An Image Registration Algorithm Based on Digital Image Fusion

Authors:	Wang Rong, Key Laboratory of Process Industry Automation, China Gao Liqun, Key Laboratory of Process Industry Automation, China
Topic:	1.1 Modelling, Identification & Signal Processing
Session:	Image Processing and Biomedical Applications
Keywords:	Image fusion; Image registration; Corner point detection; Multi-scale

Abstract

In this paper, an image registration algorithm based on both feature and intensity is proposed. Decomposing the source images with wavelet, corner points can be detected in the two approximate sub-images. The initial registration parameters can be achieved through affine transformation model, the coarse registration parameters can be obtained by the optical flow estimating and iterative refining, and the parameters are passed on to the next finer level to refine. The process will be repeated until we obtain the final registration parameters. Theory and experiments show this approach provided accurate registrations for DCIF images.