New approach for improving Mutual Information in POLSAR image registration

Aiming at the problem that information source of similarity measure is limited to image intensity usually in multi-polarization SAR registration, we propose an improved Mutual Information(MI) based on the multiple correlation information between the same polarized channels (HH, VV) in multi-polarization SAR images. In this method, by analyzing the correlation characteristics of three polarized channels in multi-polarization SAR, and using channel correlation information extracted from polarimetric covariance matrix, we compute MI in multi-polarization SAR to improve the similarity measure. Finally, the simulation results show that, in the multi-polarization SAR image registration, similarity measure information sources affect the performance of similarity measure. We compare the similarity measures with different information sources in curve sharpness and noise immunity. The results prove that the measure based on the correlation of the same polarized channels for multi-polarization SAR performs excellently.