In this paper

we present a new InSAR phase filtering method based on optimal integration. The algorithms can preserve very well the phase details while at the same time smoothing out the noise. Firstly

we use statistical method to determine the number of windows used for the filtering. It is an empirical constant associated with coherence. Secondly

eight linear directional windows are singled out

within each window a filtering is performed

and at the same time the mean coherence for each window is calculated. The proposed filtering will linearly combine a certain number （which has been determined in the first step） of the eight directional windows. However

directional windows with smaller filtering standard deviation will be given priority. Finally

the new phase value is calculated in terms of the weighted mean value of chosen linear windows. In this step

optimal integration is used to determine the weight of each directional window. The proposed filter is adaptively implemented by altering the number of the linear windows selected for filtering according to the coherence. Strategy of using both linear windows and optimal integration makes great difference in the filtering and achieve a good tradeoff between phase noise suppressing and signal preserving. Experimental results with both simulated and real data sets show that the new filter reduces the noise effectively while still minimizing the loss of signals.