Patch-Based correlation for deghosting in exposure fusion

Abstract We present a robust exposure fusion algorithm to tackle the problems of motion removal and detail preserving in dynamic scenes. Using one exposure as a reference, the algorithm detects motion in an exposure stack by comparing the structural consistency of images as measured by the degree of linear correlation between patches of each image and the corresponding patches of the reference. Then, after motion removal, a stack of latent images with consistent content is synthesized. For detail preserving, a contrast criterion is introduced to measure the quality of exposure and generate visibility maps of each latent image. Guided by the visibility maps, a tonemapped-like HDR image, which is ghost-free and has all details preserved, is produced by seamlessly merging the latent images. Exposure fusion tests on various dynamic scenes demonstrate the superiority of the proposed method over existing state-of-the-art approaches.