Chapter 2: Registration

Title: Deformable image registration using spring mass system for remote sensing images

Author(s): Jian-Kun Shen, Bogdan J. Matuszewski, Lik-Kwan Shark

Address: ADSIP Research Centre, University of Central Lancashire, Preston PR1 2HE, UK | ADSIP Research Centre, University of Central Lancashire, Preston PR1 2HE, UK | ADSIP Research Centre, University of Central Lancashire, Preston PR1 2HE, UK

Reference: Atlantic Europe Conference on Remote Imaging and Spectroscopy pp. 49 - 56

Abstract/Summary: This paper presents a novel multi-resolution registration method for remote sensing images. In the proposed method, the deformation is modelled using an elastic spring mass system, containing a number of sparse spring-points and springs interconnecting them. At the initial stage it is assumed that the system of interconnected points is in equilibrium. One of the advantages of using such a system is that the system can explicitly represent features in the image in order to improve the registration accuracy. By moving the featured masses in one image to the corresponding location in the other image, the system will no longer be in equilibrium. The remaining points will follow to the appropriate location until all the forces are in equilibrium. In the following stage, the external forces calculated from the image content are introduced into the system to locally guide movable masses to gradually improve registration accuracy under the constraints caused by the internal forces. The applied multi-resolution approach reduces the computational time and improves the registration accuracy. The method is fast, robust and offers high registration accuracy. The effectiveness of the method is demonstrated using a number of real remote sensing images.

Order a copy of this article