An image registration algorithm using a particle filters

The image registration is the process to overlay two or more images taking at different times. The image registration is a very important step in medical imaging as well as in remote sensing to quantify changes occurring over times. Most image registration algorithms measures the distances between two images and try to find the transformation that these distance is minimum. However, the distance function is usually multimodal and non-convex. Hence, the gradient-based algorithms usually stuck in one of the local optima. As a result, we propose the use of the particle filtering algorithm to search for the global optima.