Chapter 2: Registration

Title: A new line segment feature matching algorithm for branch-and-bound image registration

Author(s): Lik-Kwan Shark, Bogdan J. Matuszewski, Andrey Kurekin

Address: ADSIP Research Centre, University of Central Lancashire, Preston PR1 2HE, UK | ADSIP Research Centre, University of Central Lancashire, Preston PR1 2HE, UK | Department of Computer Science, School of Engineering, Cardiff University, Cardiff CF24 3AA, UK

Reference: Atlantic Europe Conference on Remote Imaging and Spectroscopy pp. 35 - 42

Abstract/Summary: This paper presents a new line segment feature matching method that finds the global optimal solution by implementing a branch-and-bound search technique. It is based on recursive subdivision of geometric transform parameter space and discarding the parameter subsets that does not provide good matching accuracy. The paper includes a new modified square perpendicular distance to measure the similarities between two sets of image line segments, with the assumption that the sequences of line segment features can be fragmented, grouped in a different way and shifted at presence of image background noise and sensor impairments. The partial Hausdorff distance measure is applied to reduce the sensitivity of the estimated parameters to outliers in line segment data. The performance of the developed algorithm is evaluated using radar remote sensing images.

Order a copy of this article