EVALUATING THE PERFORMANCE OF EPIPOLAR GEOMETRY FOR AUTOMATIC IMAGE REGISTRATION
No Thumbnail Available
Date
2017-04
Journal Title
Journal ISSN
Volume Title
Publisher
Department of Geography, University of Ilorin
Abstract
Image registration has being of great importance to a variety of applications in remote sensing, medical diagnosis, computer vision, and pattern recognition. Accordingly, several methods have been proposed by different researchers, with a view to enhance the performance of image registration process. This paper investigates the performance of automatic image registration using epipolar geometry. The epipolar geometry was recovered through the estimation of a fundamental matrix from matched image points. Then the rectification transform was computed using the F-matrix, and the images registered using epipolar correlation. A root mean square error (RMSE) of 0.032 and 00 was recorded for the correlated points. the paper reveals that the image registration process performed well and therefore suitable for remote sensing image registration
Description
Keywords
Image Registration, Epipolar Geometry, Fundamental Matrix, Image rectification
Citation
Ahmadu, H. A., Abdulyekeen, A. O., Alade, A. K., and Yusuf, A., (2017). An investigation into the performance and accuracy of epipolar geometry for automatic image registration. Geo-Studies forum, Vol. 9, No. 1, pp. 66 – 73