Study of change detection based on Poisson’s mixture model and component-maximum threshold algorithm

Abstract

This article proposes a new Poisson’s mixture model (PMM) and component-maximum threshold algorithm that can be used for change detection based on the difference between two satellite images of the same scene recorded at two different time points. Using the method proposed in this article, the difference image obtained through subtraction of the recorded satellite images is first modelled as the PMM, and the component-maximum threshold algorithm is then used to segment the difference image into changed and unchanged pixels according to the maximum value of each component. A final binary image indicating the changed and unchanged pixels of the satellite images can be obtained by minimizing the sum of the mean square error. The validity of the method proposed in this article is demonstrated through synthetic data and satellite images.

Acknowledgement

This work was supported by the Participation in Research Programme (PRP) of Shanghai Jiao Tong University (No. T030PRP21037).