Rigid Image Registration by Bare Bones Fireworks Algorithm

摘要

Image registration is widely used in various applications in astronomy, medicine, robotics, and others which makes it an important research topic. Image registration problem refers to image alignments and it can be defined by linear (rigid) or non-linear transformation. Finding the optimal transformation that aligns two images is a hard optimization problem. In this paper, we propose a recent swarm intelligence algorithm, bare bones fireworks algorithm, for solving rigid image registration problem. The proposed method uses normalized cross-correlation as similarity metrics and it was tested on six benchmark images. Obtained results were compared to other approaches from literature and it has been shown that the proposed technique is better in term of registration accuracy.