SAR and optical images registration using shape context

Image registration is the process of overlaying two or more images of the same scene taken at different times, from different view points, and /or by different sensors. A novel feature-based multi-sensor image registration system is developed. The system consists of two new points: first, edge features are extracted from images, and the features are dilated to suppress some certain kinds of noise arising from groves; second, the preprocessed features are matched using the improved shape context. The shape context has been found to be robust in hand written digit and object recognition, and now it is introduced into remote sensing image matching after some adjustments. The developed system is successfully applied to register airborne optical and C-band SAR images in our experiments, and the results demonstrate its robustness and accuracy.