate its validity and advantages over other classic methods