Detection of Information Change on SAR images based on entropy theory

摘要

A novel Information change detection method for SAR images processing was proposed, which is based on the entropy theory. The entropy image is an alternative image for Information change detection comparing with the coherence map. The entropy is calculated pixel by pixel using a linear combination of the images crosscovariance matrix eigenvalues. The relationship between the entropy and coherence images was discussed. The results show that the entropy method provides a better performance than the coherence map in high coherence areas. The entropy estimation is also shown to be more details than the coherence estimation with small estimation windows. The performance of the new method is evaluated by processing real SAR images from Nanjing city in China.