Rigid Image Registration based on Normalized Cross Correlation and Chaotic Firefly Algorithm
Yudong Zhang, Lenan Wu · International Journal of Digital Content Technology and its Applications · 2012
Image registration is a hotspot in the field of image processing and automatic target recognition, and it can be simplified as an optimization problem, including three input variables (two translational parameters and one rotational parameter) and one output variable as the normalized cross correlation (NCC). To solve the optimization problem, we introduced in the latest nature-inspired technique, chaotic firefly algorithm (CFA), which is based on the behavior of fireflies. The simulation experiments on 18 standard benchmarks demonstrate that, for the mean absolute error of spatial translational parameter (tx and ty) and rotational parameter (θ), the CFA achieves the lowest error as 0.0253, 0.0246, and 0.0020, respectively. Therefore, CFA is effective for the rigid image registration problem.