A fast and accurate approach for 3D image registration using the scatter search evolutionary algorithm

Nowadays, image registration (IR) is still an important and useful task in several areas such as remote sensing, medicine, cartography, and computer vision. Different approaches to solve the existing variants of the problem are commonly proposed in the specialized literature. In this paper, we focus our interest on the 3D IR problem considering similarity transformations and our proposal is based on the use of a new procedure based on the evolutionary computation framework for non-linear optimization. We apply an emergent global optimization strategy called scatter search providing a fast and accurate algorithm. To measure its performance, we design an experimental setup considering some of the most accepted and accurate classical and evolutionary techniques for the problem, as well as six different shapes, one synthetic and five magnetic resonance images, dealing with different levels of noise and occlusion in the scenarios treated.