Efficient and Robust Approach to Simultaneous Non-Rigid Registration and Gamma Estimation in the Frequency Domain

This paper presents an efficient and robust method for performing non-rigid image registration and relative gamma estimation in a simultaneous manner. The proposed method identifies regions of interest within the target image and determines the optimal correspondence and relative gamma between each region of interest and the reference image. The problem of finding the optimal correspondence and relative gamma between each region of interest and the reference image is combined and formulated such that both problems can be solved as a single minimum squared distance problem efficiently using the Fast Fourier Transform (FFT). The resulting correspondence and relative gamma information is used to estimate the geometric transformation between the target image and the reference image, as well as perform gamma correction between the images. Experimental results demonstrate that good registration and relative gamma estimation accuracy can be achieved using the proposed method, making it well-suited for registering images captured under different levels of exposure.