Improving robustness of real-time geo-registration process

In this paper, a modified correspondence matching method is proposed for geo-registration system that enables parallelism and image update detection. The image to be geo-referenced is divided into blocks which can be processed in parallel. A voting process is performed on the correlation measures obtained from different blocks to determine the match point. An image update detection method is also presented, which uses the correlation data from the proposed block based matching method to determine which parts of the image have been updated or changed with respect to an existing reference image. The experimental results to evaluate the proposed methods are presented. The proposed method gives better matching scores compared to the conventional correlation approach.